Amazon's $78bn ad machine now sets the defaults agencies buy inside

PPC Land is now also on WhatsApp, with a new PPC Land channel carrying its stories.

A PPC Land special edition on Amazon's advertising business, published November 14, 2026. It covers the period from Amazon's first public hints about advertising revenue in 2017 to Amazon Ads Support Center changes dated November 13, 2026, with most of the detail drawn from the years since the pandemic began in March 2020. Revenue figures come from Amazon's filings and earnings releases as reported in the linked coverage. Growth rates are given as each source reported them; where Landline has calculated a figure itself, the text says so.

On Friday morning, November 14, 2026, PPC Land published an account of a help page that most advertisers will never read. Amazon's Support Center guide to building a DVA+ campaign, updated the previous day, now limits each automated campaign to 20 products, offers three routes into the product with the automated one selected by default, and lists seven possible campaign objectives while showing each advertiser only those it is eligible for. It advises against budget caps on the grounds that they can cause under-delivery, warns that changing a key performance indicator after launch resets what the system has learned, and leaves audio, a format Amazon has sold programmatically for years, out of the default flow altogether. Two days earlier, another Support Center change had told agencies that the campaign-level fee setting they have used inside Amazon's demand-side platform can be set only until January 31, 2027, after which agency fees move into manager accounts that Amazon controls and describes, in its own words, with the line "Amazon doesn't bill agency fees." None of this was announced on a stage. It is the plumbing of an advertising business that took in $78.0 billion over the twelve months to September 30, 2026, and it shows where that business is now heading: from selling inventory to the people who plan media, towards deciding, by default, how the media is planned.

The Landline newsletter can also be heard. Marketing Is Calling reads each edition aloud every day, on Spotify, Apple Podcasts and YouTube. Marketing Rodeo picks the most controversial advertising story of each day, weekends included, and has some fun with it, on Spotify, Apple Podcasts and YouTube.

Twenty products and seven objectives

The DVA+ guide is a useful place to start because it is so ordinary. DVA is the name Amazon gave its demand-side platform at the unBoxed conference in San Francisco on September 29, 2026, when it retired the Amazon DSP brand and folded the buying tool and its advertising console into a conversational interface called Amazon Ads Agent. The letters stand for display, video and audio. DVA+ is the simplified, automated mode of that product, and since September it has been the default way into it. The guide PPC Land described on November 14 sets out what that default now means in practice.

There are three ways to build a campaign. The first, labelled DVA+, is pre-selected. The second, DVA+ advanced, exposes more controls. The third, a legacy mode, preserves the older demand-side platform workflow for buyers who still want it. In the default route, an advertiser picks an objective from the list the system offers, adds up to 20 products, accepts or edits AI-generated creative and sets a budget, which can be monthly, daily or lifetime. Three objectives apply to Amazon's own properties: sales on Amazon, optimised to total return on ad spend; traffic on Amazon, measured by cost per detail page view, which the guide says is reported but not optimised to; and awareness on Amazon, described as reach "at optimal frequency". Display, online video and streaming television are the default formats. Location targeting is available only for objectives that point off Amazon. Up to three frequency caps can be set for each ad group, at user or household level. Managed-service clients pay a fee calculated as a percentage and invoiced separately. And pressing the launch button does not, by itself, start a campaign.

Each of these choices can be defended on its own terms. A cap of 20 products keeps the creative system from trying to promote a whole catalogue at once. Hiding objectives an advertiser cannot use reduces errors. The warning on budget caps, that they "can cause under-delivery by blocking full budget use", is technically true of almost any automated buying system. But taken together, and set beside the rest of what Amazon changed in the same week, they describe a product designed around the assumption that the machine sets the parameters and the human approves them. PPC Land listed what the guide leaves out: no default value for the frequency caps, no percentage for the managed-service fee, no explanation of what "reach at optimal frequency" means, no rule for how product variations count towards the limit of 20, and no statement of which countries or beta stage the page applies to. It also noted that the guide gives two different sets of launch instructions.

That last point matters more than it sounds. On October 29, PPC Land reported that the conversational Ads Agent experience is still in closed beta in 25 countries, that DVA+ advanced settings are limited to some US advertisers, and that Amazon's own pages disagree on which markets keep the old DSP experience, with one list naming 21 countries. Advertisers with programmatic experience get both tiers. New advertisers and those coming from Sponsored Ads get DVA+ only, and must ask an account manager if they want the advanced controls. Agencies with existing entity access get both. DVA+ is billed on a cost-per-thousand basis with no minimum spend, and payment is by invoice only, with credit cards promised for the first quarter of 2027. Destaney Wishon, chief executive of the agency btr media, captured the confusion on LinkedIn with a line PPC Land quoted: "DSP is no longer DSP. It is DVA." She went on to suggest that advertisers who do not own their seat on the platform should move to own it sooner rather than later, which is the kind of advice agencies give when they suspect the ground under their contracts is shifting.

The agency fee moves to the manager account

The second Support Center change, dated November 11 and reported by PPC Land at 21:31 UTC on November 12, is narrower and more revealing. Agencies buying through Amazon's platform have long been able to set a fee on each campaign, so that the client's invoice reflects the agency's margin. That legacy setting can be applied until January 31, 2027 and will be removed entirely by the end of March 2027. Campaigns that keep it will not receive new features. In its place, Amazon has put agency fees for display, video and audio into open beta inside manager accounts, in all marketplaces.

The new structure is more elaborate than the old one. There are five ways to calculate a fee: four markups, based on supply cost alone, supply cost plus the Amazon audience fee, supply cost plus the total audience fee, and supply cost plus the total audience fee plus third-party pre-bid charges, and one margin, calculated as a percentage of budget. A fee can be applied to all current and future accounts, to specific accounts, to specific campaigns or to specific ad groups, and when more than one could apply, "the most specific fee is used". Amazon's worked example takes a 3.5% markup on $10,000 of supply cost and produces a $350 fee and a $10,350 total. Its percentage-of-budget formula, applied to the same figures, gives $362.69, which PPC Land calculated as 3.63% of $10,000, and which cannot be squared with the sentence that follows it on the same page. Two further lines define the relationship. "Amazon doesn't bill agency fees," the page says, and "Advertiser accounts can't set agency fees."

Neither sentence is hostile. The first means the agency invoices its client separately; the second means a brand cannot impose or remove its agency's margin from inside its own account. But the arrangement places the fee in a layer of Amazon's software that Amazon designs, and PPC Land listed the questions the page does not answer: whether clients can see the agency fees column, whether the fees apply to DVA+ campaigns at all, what happens to legacy fees on campaigns already running, and whether a change made in mid-flight applies to the whole campaign. If agency fees do not reach the automated tier that Amazon has made the default, then the default is also the one route on which the agency's margin is invisible to Amazon's billing.

The same evening, PPC Land published three more accounts of Support Center pages updated on November 11. One found that Amazon's 16 benchmark recommendations never suggest cutting budgets: the page sets out eight metrics, each with a favourable and an unfavourable set of actions, and where cost per click or cost per thousand impressions sits below the benchmark, the advice is to "test increasing bids". Another reported that advertisers will lose unmigrated reports on January 31, 2027, with the old report pages going read-only on December 17 and ten report types, including pricing transparency, conversion path and Amazon Attribution reports, unable to migrate yet. A third counted the dimensions in the replacement system and found that 85 of Amazon's 156 unified report dimensions apply to Sponsored Products, against 137 that carry the DVA+ marker, with 59 that apply to DVA+ alone.

Read separately, these are housekeeping. Read together, they describe a platform consolidating the places where an outside party used to add value: the choice of products and objectives, the selection of deals, the setting of budgets, the invoicing of margin and the export of history. That is the argument of this special edition. Amazon spent the years after 2018 building the inventory, the data and the cheapest demand-side platform in the market, in large part to win budgets that agencies controlled. It is now in a second phase, in which the money already flowing through its systems is spent inside defaults Amazon writes. The rest of this piece sets out how it got here, with the numbers, the dates and the evidence on both sides.

A decade in numbers

Amazon did not publish an advertising revenue figure of its own until February 3, 2022. Before then, the money sat inside a line called "other", which the company described in its filings as sales not otherwise included elsewhere, "such as certain advertising services and our co-branded credit card agreements". Executives said on most earnings calls that advertising made up the bulk of it, and analysts treated the line as a proxy. The table below follows that convention for 2016 to 2020 and switches to the advertising services line Amazon has reported since. The two series are close but not identical, and the break is marked.

Sources: Amazon's 2018 annual report on Form 10-K, which restates 2016 and 2017 on the same basis; GeekWire on 2018; MarTech on 2019; Digiday, Marketing Dive and GeekWire on 2020; CNBC on 2021; Intentwise and Forrester on 2022; Digital Commerce 360 on 2023; Adweek on 2024; PPC Land on 2025; Amazon on 2025 net sales; PPC Land on the trailing twelve months. Changes and shares are Landline calculations from the figures shown.

Three things stand out. The first is the scale of the compounding. On the figures above, the line Amazon used to call "other" was worth less than $3 billion in 2016; the advertising business that replaced it took in more than $78 billion in the twelve months to the end of September 2026, roughly 26 times as much. Over the same stretch, Amazon's total quarterly sales for the four quarters to September 2026 added up to about $797 billion on Landline's arithmetic, a little under six times the 2016 total. Advertising, in other words, has grown several times faster than the company around it, and its share of Amazon's sales has risen from about 2% to nearly 10%.

The second is that the growth did not come in a straight line. The leap in 2018 was flattered by an accounting change: Digiday reported in April 2018 that Amazon had moved some advertising revenue into the "other" line from January that year, adding $560 million in the first quarter alone. The pandemic years then produced the steepest real acceleration, before growth settled at around 20% a year between 2022 and 2025, a rate that would be remarkable for almost any other advertising business of that size. In 2026 the quarterly pattern has been distorted by a calendar change, of which more below.

The third is that advertising has become one of the main sources of Amazon's profit, even though the company does not say how much it earns from it. Amazon reports operating income for its North America, International and Amazon Web Services segments, and advertising is spread across the first two. Brian Olsavsky, the chief financial officer, told analysts in February 2022 that advertising "only works if we make it useful for customers", and CNBC noted that Amazon had never disclosed advertising operating income. It still has not. The margin on a sponsored product placement sold against Amazon's own search results, with no content to license and no publisher to pay, is widely assumed to be far higher than that on retail, but that assumption rests on inference rather than disclosure.

The years the number sat in "other"

The first public sign that Amazon's advertising business had become material came in October 2017. Search Engine Land estimated after Amazon's third-quarter results that advertising revenue had passed $1 billion in the quarter, roughly double the level of early 2016, on the reasoning that the "other" line was "basically advertising". Olsavsky said on the call that "advertising revenue continues to grow very quickly", and that its growth was faster than the line it sat in. Three months later, MarTech reported that "other" had grown 60% to $1.7 billion in the fourth quarter of 2017, within total revenue of $60.5 billion. Dave Fildes, then director of investor relations, said that "advertising was also a key contributor".

By April 2018 the line had reached $2 billion a quarter. Digiday's account of the first-quarter results ran under the headline "Amazon advertising is now worth $2 billion", and quoted Olsavsky's description of advertising as "a multibillion program". Marketing Dive, summarising the same results on April 30, 2018, noted that Amazon did not specifically break down its advertising revenue, and cited a ClickZ and GroupM study in which 63% of brands planned to increase their Amazon budgets, against 54% for Google and 53% for Facebook. Merkle data in the Digiday piece put the year-on-year rise in Sponsored Products cost per click at 65%. Even then, the price of an Amazon click was rising faster than the volume of clicks.

The same year brought the first reorganisation of the product line. On September 5, 2018, AdExchanger reported that the Amazon Advertising Platform had become Amazon DSP, and that Amazon Media Group and Amazon Marketing Services had been retired under a single Amazon Advertising name. Digiday's cheat sheet the following day noted that Headline Search Ads had become Sponsored Brands and that Amazon's quarterly advertising revenue had reached $2.2 billion, up 129%. GeekWire's account of the full year put "other" revenue at $10.1 billion in 2018, with $3.4 billion in the fourth quarter alone, up 95%, while reminding readers that Amazon did "not break out specific financials for advertising". In 2019 the line reached roughly $14.1 billion, up 39%, and Fildes offered the explanation that would become Amazon's standard pitch: "We're uniquely positioned to do this given our retail business."

The pitch was not only to brands. In July 2018, Digiday reported that Amazon's advertising sales team was increasingly meeting marketers and chief marketing officers directly, cutting agencies out of conversations they had once led. Seth Dallaire, then vice president of worldwide advertising sales, described Amazon's offer to brands as helping them "put together a promotional plan to help make it successful". Dan Salzman, global head of media at HP, which then ranked Amazon as its third-biggest media partner behind Facebook and Google, said that "the benefits of working directly with Amazon are greater speed and operational efficiency". Amazon told Digiday that agencies remained important, and Marketing Dive's summary the next day recorded the same assurance. Kevin Packler of the Knoxville agency Tombras Group described Amazon as "neither a pure-play competitor nor a pure-play ad platform. It's in the middle." Eight years later, that middle position is the one Amazon is now trying to leave.

Amazon finally broke the number out on February 3, 2022. AdExchanger's headline was "Amazon Breaks Out Advertising Services (At Last) - On A $31 Billion Run Rate". Advertising services brought in $9.7 billion in the fourth quarter of 2021, up by about a third, and $31.2 billion for the year, about 7% of Amazon's revenue. What remained in "other" was worth about $710 million in the quarter. "It got to the point where it was a size that we felt we needed to break it out," Olsavsky said. He also offered a line that reads differently in 2026 than it did then: "Longer term, the Amazon DSP opportunity is something we'll continue to work and refine." CNBC, reporting the same disclosure, compared it with Microsoft's advertising revenue, which Satya Nadella had said exceeded $10 billion in 2021, and with total 2021 revenue of $4.12 billion at Snap and $2.58 billion at Pinterest. Amazon's advertising business was already larger than all three put together.

March 2020: the shelf moves online

The pandemic arrived at Amazon as a logistics crisis before it became an advertising windfall. On March 17, 2020, Amazon told sellers it would temporarily prioritise household staples, medical supplies and other high-demand products at its fulfilment centres in the United States and Europe, which meant that many brands paying for Sponsored Products placements could not restock the items those placements promoted. The same day, Bloomberg reported that Amazon had cut its own spending on Google advertising, pulling back from Shopping ads at the moment when demand for its goods was outrunning its ability to deliver them. For a few weeks the question was not how much advertisers would spend on Amazon, but whether there was anything to advertise.

The answer came quickly. Digiday's Seb Joseph reported on May 1, 2020 that Amazon's advertising business had grown 44% in the first quarter to $3.91 billion, against 13% growth to $28 billion at Google and 17% growth to $17.4 billion at Facebook. Merkle's client data showed Sponsored Products sales up 70% year on year and Sponsored Brands sales up 131%, on a 118% increase in cost. Tinuiti, tracking about $400 million of US spend, saw Sponsored Products ad growth slow to 6% in the last week of March as stock ran out, before recovering. Danielle Waller of Merkle described what was happening to budgets: "we're seeing advertisers take money from their holistic budgets to support Amazon as a growth channel."

The second quarter made the shift structural. Amazon's revenue rose 40% to $88.9 billion, and the "other" line, which Marketing Dive described as mostly advertising, grew 41% to $4.22 billion. Amazon itself, Olsavsky told analysts, had cut its own marketing spend by roughly a third in the quarter because it did not need to stimulate demand it could barely fulfil. AdNews in Australia quoted eMarketer's Andrew Lipsman: "Amazon has strengthened its future position in key growth areas like grocery, health and advertising." Prime Day, normally a July event, was moved to October 13 and 14 that year, pushing one of the heaviest advertising weeks of the calendar into the fourth quarter.

By the end of 2020, the line had grown between 41% and 49% in every quarter since the third quarter of 2019, and in the fourth quarter it jumped 64% to $7.95 billion, overtaking subscription services, GeekWire reported. Olsavsky credited the move of Prime Day into the quarter, and said: "I would say that there's been a recovery in advertising spend as the year progressed." Michael Pachter of Wedbush was less impressed by the surprise than by the arithmetic: "It's not really surprising in the context of their sales growth." Across the whole year the line reached about $21.5 billion, roughly half as much again as in 2019.

What the pandemic did to Amazon's advertising business was to move a large share of everyday shopping, for groceries, household goods and pharmacy items as well as electronics, onto a site where every search results page carried sponsored placements. Brands that had treated Amazon as a distribution channel learned to treat it as a media channel, because the product listing was now the shop window. Agencies built retail media practices around it. Amazon, in turn, learned that advertising revenue tracked retail traffic closely and that the two could be grown together. That lesson informed everything it did next.

After the surge: 2021 to 2025

The first full year with a published advertising number was also the last of the pandemic boom. Advertising services reached $31.2 billion in 2021, with growth of 32% in the fourth quarter. Then retail growth stalled. Marketplace Pulse reported on February 3, 2023 that Amazon's first-party retail sales had fallen 0.9% to $220 billion in 2022, while its advertising business expanded by 21%. Juozas Kaziukenas summed up the year: "Amazon's growth in 2022 was nearly flat." Advertising kept growing anyway, which was the point. Adweek reported on February 2, 2023 that Amazon's ad business continued to surge even as Google and Meta faced slowdowns, with fourth-quarter advertising sales of $11.6 billion, up 19%. Forrester calculated that the $38 billion from advertising services in 2022 was nearly double the level of two years earlier.

In 2023 growth picked up again. Marketing Dive's headline on February 2, 2024 was "Amazon's Q4 ad sales total $14.6B as Prime Video ads begin", and the quarter's 27% rise owed nothing to video: Olsavsky said that "the strength in advertising was primarily driven by sponsored products." The full year came to $46.9 billion. The following year, 2024, was the first above $50 billion: Adweek reported on February 6, 2025 that Amazon's ad revenue was $56.2 billion, with $17.3 billion in the fourth quarter, up 18%. That was also the year Prime Video began to matter. Digiday's Seb Joseph and Krystal Scanlon reported on February 20, 2025 that, in one agency sample, the demand-side platform's share of Prime Video ad spend rose from 26% in the third quarter of 2024 to 36% in the fourth, and that Tinuiti's clients spent more than eight times as much on Prime Video in the fourth quarter as in the first.

In 2025, advertising revenue reached $68.6 billion. The quarterly figures since then show both the momentum and the first real interruption to it.

Sources: PPC Land on Q3 2025, Q2 2025, Q4 2025 and the 2025 quarters, Q1 2026, Q2 2026 and Q3 2026; Adweek on Q1 2026. Amazon reports both headline and currency-neutral growth, and outlets quote either; Amazon's chief executive put fourth-quarter 2025 growth at 22% in the company's release. The drivers column summarises the linked coverage.

The third quarter of 2026 is the anomaly in that table, and it deserves attention because it is the closest thing to a counter-signal in Amazon's advertising record. PPC Land reported on October 29 that advertising revenue rose only 10.9% to $19.64 billion, down $172 million from the second quarter. Amazon attributed the slowdown to "timing anomalies primarily related to Prime Day (and a couple other timing factors)". Prime Day ran from June 23 to 26 in the United States, inside the second quarter, where it had previously fallen in July. Olsavsky said that "Prime Day timing shifted this year, with the event occurring in Q2 for most large countries." AdExchanger's Joanna Gerber reported that Andy Jassy also pointed to the absence of NBA games in the quarter. The combined second and third quarters came to $39.45 billion, up 18.1%, which is a fairer measure of the underlying rate. For the fourth quarter, Amazon said: "We expect year-over-year advertising revenue to grow in the high teens in fourth quarter 2026." Digiday's Ronan Shields noted that Amazon expects total fourth-quarter revenue of $240 billion to $247 billion.

One explanation Amazon rejected is that AI shopping agents run by other companies are already eating into sponsored search. Adweek reported on October 30, 2026 that Amazon says external AI agents are not hurting its ad business, and quoted Jassy: "Few customers are doing much shopping this way yet." Adweek put Amazon's trailing twelve-month advertising revenue at $78 billion, the same figure PPC Land calculated. There was also a smaller warning sign from the advertising side of the calendar shift. CommerceIQ data reported by PPC Land showed that measured US Prime Day advertising spend fell 8.8% while conversion rose 17.1%, which describes fewer and better-qualified impressions, but also an event at which brands spent less than the year before.

Where Amazon sits in the market

Amazon's advertising revenue is now large enough to be measured against the two companies that have dominated digital advertising for two decades, and the comparison shows a business that is smaller but gaining. Ronan Shields' Ad Tech Briefing for Digiday on September 15, 2026 carried Madison and Wall's estimates of US advertising revenue shares in 2024 and 2025:

Source: Madison and Wall estimates, via Digiday, September 15, 2026.

Luke Stillman of Madison and Wall told Digiday that the three companies are "just going to outperform the average every year", because more spending produces better optimisation data and better optimisation attracts more spending. Globally, WARC's forecast, cited in Sam Bradley's Media Buying Briefing for Digiday on October 12, 2026, has advertising spend growing 11.9% to $1.34 trillion this year, with Alphabet, Amazon and Meta taking 59.7%, about $659.6 billion. Amazon is the smallest of the three, but it is the only one that also sells most of the products being advertised, operates the store in which they are bought and can see the purchase. That is the asset everything else in this story is built on.

Prime Video: the audience Amazon switched on

If there is one decision that explains how Amazon went from a search advertising business to a television advertising business, it was taken in September 2023 and carried out on January 29, 2024. On September 22, 2023, Amazon announced that films and series on Prime Video would carry "limited advertisements" from early 2024, starting in the United States, the United Kingdom, Germany and Canada, with France, Italy, Spain, Mexico and Australia to follow later that year. Members who wanted to avoid them would pay an extra $2.99 a month in the United States. Advanced Television quoted the company's promise that it aimed "to have meaningfully fewer ads than linear TV and other streaming TV providers".

What made the move different from every other streaming service's ad tier was the direction of the default. Netflix and Disney+ had launched cheaper plans with advertising and invited subscribers to move down to them. Amazon put every existing Prime Video member onto the advertising plan at once and asked those who objected to pay to leave it. 9to5Google reported on December 26, 2023 that Amazon had begun emailing members to say that ads would start on January 29, 2024, and that no action was required to keep watching with them. Nobody had to sign up. The largest ad-supported streaming audience in the United States was created by a setting.

Amazon presented the result at its first upfront, in New York on May 14, 2024. Its own recap claimed "an average monthly ad-supported reach of 200 million global customers, 115 million of whom are in the U.S." Paul Kotas, senior vice president of Amazon Ads, framed the pitch around scale: "It's about delivering relevant advertising in Prime Video to over 200 million customers." Mike Hopkins, head of Prime Video and Amazon MGM Studios, went further: "No other streaming service can reach the number of customers with premium entertainment that we can." The same event introduced three interactive formats, a shoppable carousel, an interactive pause ad and interactive brand trivia, and announced that Prime Video would carry five NASCAR Cup Series races in 2025. Amazon also claimed a streaming television audience of more than 175 million a month and weekly Thursday Night Football audiences of more than 10 million.

A year later, the US figure had grown again. The Hollywood Reporter reported in May 2025 that Prime Video's US ad-supported reach had reached 130 million. By the fourth quarter of 2025, Amazon said the average ad-supported audience was 315 million viewers across 16 countries, PPC Land reported on February 7, 2026, a little over half as large again as the global figure it had claimed in 2024.

The promise of fewer ads

The promise of "meaningfully fewer ads" was kept in the sense that Prime Video still carries much less advertising than broadcast television. It was not kept in the sense that the load stayed where it began. Fortune reported on June 12, 2025, citing an Adweek study, that Prime Video's ad load had risen from two to three and a half minutes an hour to between four and six minutes, against 13 to 16 minutes on linear television. Amazon's reply was that it remained "focused on prioritizing ad innovation over volume." That is roughly a doubling in about 16 months.

The price of leaving the ad tier went up too. PPC Land reported on March 15, 2026 that the US ad-free add-on would rise from $2.99 to $4.99 a month from April 10, 2026, an increase of 67%, under a new name, Prime Video Ultra, with an annual option at $45.99. The full Prime membership costs $14.99 a month or $139 a year, and a standalone Prime Video subscription $8.99. "Delivering ad-free streaming with premium features requires significant investment," Amazon said. PPC Land noted that a US class action over the original switch had been dismissed in July 2025, with the court treating the introduction of ads as a "benefit modification" that Prime's terms permitted.

Germany was less accommodating. On December 19, 2025, the Munich Regional Court ruled in a case brought by the Federation of German Consumer Organisations, the vzbv, that Amazon's February 2024 email telling German members that ads would appear unless they paid EUR 2.99 a month was misleading, and ordered the company to send a correction letter. The judgment was not legally binding at the time it was reported. It remains the clearest legal challenge to the method by which Amazon built its streaming audience: a contract that existing customers had signed without advertising, changed to one with advertising, with the cost of opting out set by Amazon.

The timeline below sets out how the Prime Video business was assembled.

A television screen that knows what was bought

The formats Amazon has built for Prime Video are designed to do something conventional television advertising cannot: connect the screen to the shopping basket. The shoppable carousel and interactive pause ads shown at the 2024 upfront allowed a viewer to add a product to an Amazon cart with a remote control. On October 31, 2025, Amazon announced zip-code-level targeting for Prime Video ads, which lets an advertiser generate thousands of local variants from one creative, available through managed and self-service buying in the DSP in the United States. PPC Land's account of the launch, published on November 12, 2025, noted that Prime Video's German reach alone was more than 17 million a month.

Six months later, on May 11, 2026, Amazon introduced Dynamic TV Creative for selected US advertisers in packaged goods, fashion and electronics, with an expansion to live sports and Prime Video Channels planned for the third quarter of 2026. The format assembles a television advertisement from components based on what Amazon knows about the household, and carries calls to action such as "Add to Cart" and "Send to Phone". PPC Land cited an IVA study of 14,518 campaigns from 9,960 advertisers in support of the approach, and quoted Rousseau, a director at CreativeX: "This moves away from a one-size-fits-all approach."

Measurement followed the inventory. In the United Kingdom, the audience measurement body Barb launched its Ads Hub on January 27, 2026 with a Prime Video integration under way, a step that matters because British television budgets are traded on Barb's currency. In the United States, Amazon's argument has been that it does not need a third party to tell it what happened, because it can see the purchase. Alan Moss, Amazon's vice president of global advertising sales, put it most directly in an interview reported by PPC Land on May 14, 2026: "Amazon's not guessing, we're knowing."

Prime Video also became the place where Amazon's demand-side platform found its most compelling inventory. Digiday's Seb Joseph and Krystal Scanlon reported on February 20, 2025 that, in one agency's data, the DSP accounted for 32% of Amazon ad spend in the fourth quarter of 2024, and that its share of Prime Video spend had risen from 26% to 36% in a single quarter. Clare Hogan of Pacvue told Digiday: "Amazon has removed the barrier of entry to the DSP." Shray Joshi of Good Peeps offered a measure of where it sat in the budget: "Amazon DSP for our brands is typically 10-15% of spend once they have maximized sponsored search spend." Prime Video, in other words, was the reason to log in to the DSP, and the DSP was the route through which the rest of Amazon's video supply would later be sold.

The Emmy deal shows how far the ambition now reaches. Amazon and the Television Academy disclosed on October 6, 2026 a six-year agreement making Prime Video the exclusive worldwide home of the Emmy Awards from the 2027 ceremony, ending a rotation among ABC, CBS, Fox and NBC that had run until the 78th ceremony on September 14, 2026. The awards began in 1949. Prime Video will be the first home outside broadcast television in their history, and the stream will need no subscription. It is the kind of tentpole event that, until very recently, only a broadcast network could sell.

Live sports as the anchor

Prime Video gave Amazon a large audience. Live sports gave it an audience that advertisers cannot easily buy anywhere else, that watches in real time and that does not skip. The investment began modestly. In 2017, Amazon took the streaming rights to Thursday Night Football from Twitter, reportedly for about $50 million for ten games. In 2021 it signed an 11-season agreement to become the exclusive home of the package from 2022, at a reported cost of about $1 billion a year. GeekWire reported that the deal would begin a year early and cover 15 regular-season games and one preseason game. Marie Donoghue, then vice president of global sports video, said: "This expedited deal is an immediate differentiator for us as a service."

The first exclusive broadcast, the Kansas City Chiefs against the Los Angeles Chargers on September 15, 2022, averaged 13 million viewers on Nielsen data, against 8.84 million for the NFL Network's Thursday games in 2021. CNBC reported that the median age of the audience was 47, against 53 for the previous broadcasts. Jay Marine, then head of sports at Prime Video, said: "By every measure, Thursday Night Football on Prime Video was a resounding success." Amazon's 2024 upfront materials claimed that the audience had grown 24% in its second season and that viewers were seven years younger than on broadcast.

The growth since then has been steady. PPC Land reported on October 31, 2025 that the 2025 season was averaging 15.3 million viewers, and on February 7, 2026 that the full season had averaged more than 15 million, up 16%, with a Wild Card playoff game between the Green Bay Packers and the Chicago Bears drawing 31.6 million. In 2026, Adweek reported on October 30 that Thursday Night Football, in its fifth season on Prime Video, was averaging 17.7 million viewers, up 17%. Adweek's Bill Bradley reported on October 24 on how Amazon made a stronger partnership play in Thursday Night Football in year 5: the number of advertisers rose 6% on the year, and 17% of the season's advertisers were new both to the broadcast and to Prime Video. Amy McDevitt, who leads live sports sales partnerships, was pushing sponsors to buy across Amazon's other businesses as well.

Basketball followed. In July 2024, Amazon agreed an 11-year deal with the National Basketball Association from the 2025-26 season, covering 66 regular-season games a year and playoff rounds, along with WNBA games through 2036. GeekWire, citing Bloomberg, put the cost at about $1.8 billion a year. Adam Silver, the league's commissioner, said that "the digital opportunities with Amazon align perfectly with the global interest in the NBA." The first regular-season game on Prime Video drew 1.25 million US viewers in October 2025. By the end of the first season, Amazon reported a peak of 6.5 million US viewers, and play-in games had drawn more than 3 million.

The deals have produced exactly what Amazon's advertising team wanted from them: new buyers. In the PPC Land interview published on May 14, 2026, Moss said that Thursday Night Football had brought in 80 net new advertisers and the NBA 30 in its first year. Moss joined Amazon in July 2020 and had the football deal done within nine months. He rejected the idea that Amazon's streaming and retail advertising are separate propositions: "I don't think of those as separate worlds." Asked whether AI agents would ever handle upfront negotiations, he answered with one word: "Never."

By the second quarter of 2026, there was nothing left to sell. Andy Jassy told analysts in late July that inventory on Thursday Night Football, NBA, WNBA and NASCAR had all sold out. The third quarter then showed the other side of a sports-driven business. With no NBA games between the end of the finals and the season opener on October 23, Jassy cited the absence of basketball as one of the reasons advertising growth slowed to 10.9%. Live sports supply is lumpy, and an advertising business that relies on it inherits its calendar.

Amazon is also building technology into the broadcasts. On October 28, 2026 it published a technical account of Prime 360, an AI system that generates new replay frames in selected NBA broadcasts, and said that running its diffusion models on Amazon's own Trainium chips cut processing latency by more than 40%, without giving a baseline. Amazon did not say whether the replays carry sponsorship.

The NBA cost is Bloomberg's figure as reported by GeekWire. Audience figures are Amazon's or Nielsen's as reported in the linked coverage.

Taken together, the two league agreements commit Amazon to roughly $2.8 billion a year in sports rights on the reported figures, which on Landline's arithmetic is a little under 4% of its advertising revenue over the twelve months to September 2026. That proportion is low by the standards of any broadcaster, and it is the reason Amazon can treat sports as a customer acquisition cost for its wider advertising business rather than as a programme that must pay for itself. Moss's 80 and 30 net new advertisers are the return Amazon talks about. The Prime memberships those games help to retain, and the products those advertisers then sell on Amazon, are the return it does not have to.

Building the cheapest DSP in the market

Amazon's demand-side platform is older than its streaming business. Amazon sold display advertising through a buying tool it called the Amazon Advertising Platform for years before the 2018 rename, and by October 2017 Digiday was reporting that buyers considered the technology good enough for Amazon's DSP to become a rival to Google, with fees already described as among the lowest in the market. For most of the period since, the DSP's main selling point was not its interface, which buyers often criticised, but two things no other platform could offer: Amazon's shopper audiences, and a price that undercut everyone else.

The price story is the most consistent part of Amazon's pitch, and it is also the most frequently misreported. In November 2024, Digiday wrote that advertisers typically paid 10% of intended spend through the DSP, against 15% for Google and 20% for The Trade Desk, although the managed service option demanded a minimum of $50,000. Amazon's own documentation, summarised by PPC Land on July 18, 2025 when it launched an ad tech activation partners programme, set the fully managed option at a $50,000 minimum, a 15% service fee and a 10% DSP platform fee.

Then the headline rates fell further. In April 2025, a beta tester told AdExchanger that programmatic guaranteed deals carried 0% DSP fees on Amazon media and 1% on a set of streaming publishers, with Amazon Marketing Cloud and data fees reducing the total savings, while The Trade Desk's take rate had run between 19% and 21% since 2017. Amazon's pitch deck, seen by Digiday in May 2025, "touts a 1% tech fee for programmatic guaranteed deals on premium open web inventory." Digiday also described Amazon courting buyers with loss-leader tactics, including an extra 10% of a new advertiser's video buy, capped at $250,000.

By October 2025, buyers told Digiday that Amazon's DSP fees typically ranged between 4% and 8%, sometimes falling to 1% and occasionally disappearing. Adweek reported on October 24, 2025 that Amazon had offered at least one agency a free head-to-head test against a rival DSP, covering inventory, technology and measurement costs. PPC Land's summary of the December 3, 2025 platform consolidation recorded no fee on programmatic guaranteed deals for Amazon-owned media and 1% on the open web.

Those numbers need to be read with care, because the 1% figure that buyers repeat has a narrower scope than it is often given. Both the zero and the 1% describe programmatic guaranteed deals, one way of buying on the platform, not the rate on everything that runs through it. Data targeting and audience fees are charged on top. When Amazon explained how its regulatory advertising fees are calculated, PPC Land reported in August 2024, it described DSP campaign costs as comprising media cost, audience fees and platform fees. Those regulatory advertising fees, it should be said, pass on digital services taxes, and are not a platform fee at all. Inside Amazon Marketing Cloud, third-party signals sit behind a subscription tier called Paid Features, and Amazon has made its own paid audience and retail purchase features free to query only until the end of 2026, without saying what they will cost from January 1, 2027.

Rates are as each source described them. The 0% and 1% figures apply to programmatic guaranteed deals; audience and data fees are charged in addition, and digital services tax pass-throughs are taxes, not platform fees.

The table carries one entry that cuts against the rest. In May 2026, an agency executive told Digiday that minimums for direct DSP access had been pushed into the "tens of millions", in Digiday's coverage of that month's upfronts. A platform that charges almost nothing on its own inventory but requires very large commitments for direct access is a platform that wants large buyers in its own interface and everyone else in its simplified products or dealing through resellers. DVA+, with no minimum spend, is the simplified product. That distinction matters for what comes later in this story.

Why low fees are affordable for Amazon

Vlad Chubakov, associate director of programmatic at Delve Deeper, put the logic in one sentence to PPC Land in August 2026: "Amazon can slash margins because they own the pipes." A standalone DSP such as The Trade Desk earns its revenue from the fee it takes on media it does not own. Amazon earns its revenue from the media itself, from the audience data advertisers buy to target it, from the measurement products it sells to evaluate it and, increasingly, from the cloud contracts that sit underneath. A low platform fee, and none at all on Amazon's own guaranteed inventory, costs Amazon little when every dollar of Prime Video or Thursday Night Football bought through the DSP is a dollar of Amazon advertising revenue in any case.

There is a cost to the advertiser that the headline rate does not show, and the comparison with Meta makes it visible. Meta has never charged advertisers a platform fee; its take is the auction itself. A buyer on Meta pays nothing for the software and everything through the clearing price. Amazon is moving towards the same model from the other direction: as more of what runs through its buying tool is Amazon's own inventory, priced by Amazon's own auctions or guaranteed deals, the platform fee becomes less relevant to what Amazon earns, and the price of the media becomes more so. Sarah Caputo, founder of the buy-side consultancy Fraction Method, found in an AdExchanger column on September 9, 2026 that Amazon's documentation describes its invitation-only Private Auction product, including Prime Video inventory, as clearing at a "variable CPM based on optimization," with no buyer-visible floor or clearing formula. "Platforms built the systems, and only platforms can fully explain how they set the price," she wrote.

Buying the pipes: the supply deals of 2025 and 2026

A cheap DSP needs something to buy. From the middle of 2025, Amazon set about connecting its buying platform to almost every large source of premium video and audio supply in the United States and many beyond it. Digiday later described the result as an eighteen-month run of supply acquisitions, in which Amazon did not build the audience but bought the pipe to it and then undercut rivals on fees.

The first and largest was Roku. On June 16, 2025, Amazon and Roku announced what they called the largest authenticated connected television partnership, reaching 80 million US connected TV households, more than 80% of the total. Amazon claimed advertisers could reach 40% more unique viewers with the same budget while cutting frequency by nearly 30%. "Our exclusive partnership with Roku is a giant leap for advertisers," Kotas said. A day later, on June 17, Disney said its inventory across Disney+, Hulu and ESPN would become available in the DSP through its DRAX programmatic platform, for all DSP advertisers from the third quarter of 2025, with Disney+ inventory in eight European markets as well. Amazon said its DSP now reached 275 million monthly US ad-supported customers. Kelly MacLean, vice president of Amazon DSP, called it "a significant leap forward in advertising effectiveness."

Netflix followed on September 10, 2025, when Amazon announced that Netflix inventory would be buyable through the DSP in 11 markets from the fourth quarter of 2025: the United States, the United Kingdom, France, Spain, Mexico, Canada, Japan, Brazil, Italy, Germany and Australia. Kotas said that the goal was to "remove the guesswork for advertisers by making it simple to manage all of their TV planning and buying with Amazon Ads." Audio arrived within weeks. SiriusXM Media, with 160 million monthly digital listeners across SiriusXM, Pandora and SoundCloud through AdsWizz, integrated with the DSP on September 16, and Spotify's global audio and video inventory followed in early October, giving access to 696 million monthly listeners in nine markets on Amazon's later count.

The most symbolic addition came from Microsoft. When Microsoft decided to wind down its Invest DSP by February 28, 2026, PPC News Feed reported on October 12, 2025 that Amazon DSP would be the preferred partner for Microsoft's advertisers, with Microsoft Monetize joining the Amazon Certified Supply Exchange through a preferred-partner integration. A company that had spent two decades trying to build its own display advertising business handed its buyers to Amazon.

The pattern continued through 2026. Triton Digital, the audio technology company owned by iHeartMedia, added Amazon DSP as a demand partner across more than 80 countries on July 15. On September 10, AWS Elemental added yield optimisation that fills unsold live ad breaks with Amazon Ads demand for streaming publishers using Amazon Publisher Services, and Amazon Ads and OpenAI opened a pilot that lets select US advertisers extend DSP campaigns into ChatGPT, with Delta Vacations the first to test it. On October 6, Sky Media made its UK and Irish streaming inventory buyable through Amazon Ads Agent, across Sky Go, NOW, Sky Glass, Sky Q and Sky Stream, with programmatic guaranteed deals available immediately. In late October, PPC Land reported that Pinterest sells ads through Amazon and Google DSPs in three regions, not Europe, with Amazon Ads DVA+ named among the buying platforms. And in early November, Adweek reported that Amazon advertisers can now target shoppers that buy elsewhere, through an integration between the data company Attain and DVA+ that uses purchase behaviour at Walmart, Dollar General, Kroger, Walgreens and Sephora.

By May 2026, Moss was claiming that the DSP reached 90% of US households through Netflix, Roku, Disney, Spotify and SiriusXM. The effect on market share was measurable. Guideline data cited by PPC Land on August 7, 2026 showed Amazon growing from under 10% to just under 20% of tracked programmatic spend in about 15 months.

Not every partnership has run to plan. PPC Land reported on October 22, 2026 that Netflix's ad revenue forecast had dropped to $2.7 billion as partner integrations slipped, without Netflix saying which integrations were late; earlier commitments had included Amazon's buying platform in all countries by June 1. And Netflix has not given Amazon exclusivity. Adweek reported in October 2026 that The Trade Desk would join Amazon and Yahoo as a programmatic buying platform for Netflix from November.

What it did to The Trade Desk

The largest independent DSP has borne the cost of Amazon's strategy more visibly than anyone. Digiday reported in December 2025 that The Trade Desk was loosening its grip on pricing amid buyer pressure, with an effective platform fee that could run from the mid-teens to roughly 20% when fully loaded. In August 2026, PPC Land reported that Amazon DSP was being placed at roughly 1% to 2% of media spend against a 12% to 15% range typically associated with The Trade Desk, which does not disclose its take rate. The comparison is less like-for-like than it looks, because Amazon's low end reflects guaranteed deals and neither range includes the audience data that advertisers buy on top. Jeff Green, The Trade Desk's chief executive, argued that rivals believe "their cheap fee will compensate for their lack of objectivity," and illustrated the point with an 8% fee against a 4% fee on $1 of media, producing $1.08 against $1.04. "It is not our goal to be the cheapest platform," he said.

The market was not persuaded. The Trade Desk reported second-quarter revenue of $715 million, up 3%, and guided third-quarter revenue to at least $650 million, about 12% below the prior year. Its shares fell 24.22% to $13.39 on August 6. In September, AdExchanger's account of why Wall Street turned against The Trade Desk quoted a description of Amazon as "the apex predator" for its budgets, and the company cut about 575 jobs. In November, PPC Land reported that Trade Desk revenue had dropped 0.5%, its first decline since its 2016 IPO, even though the figure came in $86 million above its guidance. Net income fell 67% to $38 million. Green conceded that the business was "still operating below the level of growth we expect from ourselves." PPC Land pointed to Safari blocking the company's ad domain, competition from Amazon and Walmart's non-exclusive renewal as pressures, while customer retention stayed above 95%.

Independent agencies are caught between the two. Digiday's Kimeko McCoy reported on October 22, 2026 that indie agencies want The Trade Desk and Amazon DSP to acknowledge their buying power. One 15-person agency that clears The Trade Desk's $100,000 monthly threshold still pays 17% to the platform plus 12% to a reseller. "They won't let us go direct and they won't change the fees," one executive said. For those agencies, Amazon's low fee is attractive in principle. In practice, the minimum for direct access means many of them buy Amazon through a reseller too, or through the simplified product Amazon has now made the default.

The first phase: going around the agency

The phrase "capturing ad dollars from agencies" can sound like an accusation. Amazon would describe the same process as serving advertisers directly, and much of what it did between 2018 and 2025 can be described in those terms without strain. But the record shows a consistent direction. At each stage, Amazon took a task that an outside party had performed and offered to perform it itself, usually more cheaply, and usually inside software it controlled.

The first stage was the sales relationship. As described earlier, Digiday reported in July 2018 that Amazon's advertising sales team was meeting chief marketing officers directly, and that HP valued "greater speed and operational efficiency" from the arrangement. Amazon said agencies remained important. What it built next was a service that competed with them on their own terms. Amazon's managed service for the DSP, in which Amazon's own staff plan and run campaigns, carried a $50,000 minimum and, on the documentation PPC Land summarised in July 2025, a 15% service fee on top of a 10% platform fee. That is an agency fee, charged by the platform. The Decoded entry PPC Land published on November 13, 2026 notes that the managed service "typically requires a minimum spend of USD $50,000", varying by country.

The second stage was the partner ecosystem. Amazon launched its Ads Partner Network in 2021, giving agencies and technology companies badges, training and access. By late 2025, Shopifreaks, summarising Adweek's reporting, described Amazon as pulling away from its ad tech partners to work directly with advertisers, as its own AI tools began to overlap with products those partners sold. Amazon's statement at the time was that partners "are integral to Amazon Ads, providing expertise, impactful products, and strategic guidance." The relationship with the developers who build on Amazon's interfaces was tested in the same period, when Amazon proposed fees on its Selling Partner API for high-volume users and then, Adweek reported, scrapped the planned fees after a backlash from technology firms.

The third stage was the price of direct access. A platform fee of zero on Amazon's own guaranteed inventory is a gift to whoever holds the seat. But Digiday's May 2026 report that minimums for direct DSP access had been pushed into the "tens of millions" meant that most agencies could not hold a seat of their own. They could buy through a reseller, buy through Amazon's managed service, or buy through the simplified, self-service products Amazon was building. All three routes left Amazon closer to the money than the agency.

The agency market adapted in the way markets do. Specialist Amazon agencies grew up and were bought: on August 17, 2026, the independent performance agency Chief Media acquired the Amazon consultancy AMZ Advisers and the social agency Reach Social, Adweek reported, to expand its marketplace and social commerce capability. Holding companies built commerce units of their own. Buyers kept shifting money to Amazon, because it worked: Digiday reported on October 7, 2026 that five media buyers saw client Amazon spend up 15% to 40% year on year. Darren D'Altorio, senior vice president of paid media at Wpromote, said his agency was "looking at 30% growth year on year" for brands that sell on Amazon and 40% for those that do not, and that "it is becoming increasingly more defensible to the CFO why we should be investing there." Hillary Bram, vice president of performance buying at Exverus by Brainlabs, offered a shorter version: "Full funnel on Amazon is what's working."

How agencies make money, and why it matters here

To understand why the second phase matters, it helps to recall how agencies now earn their margin. The traditional model was a commission or fee on media bought on a client's behalf. Over the past decade, a growing share of agency income has come from principal media, the practice in which agencies buy inventory themselves and resell it to clients at a price the client does not see broken down. The Association of National Advertisers reported on March 19, 2026 that 58% of the marketers it surveyed had used principal media in the past year, up from 47%, that 74% bought television this way, and that 90% were uncertain whether their agencies' recommendations served their interests, up from 79%. Forrester estimated in January 2026 that principal media made up about a third of agency billings.

The other model, used across Amazon's platform, is the markup: the agency buys on the platform and adds a fee to what the client is invoiced. Until now, Amazon let agencies set that fee on each campaign inside the DSP, so that the client's bill reflected the agency's margin. The November 11, 2026 Support Center change moves that setting into manager accounts, defines five ways to calculate it and states, in Amazon's words, that "Amazon doesn't bill agency fees." The change does not tell agencies what to charge. It does, however, place the calculation of their margin in a layer of software Amazon designs, define the base on which it is calculated, and leave open whether it applies at all on the automated route Amazon has made the default.

There is nothing unique to Amazon in a platform going directly to brands. Digiday's Kimeko McCoy reported on October 23, 2026 that Walmart Connect was pitching brands directly without their agencies, offering on-site brand lift studies to win budgets from Amazon. Walmart's global advertising business grew 38% in the second quarter, and eMarketer puts US retail media spending this year at $69.33 billion. Google has long sold to large advertisers directly, and Meta's automated Advantage+ products were designed to need little outside management. What distinguishes Amazon is the combination: it owns the store, the inventory, the buying tool, the measurement and now the place where the agency's fee is set.

Alex Brownstein of AHB Advisors described the result to Digiday after Amazon's third-quarter results in a sentence that summarises the whole strategy: "Amazon is open where openness brings it more of the budget and closed where its advantage sits." The supply deals described above are the open part. The defaults described in the next section are the closed part.

The second phase: owning the defaults

Amazon learned what a default is worth from its own customers. On January 29, 2024, it did not ask Prime members whether they wanted advertising. It switched advertising on and charged those who wanted it off. Behavioural economists have documented for decades that most people stay with whatever option is pre-selected, whether it is a pension contribution, an organ donor register or a privacy setting. Prime Video turned that observation into one of the largest ad-supported streaming audiences in the world within a few months. The second phase of Amazon's advertising strategy applies the same principle to advertisers.

The sequence is easiest to see as a list, because each step on its own looked modest.

Three of these changes deserve closer attention, because they show the mechanism most clearly.

Who picks the deals

The first is automatic deal selection. For most of the history of programmatic television, the choice of which deals to buy was where an agency trading desk earned its keep: knowing which publisher's inventory performed, which private marketplace carried the right audience, which guaranteed deal was worth its price. On April 15, 2026, Amazon launched automatic deal selection for streaming television campaigns in the United States, for streaming TV, awareness and reach objectives. In the platform's interface, the deals Amazon chooses carry a source type called "AMAZON_CURATED_AGENT". Nicole Goksel, a senior technical product manager at Amazon, described the change as making the deal ecosystem "more meritocratic", because a deal that performs gets selected. Kartal Goksel, chief technology officer of Seedtag, welcomed it: "Activating on performance signals instead of relationship inertia flips the incentive in exactly the right direction."

Three months later, on July 22, the feature became a default. Amazon announced ten capabilities for its Brand+ and Performance+ products, four of them on by default, including automatic deal selection for new Brand+ streaming TV prospecting lines. PPC Land's analysis, published on July 26, ran under the headline "Amazon kills single-deal STV buying in Brand+ as 10 AI features land". Goksel explained the change: "Previously, Brand+ campaigns ran on a single, pre-configured deal." Now they would run on "a continuously optimized set of deals across Amazon and the open internet." The argument for the change is that a machine with Amazon's data can pick deals better than a person can. The consequence is that the person who used to pick them, and who could be paid for doing so, no longer does.

Who chooses the channels

The second is the Full-Funnel Campaign. Announced in November 2025 and expanded at unBoxed on September 29, 2026, the product collapses sponsored ads, display, video and streaming television into one campaign. The advertiser supplies products, creative and a budget; Amazon's models allocate the money across formats and optimise towards outcomes including new-to-brand customers and long-term sales. Advertisers approve the creative but "cannot select individual channels or specific audiences," Digiday reported, and visibility is offered at the level of formats only. MediaPost added that Full-Funnel Campaigns require a minimum daily budget of $100 and are available to all advertisers in the United States. PPC Land's account of a column by ADOTAT ran under the headline "Amazon's $100-a-day Full-Funnel campaigns hand targeting to AI", and noted that $100 a day comes to about $3,000 a month.

Amit Bhattacharyya, Amazon's vice president of full-funnel products, AI and modelling, put the reasoning plainly: advertisers were already stitching such campaigns together by hand, so "what if we made it really simple, and yet very powerful for advertising?" Amazon claims that, in beta, Full-Funnel Campaigns produced up to 67% higher long-term return on ad spend and 29% lower cost per new-to-brand customer, figures Digiday reported after the third-quarter results. Asked about the opacity, Bhattacharyya told Digiday: "We are figuring out how to give them more transparency on the audience side." The decision about where the money goes, between a search ad, a display ad and a Prime Video spot, is the decision a media planner used to make. In this product, it belongs to Amazon.

Who sees which controls

The third is the tiering of access. The PPC Land report of October 29, 2026 set out who can see what. Advertisers with programmatic experience get both DVA+ and DVA+ advanced. Agencies with existing entity access get both. New advertisers, and advertisers coming from Sponsored Ads, get DVA+ only, and must ask an account manager if they want the advanced controls. The guide updated on November 13 adds that, in the default route, advertisers see only the objectives they are eligible for.

The design protects incumbents. The agencies that already run large Amazon budgets keep their controls, for now, and the legacy mode preserves the old DSP workflow. But the growth in Amazon's advertising business comes disproportionately from new advertisers, including the 80 and 30 net new advertisers Moss credited to football and basketball, and from sellers moving up from Sponsored Products. Those are precisely the advertisers who now enter through the automated tier. Over time, the share of Amazon's display and video revenue that runs inside defaults Amazon writes will grow, not because anyone was moved, but because the newcomers started there. It is the Prime Video method applied to the advertiser base.

Amazon's own descriptions of the product make the intent explicit. Its September 29 announcement of the four-step DVA+ builder, reported by PPC Land, promised "same infrastructure, same signals, same measurement" and "no DSP expertise required." Kelly MacLean told MediaPost: "With DVA+, we have brought the power of programmatic to advertisers of all sizes." In January 2025, she had described the DSP to Digiday as "the largest AI-driven buying platform for all digital marketers." The expertise that is no longer required is, in most cases, the expertise an agency sold.

The names keep changing

The rename is part of the pattern. Amazon has changed the name of its main buying interface twice in under a year: from the DSP and the Ads Console to Campaign Manager on December 3, 2025, when Kelly MacLean said that "advertisers of all sizes can now very simply manage campaigns across the funnel through one global entry point", and to Ads Agent on September 29, 2026, when the DSP itself became DVA. AdExchanger's headline was "Amazon Ads And The Amazon DSP Get An Agentic Rebrand".

The closest precedent is Google's. On June 27, 2018, Google renamed AdWords to Google Ads after almost 18 years, folded its DoubleClick advertiser products into Google Marketing Platform and introduced Display & Video 360 as the consolidated buying product. Digiday reported that the same announcement made Smart Campaigns the default for new advertisers. Google renamed its demand-side platform and made automation the default for newcomers in a single move, eight years before Amazon did the same.

The single global account fits the same frame. On July 30, 2026, Amazon began moving existing DSP accounts into one advertiser account across 34 countries: three in North America, Brazil, 17 in Europe, seven in the Middle East and six in Asia Pacific. PPC Land's headline asked whether the agency was the next thing it would remove. A single login is a convenience for a global brand. It is also the end of the regional account structures that local agencies managed, and a single place in which Amazon can present its defaults to an advertiser in every market at once.

The data only Amazon holds

Every argument Amazon makes to advertisers comes back to the same asset: it sees the purchase. Its measurement products are built on that fact, and the history of those products is, in effect, a history of Amazon defining the metrics by which its own advertising is judged.

The earliest of them is new-to-brand. Introduced in January 2019, the metric counts a purchase as new-to-brand if the shopper has not bought from the brand on Amazon in the previous 12 months, and it looks only at Amazon purchases. PPC Land's Decoded entry, published on November 13, 2026, works through an example in which $5,000 of spend produces 400 purchases worth $16,000, 240 of them new-to-brand and worth $9,000, for a return on ad spend of 3.2 and a new-to-brand return of 1.8. Three of the eight metrics in Amazon's new benchmark tool are new-to-brand metrics, and Full-Funnel Campaigns optimise towards them. A metric that started as a reporting column has become an optimisation target.

Amazon Marketing Cloud, the clean room in which advertisers can query Amazon's signals against their own, has been extended steadily. AWS announced on July 31, 2024 that it had become generally available on AWS Clean Rooms, linking Amazon's advertising data to its cloud business. In 2026, Amazon extended Marketing Cloud's lookback window from 13 months to 25. Its own paid audience and retail purchase features became free to query until the end of 2026, with no published price from January 1, 2027. On November 4, 2025, Amazon unified in-market audience definitions across 32 countries. On May 15, 2026, it opened self-service access to more than 50 third-party measurement products across 18 countries. And in early November 2026, the Attain integration brought purchase data from other retailers into DVA+.

The benchmarks show the same reach in a different form. Amazon's benchmark tool became generally available on May 18, 2026 in 18 countries, built on a white paper published on March 27 that capped any single brand at 49% of a benchmark cohort and required at least five brands. When PPC Land read the help page updated on November 11, it found eight metrics, each with a favourable and an unfavourable set of suggested actions, and that none of the 16 ever suggests cutting budgets. Where cost per click sits below the benchmark, the page advises: "Test increasing bids to capture more premium placements." Where cost per thousand impressions is below the benchmark, it advises: "Test increasing bids to access premium inventory while maintaining efficiency." Amazon adds that the suggestions "serve as general guidance to help inform" strategy "but are not exhaustive."

Then there is the impression itself. PPC Land reported on November 4, 2026 that Amazon will drop off-screen ads from impression counts and expects 63% fewer. From January 1, 2027, display impressions on Amazon's own store will be counted only when the ad appears on the shopper's screen, rather than when it starts to download. The change covers homepage, product detail, search results and order pages, and Amazon expects counts to fall by about 63% on average, with larger drops lower on the page. Saul Jacobs, chief marketing officer at Elevate33, observed that "it's definitely a departure from IAB standards." By PPC Land's arithmetic, a $5.00 cost per thousand impressions would become $13.51 if spend and clicks stayed the same. The change is in some ways a gift to advertisers, who will no longer be counted as having bought impressions nobody saw. It is also a reminder of who decides what an impression is on Amazon.

The reporting migration completes the picture. When unified reporting reached general availability on June 8, 2026, it brought every Amazon ad product into one schema of 156 dimensions in 11 categories. On PPC Land's count, 137 of them carry the DVA+ marker, 85 apply to Sponsored Products, 81 to Sponsored Display and 73 to Sponsored Brands, and 59 apply to DVA+ alone. "When a dimension does not apply to an ad product, its value appears as null," Amazon's documentation says. The old report pages go read-only on December 17, and "Sponsored ads reports and Display, Video, and Audio campaigns reports are shutting down on January 31, 2027." Only active saved or scheduled reports from the past 24 months will migrate, and ten report types, including pricing transparency, conversion path and Amazon Attribution reports, cannot migrate yet. An agency that has built its client reporting on the old exports has about two and a half months to rebuild it.

The auction case

The most serious challenge to Amazon's advertising business has nothing to do with agencies, but it goes to the heart of the trust the second phase requires. On August 31, 2026, the Federal Trade Commission and 22 states filed a 181-page complaint in the Western District of Washington, case 2:26-cv-03097, alleging about $20 billion in hidden surcharges on roughly 1.2 million US advertisers, more than 500,000 of them small and mid-sized businesses. The alleged mechanism is a "soft reserve" price added on top of a second-price auction and capped only by the winning bid. The FTC alleges that the share of clicks billed at the advertiser's own bid rose from about 4% in late 2020 to 79.1% in 2024, after years in which advertisers were told they would pay a penny more than the next-highest bidder. Among the names the complaint says Amazon considered for what it alleges was a surcharge was "profit extraction lever"; the team settled on "performance premium."

Amazon disputes the case. It says that "in no scenario does an advertiser pay more than their bid," that Sponsored Products cost per click stayed flat in inflation-adjusted terms between 2019 and 2024, that relevance-weighted auctions saved advertisers more than $8 billion between 2021 and 2025, and that about 92% of selected Sponsored Products ads in 2024 were not the highest bid. No liability has been found. The FTC had opened consumer protection inquiries into both Amazon and Google over search advertising pricing disclosures in September 2025, so the case did not come without warning.

The case covers Sponsored Products, Sponsored Brands and Display Ads bought through the console. AdExchanger's Anthony Vargas noted on September 3 that the alleged behaviour applies to ads bought through the console, not to Amazon DSP. But the questions it raises apply to any system in which the seller of the inventory, the operator of the auction and the reporter of the results are the same company. In Digiday's account on September 8, an agency executive said buyers were "probably the frogs in the boiling water", and another said auction type is not passed in log-level data roughly 70% of the time.

The connection to DVA+ is direct, even though the case does not concern it. An automated tier in which Amazon picks the deals, allocates the budget across channels, writes the benchmarks, counts the impressions and reports the results asks advertisers to accept Amazon's judgment at every step. The FTC complaint is an allegation, not a finding. But it describes what can happen when a platform's incentives and its customers' interests diverge inside a system the customer cannot inspect, and it arrived four weeks before Amazon made such a system the default.

The regulatory perimeter

The auction case is the largest legal challenge to Amazon's advertising business, but it is not the only one, and the others are worth listing because several of them touch the same mechanisms the second phase relies on: defaults, disclosures and data.

The Prime membership itself, the gateway to Prime Video's ad-supported audience, has already cost Amazon a settlement. In the third quarter of 2025, Amazon recorded a $2.5 billion settlement with the FTC in the case over how Prime memberships were sold and cancelled, PPC Land reported on October 31, 2025. That case concerned consumer sign-up flows, not advertising, but it is the same membership whose terms Amazon changed in January 2024 to add advertising by default.

The FTC's interest in how Amazon prices its ads predates the auction complaint. In September 2025, the agency opened consumer protection inquiries into both Amazon and Google over search advertising pricing disclosures. In Germany, the Munich Regional Court's December 2025 ruling on the Prime Video email showed that consumer law can reach the way a default is introduced, even where the default itself is permitted.

Taxes are a separate category, and they are often confused with fees. Amazon passes on digital services taxes to advertisers in several countries as line items it calls Regulatory Advertising Fees; PPC Land reported in August 2024 that it was extending them to Canadian advertisers from August 15. These charges reflect government levies on digital advertising revenue. They are not a platform fee and say nothing about Amazon's own margin.

European AI law now reaches the store. Following the AI Act's general application on August 2, 2026, PPC Land reported that the EU AI Office has opened three complaint routes, one of which covers AI systems embedded in designated platforms, a list that names the Amazon Store alongside Facebook, Instagram, YouTube, Google Search, Bing, LinkedIn and TikTok. An automated campaign builder and a shopping assistant that places sponsored prompts are the kind of systems such routes could one day be asked to examine.

The supply partnerships bring their own exposure. On September 13, 2026, Florida sued Netflix over children's data, seeking up to $50,000 per violation of the state's Digital Bill of Rights, and named Amazon DSP among the recipients of behavioural data, alongside other DSPs and data companies. And in Japan, the Fair Trade Commission approved a plan under which Amazon Japan faces a five-year monitor over a rule that pegged sellers to rival prices for the Featured Offer, the slot that decides which seller's offer a shopper sees first and therefore which offer Sponsored Products ads promote.

Sponsored Products: the engine underneath

For all the attention paid to Prime Video, sports and the DSP, the largest part of Amazon's advertising revenue still comes from the least glamorous product: the sponsored listing on a search results page. When fourth-quarter 2023 advertising revenue rose 27%, Olsavsky said that "the strength in advertising was primarily driven by sponsored products", even in the quarter Prime Video ads began. The drivers Amazon named in 2025 and 2026 kept returning to the same place. Sponsored ads are where most sellers start, where the auction sets the price and where the FTC's case is focused.

The price of those listings has been rising for most of the period covered here. In 2018, Merkle data put the year-on-year rise in Sponsored Products cost per click at 65%. During the pandemic, Merkle saw a 118% increase in cost alongside a 70% rise in Sponsored Products sales. The FTC alleges that the share of clicks charged at the advertiser's own bid rose from about 4% in late 2020 to 79.1% in 2024; Amazon replies that cost per click was flat in inflation-adjusted terms between 2019 and 2024. Amazon's own educational material has not settled the question. PPC Land reported in October 2026 that Amazon Ad Lab names four bidding metrics but publishes no test results: an October 15 video compared fixed and dynamic up-and-down bidding with a hypothetical example and concluded that "neither strategy is universally better." Up-and-down bidding can raise bids by up to 100%, and placement adjustments by up to 900%. The video had 55 views.

The second phase looks different in sponsored ads because the customer is different. Most Sponsored Products advertisers are sellers who manage their own campaigns, or who pay specialist agencies and software tools to do it. For them, the defaults arrive as conveniences. Free AI prompts for Sponsored Products and Sponsored Brands opened in beta on November 11, 2025 and became billable under cost-per-click bidding on March 25, 2026. Advertising costs began deducting from seller proceeds on April 15, 2026, so that Amazon collects payment before the seller receives its money. On June 8, the same day unified reporting became generally available, an off-Amazon serving column arrived in Sponsored Products bulksheets, a sign that a product built for Amazon's own search pages now runs beyond them. On August 10, Sponsored Products campaigns began appearing inside creator content. And in November, Sponsored Products ads began appearing in Alexa for Shopping's text replies.

Each of those changes extends the reach of a sponsored listing without asking the advertiser to do anything. A seller's campaign that was built for Amazon search can now appear in a chatbot answer, a creator's video, a text message thread and, through the off-Amazon toggle, outside Amazon's own properties. That is good for sellers who want reach. It also means that the meaning of a Sponsored Products budget is being redefined by Amazon, one placement at a time.

The boundary between sponsored ads and the DSP is also dissolving. Digiday's Ronan Shields reported on September 29, 2026 that DVA+ consolidates Sponsored Display, Sponsored TV and programmatic display, video and audio buying into one default campaign setup, running on Performance+ for sales and conversions or Brand+ for video branding and prospecting. Buyers who want control over targeting, deals, supply sources, frequency and bidding must switch on a toggle labelled advanced settings. Kelly MacLean described the approach as "human-led AI supported." A seller who used to buy Sponsored Display now meets DVA+, and, as a Sponsored Ads advertiser, gets the default tier only. The unified reporting schema reflects the same merger: 85 of its 156 dimensions apply to Sponsored Products, and 137 carry the DVA+ marker.

New surfaces: the chat, the text message and the order page

While the buying tools were being automated, the places where Amazon can show an advertisement multiplied. Some of the new surfaces are conversational, some are transactional, and some sit on pages that used to carry no advertising at all.

The largest is the shopping assistant. Amazon's Rufus assistant had 250 million users by the third quarter of 2025, PPC Land reported. In August 2026, Amazon merged Rufus and Alexa+ into a single shopping assistant serving 350 million trailing-twelve-month users, and by November it was claiming that more than 400 million customers had used Alexa for Shopping in the previous 12 months. On November 6, 2026, a small invite-only group of US customers began texting the assistant to reorder household items or build a shopping list from a recipe photograph. Digiday's Krystal Scanlon reported that Amazon's shopping agent now takes orders by text, with Sponsored Product ads in the replies. The ads appear only when the agent offers options; purchases the agent makes on its own stay free of advertising, and Amazon is not yet charging for the placements.

Inside the assistant, Amazon has built two formats. Sponsored prompts, short suggested questions that a brand pays to place, have been billable since March 25, 2026, as described above. Jassy told analysts in July that "shoppers who click a sponsored prompt convert to a sale 48% more often and spend 21% more on average." At unBoxed in September, Amazon added Branded Conversations, a multi-turn unit inside Amazon's chatbot surface. Amjad Jbara, director of applied science for sponsored products, described it as creating "multi-turn shopping conversations to help shoppers find the right products informed by brand-provided inputs." Advertisers steer it with what Amazon calls Conversational Briefs, but cannot choose which reviews the AI surfaces.

Reviews themselves became an advertising product. PPC Land reported on November 12, 2026 that Amazon's Ads API had gained review request ads, billed only when buyers rate. The product, in US beta, asks shoppers to review items with fewer than 1,000 ratings; Amazon "pays only when a shopper leaves a rating or review," in its own words, although no price was published. It ran a closed beta from June 3 to 21, 2026, and Amazon claimed a threefold improvement on a 19-day comparison. The Federal Trade Commission finalised its rule against fake reviews on August 14, 2024, which is why the distinction between asking for a review and paying for one matters.

The order page is now inventory too. On September 29, 2026, Amazon and Yelp launched Sponsored Services, putting ads for plumbers, electricians, landscapers, home cleaners, roofers and movers onto product detail pages, order confirmation pages and package tracking pages in the United States, with Call and Quote buttons that keep the interaction inside Amazon. Amazon's low-price storefront, Haul, is moving closer to the main store: PPC Land reported on November 9 that Haul's buyer count had grown by nearly 350%, its selection had grown 25-fold in two years to more than 10 million products, and that "relevant Amazon Haul products now surface within Amazon search results," where they compete with sponsored listings. And on November 4, Amazon launched the Shop Direct Merchant Workspace, through which any eligible store can list in Amazon's search for free, even if it does not sell on Amazon. "There are no fees to participate in Shop Direct, and merchants can opt out at any time," Amazon said. Every new listing in Amazon search is another product that may one day want to advertise there.

Television, too, is becoming a shopping surface. Amazon made Alexa+ free on compatible US Fire TV devices on August 19, 2026, and said that users select the assistant's top recommendation 40% more often than with the original Alexa. A recommendation that is accepted more often is worth more to whoever can pay to influence it.

Amazon has also defended the perimeter of its store. It went to court to keep Perplexity's Comet browser assistant out of Amazon shopping accounts, and lost: on August 4, 2026, a federal appeals court vacated the preliminary injunction that had kept the assistant out, finding that all four injunction factors went against Amazon. Agents that shop on Amazon without seeing its sponsored listings are the most direct threat to the business described in this special. For now, Amazon says, they are not a material one. Jassy told analysts on October 29: "Few customers are doing much shopping this way yet."

There is a symmetry with where this story began. In March 2020, Amazon cut its spending on Google ads at a moment when it could not deliver much of what those ads promoted. In July 2025, it left Google's Shopping auctions altogether. PPC Land reported that Amazon had exited Google Shopping globally, and that cost per click collapsed by around 30% for the retailers left bidding. The company that had once been one of Google's largest shopping advertisers had become one of its largest advertising competitors, and no longer needed to pay Google for traffic to its own store.

What Amazon claims for its automated products

Much of the case for the defaults rests on performance figures that Amazon has published itself. They are summarised below. All are Amazon's own figures as reported; none was accompanied by an independent audit in the linked coverage, and several rest on test periods that predate the products' launch.

The pattern in the table is worth stating plainly. Each claim compares an automated product with a manual alternative, and in each case the automated product wins. That may well be true. Amazon has more data about its shoppers than any agency can bring to a campaign. But the claims are made by the company that sells the automated product, measured by the company's own metrics, on the company's own store. The benchmarks that tell advertisers whether they are doing well are written by the same company, and none of their 16 suggested actions involves spending less.

What the buyers say

The people who buy on Amazon are not unanimous, and their views have shifted as the second phase has become visible. In the spring of 2025, the dominant tone was enthusiasm about price and access. Pacvue's Clare Hogan told Digiday in February 2025 that Amazon had removed the barrier of entry to the DSP. By the autumn of 2026, the tone had become more guarded.

Destaney Wishon of btr media, quoted by PPC Land in October, recommended that advertisers who did not own their DSP seat make the transition to owning it "sooner rather than later." Vlad Chubakov of Delve Deeper explained Amazon's low margins by its ownership of the pipes. Brian O'Kelley, the programmatic pioneer now at Apostra, was quoted in Ronan Shields' Ad Tech Briefing for Digiday on October 13, during Advertising Week, saying that "Amazon is acknowledging that a programmatic-only DSP makes little sense outside a web-centric universe." Sarah Caputo asked what DSP buyers need to know about auction disclosure. And the independent agencies Digiday interviewed in October wanted negotiating power they did not have.

Some buyers are responding by building their own layer above the platforms. Olyzon, a Paris company, put its agentic orchestration platform into full commercial release on October 7, 2026. Its agents plan, activate, measure and move connected TV budgets across The Trade Desk, Amazon DSP, DV360 and Yahoo DSP, while buyers keep their own seats and deals and every action is written to an audit log. "The agent does the orchestration. The human keeps the decision," said Jules Minvielle, its co-founder and chief executive. It is the mirror image of Amazon's approach: automation that sits with the buyer, across platforms, rather than with one platform, inside it.

The money, meanwhile, keeps moving towards Amazon. Tinuiti put DSP-managed spend up 25% to 30% on 2025, Digiday reported in October. Measured on Madison and Wall's estimates, Amazon's share of US advertising revenue rose from 8% to 9% in 2025. Luke Stillman of Madison and Wall told Digiday that the three largest platforms are "just going to outperform the average every year." The buyers who worry about Amazon's defaults are, in most cases, spending more with Amazon than they did a year ago.

Advertising and the AI bill

There is a financial reason why the second phase is arriving now, and it has little to do with agencies. Amazon is spending more on artificial intelligence infrastructure than at any point in its history, and advertising is one of the few businesses inside the company that throws off cash without needing new data centres to do it.

The scale of the spending is set out in Amazon's own statements. In its fourth-quarter 2025 release, the company said it expected about $200 billion in capital expenditures in 2026, after the roughly $125 billion it had forecast for 2025. PPC Land's analysis of the first-quarter 2026 results, published on May 1, found that trailing twelve-month free cash flow had fallen to $1.2 billion, from $25.9 billion a year earlier, that long-term debt had risen from $65.6 billion to $119.1 billion, and that Amazon had raised $53.8 billion of debt in March 2026. The same quarter included a $16.8 billion non-cash gain on Amazon's investment in Anthropic and a $15 billion investment in OpenAI. In the second quarter, Amazon reported $53.4 billion of non-operating income, "primarily from the company's investments in Anthropic," which on PPC Land's calculation was more than two and a half times its advertising revenue in the same period. AdExchanger's Joanna Gerber summarised the third quarter under the headline "Unrealized AI investments drag Amazon's income, but AWS and ads are still growing".

Sources: PPC Land on Q4 2025, Q1 2026, Q2 2026 and Q3 2026; Marketing Dive on Q4 2025 net sales. Shares are Landline calculations.

The table shows two businesses moving at different speeds. Amazon Web Services grew 39.3% in the third quarter of 2026, accelerating from 28% in the first quarter on demand for AI computing. Advertising grew 10.9% in the same quarter, distorted by the Prime Day shift, and about 18% across the second and third quarters combined. AWS is the business that needs the capital. Advertising is the business that does not: the marginal cost of an extra sponsored product placement on an Amazon search page, or an extra advertising minute in a Prime Video episode already licensed, is close to zero. Amazon does not disclose advertising profit, but advertising revenue of nearly $20 billion a quarter, earned largely on Amazon's own pages and screens, is the kind of income that helps carry a $200 billion capital programme.

That context explains several of the choices described in this special edition. A demand-side platform that charges almost nothing on Amazon's own inventory makes sense if the purpose of the platform is to fill that inventory, at prices set by Amazon's auctions, rather than to earn a fee. Automated defaults that steer budgets towards Amazon's own media serve the same end. And the cloud connections run in both directions: Amazon Marketing Cloud runs on AWS Clean Rooms, the Prime 360 replays run on Trainium chips, and AWS Elemental now fills unsold live ad breaks with Amazon demand.

The fourth quarter will test the arithmetic. Amazon has guided advertising growth to the "high teens" and total net sales of $240 billion to $247 billion, with the NBA season, Thursday Night Football and the holiday shopping period all in the quarter. Growth in the high teens on the $21.3 billion of the fourth quarter of 2025 would mean roughly $25 billion in the quarter and about $82 billion for 2026 as a whole, on Landline's arithmetic, against $68.6 billion in 2025. The figure matters less than the trend it confirms: an advertising business adding more than $10 billion of revenue a year, inside a company that is borrowing to build data centres.

Three routes for one advertising dollar

The clearest way to see what the second phase changes is to follow a single display or video budget through the routes Amazon now offers, and to ask, at each step, who makes the decision. The table below draws on the Support Center pages and coverage cited throughout this piece, and leaves out DVA+ advanced, the middle tier with more controls, because Amazon has published little about it beyond saying that it is limited to some US advertisers. Where a cell says "not stated", Amazon's published material does not answer the question.

Read across a row, the pattern is consistent. Each step to the right removes a decision from the buyer and gives it to Amazon. Read down the agency fee row, the pattern is also consistent: the further right a budget sits, the less Amazon's documentation says about where an agency fits. None of this prevents an agency from advising a client who uses DVA+ or Full-Funnel Campaigns, writing the brief, producing the creative or invoicing for its time. It does change what the agency is paid for. An agency used to be paid, in part, for doing the buying. On the default route, Amazon does the buying.

The legacy fee setting shows how the transition is being managed. Campaigns that keep the old per-campaign fee will not receive new features, and the setting disappears entirely by the end of March 2027. The new features, meanwhile, arrive in the automated tiers first: automatic deal selection, AI-generated creative, Full-Funnel optimisation. An agency that wants its clients to have the newest tools must move them onto Amazon's new fee structure, and a client that wants the newest tools without an agency can find them in DVA+ with no minimum spend.

The case against the thesis

The argument of this special edition is that Amazon is using defaults to move control of budgets from agencies to itself. There is a reasonable case that this overstates what is happening, and it deserves to be set out in full.

First, much of what has been announced is not yet in general use. The conversational Ads Agent experience is in closed beta in 25 countries, DVA+ advanced settings are limited to some US advertisers, and DVA+ itself is not due to reach general availability in 35 markets until the first quarter of 2027. Amazon's own pages disagree about which countries keep the old DSP experience. PPC Land found that the November 13 DVA+ guide gives two different sets of launch instructions. A strategy that is still being documented inconsistently is not yet a strategy that has changed the market.

Second, the incumbents keep their controls. Agencies with existing entity access get both DVA+ tiers. The legacy mode preserves the old workflow. Advertisers with programmatic experience see the advanced settings without asking. If Amazon wanted to force agencies out, it has chosen a slow and visible way to do it.

Third, the largest television budgets are still negotiated by people. Moss's one-word answer to whether AI agents would handle upfront negotiations, "Never," describes the part of Amazon's business that grew fastest in 2024 and 2025. Sports inventory that sells out in a quarter is sold through relationships, and those relationships run through agencies and their holding companies.

Fourth, Amazon's agency fee changes can be read as better tooling rather than worse. Five calculation methods, four levels of application and a rule that the most specific fee wins give agencies more ways to express their margin than the old per-campaign setting did. "Amazon doesn't bill agency fees" means that Amazon takes no share of them. "Advertiser accounts can't set agency fees" means a client cannot strip its agency's margin out from inside its own account. Both statements protect the agency as much as they constrain it.

Fifth, Amazon is far from dominant in programmatic buying. Guideline's data puts it at just under 20% of tracked programmatic spend. Netflix added The Trade Desk as a buying platform in November. Pinterest sells through four DSPs. Orchestration tools such as Olyzon let buyers keep seats across several platforms and move money between them. The Trade Desk's revenue fell 0.5% in the third quarter, but its customer retention stayed above 95%.

Sixth, the business is not immune to the calendar or to competition. Third-quarter advertising growth slowed to 10.9%, measured US Prime Day advertising spend fell 8.8% and Netflix's partner integrations, including Amazon's, slipped. Walmart Connect is pitching brands directly and growing faster than Amazon in percentage terms. AI shopping agents run by other companies, which Jassy says are not yet a factor, could become one, and a federal appeals court has refused to let Amazon keep one of them out.

Seventh, and most important, advertisers are choosing to spend more. Buyers reported Amazon spend up 15% to 40% year on year. If the defaults produce better results, as Amazon claims, then moving decisions from agencies to Amazon is what advertisers want, and the agencies' value will shift to strategy, creative and cross-platform planning, where it arguably always belonged.

Each of these points has force. Together, they suggest that the second phase is a direction rather than a fait accompli, and that agencies retain real leverage. They do not change what the direction is. Every product Amazon has launched since January 2025 has moved a decision from the buyer to the platform, and every default it has set has favoured its own automated tier. The beta labels will come off. The legacy settings have end dates. The newcomers enter through DVA+.

Who said what

The arguments in this special edition are made, in the end, by people, and their own words trace the arc from the first phase to the second better than any summary. The table collects the lines quoted in the coverage cited above, in date order.

Read in order, the lines describe a company that first explained why it was different from other media owners, then claimed the audience to prove it, then cut the price of buying that audience to almost nothing, and finally began to describe its tools as making decisions on its customers' behalf. The outside voices move in the opposite direction, from curiosity about where Amazon fits to questions about who sets the price.

A timeline, 2017 to 2027

The table below brings together the dates in this special edition, from Amazon's first public hints about its advertising revenue to the deadlines that fall in the first quarter of 2027.

The dates ahead

Several of the changes described here have deadlines in the next five months. For anyone with budgets on Amazon, these are the dates that matter.

What the defaults decide

Amazon's advertising business took a little under ten years to grow from less than $3 billion to $78 billion a year. The first phase of that growth was built on things Amazon had and others did not: a store where shoppers search with the intent to buy, a record of what they bought, a streaming service with a large audience it could switch to advertising overnight, and a demand-side platform cheap enough to make every other buying tool look expensive. In that phase, Amazon needed agencies, because agencies controlled the budgets it wanted, and it courted them with low fees, partner badges and Prime Video inventory.

The second phase is built on something different. It is built on the money already inside Amazon's systems, and on the observation, first tested on Prime members in January 2024, that most people accept whatever is pre-selected. In this phase, Amazon writes the defaults: which products a campaign promotes, which objectives an advertiser can see, which deals are bought, which channels receive the budget, how impressions are counted, which benchmarks define success, where the agency's fee is calculated and which reports survive. Each default can be defended. Each is presented as a simplification. Together, they move the decisions that agencies used to be paid for into software that Amazon owns.

None of this is irreversible, and much of it is unfinished. The FTC case could change how Amazon's auctions are disclosed. A court in Munich has already questioned how Amazon changed the terms of a contract with its customers. Buyers are building tools that sit above the platforms. And the largest budgets are still negotiated by people across a table. But the direction is clear from the plumbing, if not from the press releases. On November 13, 2026, a help page told advertisers that a campaign could include 20 products, that they would see only the objectives they were eligible for, and that budget caps could cause under-delivery. That is what a platform says when it expects to make the decisions, and expects its customers to approve them.

Did you know

Amazon DSP, the name buyers still use for Amazon's demand-side platform, officially became DVA, for display, video and audio, at unBoxed on September 29, 2026. PPC Land's Decoded entry, published on November 13, notes that Amazon's managed service, in which Amazon's staff run the campaigns, "typically requires a minimum spend of USD $50,000", varying by country, and that DVA+ is due to reach general availability in 35 markets in the first quarter of 2027.

Principal media is "the practice in which agencies purchase media inventory themselves and resell it to clients." The Association of National Advertisers based its March 19, 2026 report on a survey of 114 marketers conducted between October 10 and December 19, 2025, and Forrester estimated in January 2026 that the practice accounts for about a third of agency billings.

New-to-brand is Amazon's label for a purchase by a shopper who has not bought from the brand on Amazon in the previous 12 months. Amazon introduced it in January 2019, extended it to Sponsored Display on October 14, 2021, and on May 6, 2025 began letting Sponsored Brands advertisers in 21 countries raise bids automatically for new-to-brand shoppers.

Supply cost, in Amazon's definition, "is the cost of the media the ads run on." It is the base for four of the five agency fee methods in Amazon's new manager-account system. PPC Land's Decoded entry points out that a 3.5% markup on $10,000 of supply cost, a $350 fee, works out at about 3.38% of the client's combined spend of $10,350.

By the numbers

20
Maximum number of products in one automated DVA+ campaign, under Amazon's November 13 guide.
Amazon caps its automated DVA campaigns at 20 products

$362.69
Fee Amazon's own percentage-of-budget formula produces on $10,000, beside a $350 markup example.
Agencies lose Amazon's legacy DVA fee setting after January 31, 2027

16
Benchmark recommendations on Amazon's help page, none of which suggests cutting a budget.
Amazon's 16 benchmark recommendations never suggest cutting budgets

Nearly 350%
Growth in Amazon Haul's buyer count as its $0.11 holiday deals opened.
Amazon Haul buyer count gains nearly 350% as $0.11 holiday deals open