Under-delivery is the gap between what an advertising order was supposed to deliver and what it actually delivered. The shortfall can be measured in impressions, ratings points, views or unspent budget. In a guaranteed purchase it is a contractual failure that the seller must remedy. In an auction campaign it is an operational one: the money stays in the account, but the planned reach never arrives.

The term exists because almost every advertising order is a forecast. A publisher promises 2 million impressions in March before March has happened, and a buyer sets a daily budget assuming the auction will offer enough qualifying users at an acceptable price. When traffic, ratings or targeting disappoint, delivery falls short.

Two kinds of shortfall

The term covers two related but distinct situations.

Guaranteed under-delivery applies to reserved buys: direct insertion orders, sponsorships, upfront television deals and programmatic guaranteed deals. The seller has committed to a volume, usually priced on cost per thousand impressions (CPM), and the shortfall becomes a debt. The standard remedy is a makegood: extra inventory later at no charge, an extended flight or a credit.

Non-guaranteed under-delivery, often called underspend or underpacing, applies to auction buying through demand-side platforms (DSPs), search and social. Nothing is owed. A line item that spends 60% of its budget simply bought fewer impressions, and the advertiser pays only for what ran. The cost is missed reach and lost seasonal windows.

How delivery is paced and where it breaks

Delivery systems spread an order across its flight using pacing rules. In Google Ad Manager, the publisher ad server, a line item's "Deliver impressions" setting defaults to Evenly, which according to Google's help centre "could deliver up to 5% higher than the goal as a safeguard against underdelivery." Frontloaded starts as much as 40% ahead of schedule. Microsoft's ad server sets daily goals at 105% of even lifetime delivery by default, according to its documentation. With a 20,000-impression daily goal and a 100,000 shortfall, its ASAP catch-up setting targets 120,000 impressions the next day.

When delivery still falls short, the causes fall into a small number of categories that every major platform now reports.

  • Insufficient inventory. Google's Ad Manager documentation calls it "the primary reason for underdelivery." Narrow geography, small segments and strict brand-safety filters all shrink the eligible pool.
  • Competition. Line items lose to others of similar priority, or bids fall below floors and rival bids.
  • Frequency caps. Tight caps on small audiences limit volume mathematically.
  • Creative problems. Rejected or paused creatives, wrong dimensions or video with too many VAST (Video Ad Serving Template) wrappers. Ad Manager flags render rates below 80% for display, 50% for video and 30% for apps.
  • Buyer behaviour in programmatic guaranteed. In OpenRTB (Open Real-Time Bidding), the seller sends a must_bid signal set to true when a deal is behind pace. If the buyer's DSP still does not respond, Ad Manager warns that a "low response rate from the buyer can cause line items to underdeliver."
  • Budget and bid constraints. On Meta's platforms, the Marketing API documentation states that "for under-delivery, your bid price might be too low or your audience too narrow," and warns that switching to accelerated delivery will not help in those cases.

Search adds its own vocabulary. Google Ads labels a bid strategy "Limited" when inventory (search volume), bid limits or budget restrict it. Google Ads may spend up to twice the average daily budget on a given day but no more than 30.4 times it across a month, so early-month underspend leaves room the system tries to fill later.

Diagnostics and alerts

Ad Manager forecasts underdelivery before a line item starts, provided it begins within 30 days, ends within a year and has an absolute goal. Flagged items carry an "Expected to underdeliver" status. Once live, a Troubleshoot tab breaks down non-delivery causes.

On the buy side, Google's Display & Video 360 (DV360) shows a waterfall from available requests through targeting, creative, frequency, budget and pacing filters to bids won and lost. Originally a deal troubleshooter, the tool was renamed Troubleshooter in February 2019 and extended to open auction line items. A June 2020 update added which frequency cap in the hierarchy was blocking impressions and losses from bid requests without cookies. The data has gaps: according to Google, DV360 shows nothing for line items with fewer than 200,000 valid queries in 7 days, and some exchanges send no loss reasons.

Amazon introduced private marketplace deal troubleshooting in Amazon DSP on November 1, 2020, explicitly to "analyze under-delivering deals," according to Amazon Ads. A November 8, 2022 release added 15 missed-bid reasons, including "bid below deal floor." In 2024 the DSP's new overview page began to highlight campaigns most at risk of underdelivering on budget goals.

Origin and evolution

Under-delivery is older than digital advertising. Television networks have sold audience guarantees for decades, and when ratings fall short they owe makegoods, often delivered through audience deficiency units (ADUs), defined by Simulmedia as advertising slots provided at no cost to cover a viewership shortfall. As PPC Land's upfront explainer notes, a network that overestimates ratings "owes make-goods it did not budget for."

Digital media codified the practice in the IAB and 4A's Standard Terms and Conditions for Internet Advertising, version 3.0, published in 2010. Section VI requires the media company to warn the agency of likely under-delivery no later than 14 days before the insertion order (IO) ends. If guaranteed deliverables fall short, the parties try to agree a makegood flight; failing that, the agency may take a credit. For cost-per-click, cost-per-lead and cost-per-acquisition deliverables, it states, "guaranteed delivery, even delivery, and makegoods are not available." That carve-out is still the dividing line between the two kinds of shortfall.

Social platforms framed the issue for performance advertisers. Facebook published holiday guidance in November 2017 warning that "some factors can cause your ad sets to under-deliver, leading to lost opportunities," according to Social Media Today. In October 2017, Google also allowed AdWords campaigns to spend up to twice their daily budget.

Programmatic guaranteed buying brought contractual guarantees into automated pipes: DV360 added PG support for FreeWheel, Kargo, Smartclip and Xandr in September 2020, and Amazon introduced Guaranteed Deals for Prime Video Ads in August 2024.

Why it matters

For publishers, under-delivery is a liability: makegoods consume sellable inventory and refunds reduce revenue. For buyers, it distorts planning, since a campaign built to reach 60% of an audience at an average frequency of four reaches far fewer people if half the impressions never run.

Streaming made the problem public. In December 2022, weeks after launching its ad tier, Netflix let advertisers take back money for ads that had not run. According to Digiday, citing five agency executives, Netflix had in some cases delivered "roughly 80% of the expected audience."

Private deals are another chronic source. The Trade Desk said in June 2025 that 90% of deal IDs fail to scale effectively, a figure from internal analysis that has not been independently verified.

Limitations and disputes

Under-delivery is only as clear as the measurement behind it. When the seller's ad server and the buyer's verification vendor disagree, both can claim to be right. The 2026 IAB draft framework addresses this: if the controlling measurement is more than 10% lower than another source and reconciliation fails, the discrepancy is treated as under-delivery under the remedies section, as PPC Land's coverage of the framework describes.

Makegoods themselves are contested. Extra impressions in January do not replace a missed Black Friday weekend. Aaron Goodman of BakerHostetler wrote in August 2024 that makegoods are often inadequate for unique live events. Cliff Campeau of AARM, writing on the ANA's blog in February 2021, after television ratings fell 20% to 30% season-to-date, argued that "a guarantee is a guarantee."

Diagnostics also reflect the interests of the platforms that build them. Most tools list advertiser-side causes such as narrow targeting, low bids or caps, while auction dynamics, internal priority rules and model behaviour are reported in aggregate or not at all. Overcorrection carries its own risk: catch-up pacing pushes spend into the last days of a flight, often at higher prices.

Not the same as

Non-delivery means zero impressions, usually from a disapproval, a paused creative or a broken tag. Under-delivery means some delivery, but below goal.

Overdelivery is the opposite. In Google Ads it describes served costs above the daily or monthly limit, which Google says are never billed. In guaranteed buys it means bonus impressions; IAB v3.0 caps bonus delivery at 10% above the IO without consent.

Lost impression share measures auctions an advertiser was eligible for but lost to budget or rank. It is one cause of search underspend, not a shortfall against a goal.

Learning Limited is Meta's status for ad sets that lack enough optimisation events. It often coincides with under-delivery but describes data scarcity rather than a volume shortfall.

Recent developments

The rules governing guaranteed shortfalls are being rewritten. In February 2026, the IAB released a draft contract framework to replace the 2010 terms. According to PPC Land's report, sellers must notify buyers within two business days if inventory becomes unavailable, deliver "in a reasonably balanced manner over the flight period," and provide a makegood, credit, refund or other remedy if material under-delivery is not corrected within 10 business days.

In search, Google began changing how campaigns with ad schedules pace from March 1, 2026. Under the change, campaigns attempt to reach the 30.4 monthly limit regardless of ad schedule. Ginny Marvin, Google's Ads Product Liaison, said a $100 daily budget scheduled Monday to Thursday would now carry a $3,040 monthly ceiling, compared with roughly $1,700 before.

Counting rules are moving too. Google will drop unrendered ads from DV360 and Campaign Manager 360 billing in February 2027, which changes what counts as a delivered impression.

Timeline

  • 2010: IAB and 4A's publish Standard Terms and Conditions v3.0, with Section VI on under-delivery notice and makegoods
  • October 2017: Google AdWords allows campaigns to spend up to twice their average daily budget per day
  • November 2017: Facebook publishes holiday guidance on ad set under-delivery
  • February 2019: DV360 renames Deal Troubleshooter to Troubleshooter and extends it to all inventory
  • June 2020: DV360 Troubleshooter adds frequency cap and cookie-loss reasons
  • September 2020: DV360 adds programmatic guaranteed support for FreeWheel, Kargo, Smartclip and Xandr
  • November 1, 2020: Amazon DSP launches PMP deal troubleshooting
  • November 8, 2022: Amazon DSP adds 15 missed-bid reasons for deal inventory
  • December 2022: Netflix lets advertisers reclaim money after delivering about 80% of expected audiences in some cases
  • September 28, 2023: Amazon DSP adds budget at risk metric
  • May 2024: Amazon DSP overview page highlights campaigns at risk of underdelivering
  • August 2024: Amazon launches Guaranteed Deals for Prime Video Ads
  • June 2025: The Trade Desk launches Deal Desk, saying 90% of deal IDs fail to scale
  • February 12, 2026: IAB opens public comment on its new contract framework
  • March 1, 2026: Google Ads begins changing budget pacing for scheduled campaigns
  • February 2027: Google to stop billing unrendered ads in DV360 and Campaign Manager 360

Summary

Who. Publishers, broadcasters and streaming services that sell guaranteed inventory owe remedies for under-delivery. Advertisers, agencies and traders on DSPs, search and social platforms absorb its costs in lost reach. Google, Meta, Amazon, Microsoft and The Trade Desk build the pacing systems and diagnostics, and the IAB and 4A's write the contract terms.

What. A shortfall between an order's goal and its actual delivery, measured in impressions, ratings, views or spend. In guaranteed buys it triggers makegoods, credits or refunds; in auction buys it means unspent budget and missed campaign objectives.

When. Television makegoods long predate digital media. The IAB/4A's terms codified digital remedies in 2010, platform diagnostics arrived between 2017 and 2024, and the IAB's replacement framework was opened for comment in February 2026.

Where. In publisher ad servers such as Google Ad Manager, in DSPs including DV360 and Amazon DSP, in Google Ads and Meta Ads Manager, in programmatic guaranteed and private marketplace deals, and in the insertion orders and upfront contracts that govern them.

Why. Every advertising order is a forecast of traffic, ratings and auction prices. When the forecast fails, under-delivery determines who carries the cost, and disputes over measurement, makegood value and platform transparency decide whether the remedy is fair.