Nectar, a New York-based Amazon advertising agency, on October 14, 2026 released Customer Atlas, a service that scores a brand's Amazon purchase history against every one of the 6,728 audience segments Amazon exposes and then uses the results to raise or hold back bids in Sponsored Products and Sponsored Brands search campaigns.

In Short

An agency called Nectar now looks at who has bought a brand's products on Amazon and works out which of Amazon's thousands of shopper groups those buyers tend to belong to. It then tells Amazon to bid more on search ads when a shopper from a promising group searches, and not to chase shoppers from groups that tend to buy once and disappear. Amazon already allowed this kind of audience bid boost in search, so the change is less a new tool than a claim that most advertisers have not been using the one that exists.

What Nectar released on October 14

The announcement was distributed through Business Wire at 11:28 AM Eastern Daylight Time on October 14, 2026, under a New York dateline. Nectar describes itself in the release as "an Amazon Advanced Advertising Partner managing more than $2.5 billion in annual client sales." Customer Atlas, according to the company, is "an Amazon Marketing Cloud capability that changes the unit of targeting in Amazon search advertising from the keyword to the shopper."

The argument behind it is short. Amazon search campaigns bid on words. A brand sets a price for a category term and, according to Nectar, "pays approximately that price for every click it buys, whether the shopper is a likely high-value buyer or someone who will never purchase at that price point." AMC, Amazon's clean-room analysis service, made it possible to tell those shoppers apart. But in practice, Nectar argues, that capability "has been aimed almost entirely at Amazon DSP, which for most advertisers is the smaller half of their spend. Customer Atlas points it at the larger half."

Nectar does not publish the split between search and display spend that underpins "the larger half," either for its own clients or for the market. Amazon does not break it out either. When Amazon reported second-quarter 2026 results, advertising services revenue reached $19.8 billion, up 26%, and Andy Jassy called Sponsored Products "our largest offering," without giving figures for Sponsored Products, display or the DSP separately.

The company lists three parts to the service.

A full segment index

The method "analyzes up to five years of a brand's Amazon purchase history and indexes it against all 6,728 audience segments Amazon exposes, 5,268 in-market and 1,460 lifestyle, rather than selecting the few category segments a media plan would reach for," according to Nectar.

A statistical adjustment sits inside that index. Repeat buyers, according to the release, "belong to roughly 10% more segments by default, an adjustment that otherwise manufactures a false signal in every segment measured." The logic is that a loyal customer who shops Amazon often accumulates more segment memberships simply through activity. Without a correction, almost any segment would appear to over-index among a brand's best customers. Nectar does not explain how the 10% figure was derived, whether it varies by category, or whether it is measured per brand.

Two lists, not one

The index produces two outputs. One is a set of segments that concentrate among a brand's high-value buyers. The other is a set of segments that concentrate "among people who buy once and never return," according to Nectar. "Most media plans carry the first. Almost none carry the second, which means most brands pay full rates to reach shoppers their own data already identifies as unlikely to return."

That second list functions as a suppression audience, though in search it does not exclude anyone outright. Under the activation model Nectar describes, shoppers in the low-value segments simply do not receive a bid boost.

Both lists deploy "as audiences in Amazon DSP and as audience bid modifiers in Sponsored Products and Sponsored Brands, using a low base bid with a boost applied only when a targeted shopper searches," according to the release. "No new campaigns are required, and the budget is already committed."

That last sentence describes the commercial pitch as much as the mechanics. The change sits on top of existing Sponsored Products and Sponsored Brands campaigns and does not ask a brand to find new money.

The 24.9 million shopper segment

The release offers one example of what the method produces. "In one engagement, the largest audience segment available in the category reached 24.9 million Amazon shoppers and showed no relationship at all to repeat purchase, while far smaller segments predicted it strongly," according to Nectar.

The client, the category, the segment's name and the smaller segments that performed better are not disclosed. Nor does the release define what "predicted it strongly" means in numbers. As a description of the problem Customer Atlas is meant to address - that the obvious, largest category audience may be the least useful one - the example is clear. As evidence, it rests on a single unnamed case.

The Amazon features underneath

Customer Atlas is an agency method layered on Amazon's own tools, and most of those tools have existed for some time.

The capability to raise search bids for AMC audiences was announced by Amazon in October 2024. At unBoxed that year, Amazon said advertisers using AMC could boost bids in Sponsored Products and Sponsored Brands for audiences that had seen their streaming TV ads or visited their website. Nectar's release does not mention this history. Its framing that audience targeting "got classified as a display tactic years ago and nobody went back to check" is a characterisation of how advertisers used the product rather than of what Amazon offered.

The data access has widened since then. In September 2025, Amazon opened AMC to sponsored ads advertisers directly inside the Amazon Ads console, with no-code templates and without a partner or account representative. AMC custom audiences at that time were activated through Amazon DSP. In June 2026, Amazon made first-party paid feature signals in AMC free to query until December 31, 2026, covering Amazon Audience Segment Insights and Amazon Retail Purchases, both for audience creation and measurement. Amazon has not said whether charges resume on January 1, 2027.

The segment library itself has also grown. In November 2025, Amazon extended in-market audiences in its DSP to 32 countries, without disclosing how many behavioural segments the library contained. Nectar's count of 6,728 - 5,268 in-market and 1,460 lifestyle - is therefore the company's own figure for what Amazon exposes. The release does not say in which country, on which date, or through which interface the count was taken.

Scoring a customer base by value is not new inside AMC either. Amazon released a High Value Audiences solution in AMC in 2024, which grouped customers into percentile buckets by historical spend and made audiences available for activation on Amazon DSP, typically within 48 hours.

What Nectar adds, on its own account, is the scope (every segment rather than a shortlist), the correction for repeat buyers' segment inflation, the explicit second list of segments to avoid, and the decision to route both lists into search bids rather than only into display.

Five years of history

One detail in the release does not obviously match what PPC Land has reported about AMC. Nectar says the method analyzes "up to five years" of purchase history. Amazon extended AMC's lookback window from 13 to 25 months on November 11, 2025. The release does not say which dataset supplies the longer history, whether it comes from AMC paid features, from the brand's own sales records, or from Nectar's analytics platform. That matters for anyone trying to reproduce the analysis, since the advertised control cell is meant to be something "any advertiser could run without Nectar."

The cost-per-click claim

Nectar reports one performance result. "In the first live deployment, campaigns using audience bid modifiers cut cost per click by a third against the same brand's standard campaigns," according to the company.

That is Nectar's own figure. The release names no brand, category, time period, spend level, click volume or statistical test. It reports no conversion rate, return on ad spend, sales or new-to-brand outcome for the same campaigns.

The structure of the bidding model also bears on the number. Under the design Nectar describes, campaigns run "a low base bid with a boost applied only when a targeted shopper searches." Every click from a shopper outside the targeted segments is bought at the low base bid. On PPC Land's reading, a lower average cost per click is close to what such a design would produce by construction, whether or not the boosted clicks convert better. The question that decides whether the method works is what happened to sales and cost per acquisition, and that is not in the release. Neither is any measure of incrementality, which would show whether the boosted shoppers bought because of the ads or would have bought anyway.

The release does describe how future engagements will be tested. "Every Customer Atlas engagement ships with a pre-registered test design: one variable per cell, equal budgets across cells, a control cell any advertiser could run without Nectar, and decision gates written before the first impression serves," according to Nectar. Pre-registration - writing down what will be measured and what counts as success before a test starts - is unusual in agency marketing claims. Nectar has not published any pre-registered design or results from one.

Why search bids are contested ground

The release lands in a period of scrutiny over how Amazon prices sponsored ads clicks. On August 31, 2026, the FTC and 22 states sued Amazon over about $20 billion in alleged hidden ad surcharges affecting roughly 1.2 million US advertising customers. The complaint alleges that the share of clicks charged at the advertiser's full bid rose from about 4% in late 2020 to 79.1% in 2024. Amazon denies overcharging and says advertisers saved more than $8 billion between 2021 and 2025.

That dispute has a direct bearing on any strategy built on bid modifiers. If a large share of clicks clear at or near the advertiser's own bid, then raising a bid for a targeted shopper raises what is paid for that click almost one for one. Holding the base bid low for everyone else does the reverse. Nectar's release does not discuss auction pricing, Amazon's reserve price mechanism, or how audience modifiers interact with Amazon's own automated bidding, which according to the FTC complaint can raise a bid by up to 900% for certain Sponsored Products placements.

The release also arrives as Amazon pushes search and display closer together on its own side. Amazon began testing a Campaign Manager that combines sponsored ads and Amazon DSP in one platform in November 2025. An agency method that sends the same AMC audiences into both channels fits that direction.

Who Nectar is

According to the company's boilerplate, Nectar is "a full-service e-commerce performance marketing agency for mid-market and enterprise brands, managing more than $3 billion in Amazon sales across 200+ brands and $250M in ad spend." It was founded in 2018 by former Amazon sellers and runs an in-house creative studio, marketplace advertising, retail operations and a proprietary analytics platform called iDerive.

The release gives two different figures for the business it manages. The opening paragraph says "more than $2.5 billion in annual client sales." The boilerplate at the end says "more than $3 billion in Amazon sales." The release does not explain the gap, which could reflect different time periods, the difference between annual and cumulative sales, or an outdated boilerplate. PPC Land has not been able to reconcile the two from the document.

Two executives are quoted. "Audience targeting on Amazon got classified as a display tactic years ago and nobody went back to check," said Youval Peltier, Chief Strategy Officer at Nectar. "The question was never which keywords to buy. It was which shoppers are worth paying more for, and the answer is now something you can act on inside search."

Harriet Carson, VP of Advertising at Nectar, framed the service as an overlay rather than a replacement. "It is a change to bids on spend that is already happening," she said. "It works alongside whoever runs the search account today."

That positioning is commercially significant. Nectar is offering Customer Atlas to brands whose search accounts may be managed by another agency or in-house team. Engagements, according to the release, "begin with an audience analysis of the existing account." No pricing, minimum spend or country availability is stated.

What the release leaves open

Several questions remain unanswered by the document. It does not say whether Customer Atlas works outside the United States, though the 6,728 segment count presumably reflects one marketplace. It does not describe how often the index is refreshed, how many segments typically end up on each list, or what size the bid boosts are. It does not say how many brands have used the service beyond the "first live deployment," and it gives no result other than cost per click.

For a sponsored ads manager, the substance of the announcement is the method rather than the tool: a full scan of segments instead of a handful, a correction for frequent shoppers appearing in more groups, and a deliberate list of shoppers not to pay extra for. Amazon's interfaces already allow each step. Whether doing them in that order improves sales, rather than only lowering average click prices, is the part Nectar has not yet shown.

Timeline

Summary

Who: Nectar, an Amazon advertising agency founded in 2018 that says it manages between $2.5 billion and $3 billion in client Amazon sales, with Chief Strategy Officer Youval Peltier and VP of Advertising Harriet Carson quoted. Amazon brands running Sponsored Products and Sponsored Brands campaigns are the intended clients.

What: Customer Atlas, a service that indexes up to five years of a brand's Amazon purchase history against 6,728 Amazon audience segments, corrects for repeat buyers belonging to about 10% more segments, and produces one list of segments to bid up and one of segments to avoid. Both are applied as audience bid modifiers in search campaigns and as audiences in Amazon DSP. Nectar's only reported result is a one-third lower cost per click in one unnamed deployment.

When: Released October 14, 2026, at 11:28 AM Eastern Daylight Time.

Where: Announced from New York through Business Wire. The release does not state which Amazon marketplaces the service covers.

Why: Nectar argues that AMC audience data has been used mainly in display while most Amazon ad budgets sit in keyword-based search, where a brand pays roughly the same for every click regardless of the shopper's value. The method relies on Amazon features available since 2024 and its effect on sales has not been disclosed.