Why people see the same ads so often—behind the ad industry’s murky secret

If you ever wondered why you see the same ads so often, here are some explanations for why. And if you’re a marketer who’s trying to cap frequency in ad campaigns, here are some of the same reasons it’s probably not working as well as you think. Frequency capping is a cornerstone of efficient media buying. But one of the murkiest secrets is that despite all the effort, it often doesn’t work. And that’s not entirely an accident, because high ad frequency is baked into the media market in ways that are often poorly understood or intentionally downplayed. NextGen Marketing Summit Join us for a crash course on Gen Z, DTC marketing and the creator economy Register today Case in point is a story from a sales rep for a measurement firm who was scolded by agency executives worried that clients or higher ups would freak out when they saw high frequency numbers being reported for campaigns. The firm ultimately kept reporting the frequency data on dashboards, the executive said, but made it harder to find, as a sort of compromise. Both TV and digital media are dogged by restrictions on identity tracking, so marketers can’t easily know when the same person sees the same ad on TV that appears on a website or a social platform. But even when identity data is solid, the setup of programmatic marketplaces means simultaneous bids for the same ad inventory can sidestep frequency caps. On the TV side, the issues include a little-understood industry convention that undercounts frequency of 15-second spots, and the growing amount of money going into an ad-supported streaming market that isn’t really as big as it seems. TV’s frequency issues Many marketers, agency buyers and even those working in analytics don’t know how the standard way of trading TV can undermine frequency management. Under the average commercial minutes (ACM) system used for most TV trading today and for many years, 30-second spots are the basic unit of measurement, and 15-second spots count as half impressions sold at half the price, said Kelly Abcarian, a former Nielsen and NBCUniversal executive. Media measurement blog Tracking TV, social and digital updates Click here “For example, say the ACM value for a 30-second ad is based on 1 million impressions,” Abcarian said. “To achieve the same value with 15-second ads, publishers would need to deliver 2 million impressions, as each 15-second ad is credited with only half the impressions and value of a 30-second spot. This approach often leads to over-frequency, which can result in real brand damage. The solution is to adopt metrics that measure actual delivery and effectiveness, rather than discounting impressions based on ad length, which misrepresents the true impact of shorter ads.” While a 15-second ad counts for only half of a 30-second one under the predominant TV currency used today, it’s probably worth more than that. Research by YouTube in 2018 found ad recall rates increase steadily for up to 10 seconds of continuous video viewership but not after that, which suggests advertisers can get all the impact from a 15-second commercial that they get from a 30-second one. A 2021 report by Ebiquity, using Lumen and TVision data, found an average 15-second spot gets 7.5 seconds of “eyes on” attention, or 54% of the 13.8 seconds that 30-second spots get. If ad units were based on their true value :15s would likely be priced well over half the cost of :30s, said Dave Morgan, CEO of Simulmedia, the TV ad targeting platform. Regardless, under the current pricing model, :15s are definitely winning market share. People familiar with the industry say 15-second spots have grown steadily to become the predominant format over recent years. Yet, for frequency counting purposes, those :15s often get reported as half impressions. Even most advertising researchers don’t appear to understand how this has led to underreporting frequency, said Leslie Wood, former chief research officer of iSpot and NCSolutions, after having raised the issue earlier this year with a group of fellow researchers. “When you buy :15s, you double the impressions, so your frequency is twice as high,” Wood said. “And with more and more campaigns being predominantly :15s, there’s a lot more frequency than people realize. It’s a flaw in the logic.” Streaming presents new issues While almost all linear TV trades on average commercial minutes and the ad equivalization system where two :15s equal a :30, probably only around a third of connected TV trades that way, Simulmedia’s Morgan said. But that doesn’t mean the shift to streaming solves frequency problems. In fact, it exacerbates them, he said. Shifting TV budgets to CTV is seemingly a no-brainer. After all, Nielsen’s The Gauge shows streaming now accounts for more than 40% of TV viewing. But while viewers are steadily shifting from linear to streaming, available ad impressions aren’t moving nearly as fast. Streamers’ share of all ad-supported TV inventory is only in the 11% to 13% range (including digital behemoths YouTube, Amazon and Netflix), according to analyses by Simulmedia and media analyst Evan Shapiro for Comscore. (Simulmedia pegs it at 11%, Comscore at a slightly more generous 13%). Also, the viewers marketers are following into streaming are usually still watching linear TV, too, so they don’t represent unique reach. Of the 63% of U.S. households that subscribe to at least one ad-supported streamer, fewer than half were only watching TV via streaming, according to the most recent Spring 2024 data from the Device Usage and Account Sharing (DASH) study of the Advertising Research Foundation. Only 23% of U.S. households could be reached through at least one ad-supported streaming platform but weren’t also watching linear. More: The sun won’t set on Nielsen ratings Streaming networks also generally have less commercial time per hour than linear TV, which means less inventory to reach those streaming-only viewers. So available linear ad impressions outnumber CTV ad impressions by nearly an 8-to-1 margin, according to Simulmedia. Most of those CTV ads go to the heaviest viewers, a problem that’s worse in streaming than linear. The 20% of heaviest CTV viewers see 61% of ad impressions, according to Simulmedia’s analysis, which draws on Nielsen data plus direct viewing data feeds from set-top boxes and smart TVs. That’s a bigger proportion than the half of ads seen by the heaviest 20% of linear viewers.. Mashing up the data from DASH and Simulmedia— with 63% of households seeing streaming programming and 61% of CTV ads going to the heaviest 20% of viewers—most CTV ads may be seen by under 13% of U.S. households. So even at some of the highest CPMs in the industry, CTV can easily inundate a relative handful of viewers with ads, Morgan said. Read more: Cable TV’s breaking point—inside the numbers The challenge of identity resolution It’s particularly hard to avoid that when there’s no way to reliably track viewers across networks or platforms. Using retail media data to target ads, where identities can be matched between the retailer and the network and possibly an advertiser’s first-party data, can help. But it’s not failsafe, particularly if it relies on third-party data for the match. A 2023 study by data-vetting firm TruthSet found that nearly half the data used for ad targeting, specifically to match email with postal addresses, was wrong—though the firm has recently launched a product allowing data brokers to eliminate the most questionable segments of their databases.One publisher executive recounts running an experiment by using AI to generate thousands of fake email addresses, then asking data brokers how many of those addresses they could match in their databases. They came back with match rates of 30% to 40%, the executive said, essentially failing the test by serving up large numbers of bad matches, though one smaller broker won the accuracy award by admitting to not being able to find many matches. Even the biggest efforts to bridge identities across media sellers and buyers, such as The Trade Desk’s Unified ID 2.0, only have limited reach, said John Piccone, president of the Americas for ad tech company Adform, which has built an ID fusion system to work across many of the systems in the marketplace. “There’s a misconception in the marketplace that companies who want to match their first-party data with companies like LiveRamp assume their match rate is the same as their activation rate,” Piccone said. It isn’t, because various demand-side platforms and supply side platforms don’t operate on common ID frameworks, he said. One computer hardware marketer who spoke on the condition of anonymity, said the company switched to Adform to take advantage of its ID resolution and likes the results so far. But the challenge for this advertiser, who wasn’t authorized by the company to speak to the media, isn’t capping to prevent excessive frequency but finding target customers enough to get frequency levels to double digits, where the company has seen the best results. TV and streaming ad prices in 2024-25 Primetime TV ad prices Cost of ads on Netflix, Disney+ and Hulu TV ad cost trends over the past decade More broadly, identity resolution is an issue the industry increasingly is trying to solve because of its central role in preventing excess frequency and ensuring money gets spent on reaching more people rather than the same ones over and over. However, these efforts are likely years away from having an impact. The Coalition for Innovative Media Measurement and IAB recently issued a roadmap around identity resolution, but it’s not clear yet how or when it can be implemented or who will be in charge, since the groups said those are among areas to be determined. The Association of National Advertisers launched Aquila earlier this year as an offshoot to develop a privacy-safe way of measuring frequency and reach across platforms. But just getting Aquila off the ground and commissioning Kantar to start forming a calibration panel to implement it took three years, and a pilot isn’t set to start until early next year. Aquila hasn’t publicly outlined exactly how its virtual ID system is being constructed or will work. Participating media companies so far include Google, Meta, Amazon and TikTok, and Aquila is preparing to name a third-party measurement company to help extend its work to TV and CTV. However, networks have not yet signed on to share their first-party streaming data. While the platforms are picking up 60% of the tab for operating Aquila, the project is still recruiting marketers to join, and those who do will pay a toll based on a percentage of their media spending. Frequency problems baked into digital Within the owned properties of big social platforms such as Meta, Amazon and TikTok, it’s much easier to manage frequency, although that’s often based on reporting from the platforms themselves rather than third-party verification, said Krzysztof Franaszek, CEO of measurement firm Adalytics. Adalytics, best known for a series of muckraking reports about shortcomings in digital media trading, also has a day job with paying clients, and much of it is focused on helping them manage frequency. It isn’t easy, particularly on the open web. “In some cases, third-party ad servers allow media buyers to approximate frequency of exposure per device ID, but in many cases measuring ad delivery independently via an independent ad server is not an option,” Franaszek said. “We have observed many instances where brands had their ads served hundreds or thousands of times to the same consumer. For example, earlier this year, a major brand had 2,117 impressions served to one consumer on a single MFA article in less than one hour.” Sometimes, excess frequency just happens when a brand or its agency simply forgets to configure the frequency cap, he said. “Empirical data suggests that even when a media buyer configures a frequency cap in a media platform, those frequency caps are not universally respected,” he said. “This can be due to various technical and software reasons that make it difficult for an ad buying platform to correctly manage frequency, particularly in an environment with quick ad refreshes and significant programmatic bid duplication.” The latter happens because ad auctions occur in milliseconds to decide what ads get served to what consumers, and in some cases dozens of SSPs convey bid requests to DSPs simultaneously for the same ad slot. “It can be difficult for a media buying platform to de-duplicate bid requests arriving at the same time and/or to synchronize the media buying platform’s user and ad frequency databases,” Franaszek said. “This problem can be compounded in situations where a very large number of intermediaries are representing the same ad inventory and consumer to media buying platforms at the same time. Selective participation in ad auctions can help reduce the scale of this problem for media buying platforms.” What’s ideal? Provided it’s possible to execute, there isn’t a one-size-fits-all standard for what frequency is best. But there are some recommendations. A good rule of thumb for frequency is one impression per week, said Wood, based on research from her NCS days that focused on packaged goods brands. Not all frequency is created equal, according to Wood. Bunching up impressions close together has less sales impact than spreading them out equally over the course of a campaign. Spreading out impressions means exposure is more likely to come close to the time of a purchase decision, she said. Wood’s NCS research was based largely on well-established packaged goods brands; Other brands in other categories or new brands seeking to establish awareness for the first time might have different targets. The Trade Desk, which gets paid based on the volume of ads it places, has commissioned third-party research from Lucid recommending far more aggressive frequency targets for various media, based on evaluating e-commerce campaigns and incremental impact on such things as awareness, consideration and purchase intent. TTD’s recommendations, broken down by medium, call for six CTV, 10 online video, eight online audio, 20-plus online display or 12 native advertising exposures weekly. Procter & Gamble Co. worked on a standard of trying to reach people on digital no more than three times monthly in 2018, and more recently in a 2023 speech Chief Brand Officer Marc Pritchard noted an overall cross-platform frequency of “a couple of times a week” as sufficient to maintain brand awareness, as the company works to expand the reach of its brands to more people. But that discipline doesn’t always work, according to Adalytics, which among other things earlier this year showed a screen grab with the same Pampers ad playing simultaneously on small outstream video players in two corners of the same web page. Even for the biggest of advertisers with the best of intentions, managing frequency can be pretty tough.

Why people see the same ads so often—behind the ad industry’s murky secret

Featured Stories