The frequency illusion is a quirk of perception in which a thing a person has only just noticed suddenly seems to turn up everywhere. A new word, a car model, a brand name: once it registers, every further encounter is spotted and remembered, while the many earlier encounters that passed unnoticed are not counted. The result is a confident but inflated sense of how often the thing occurs. It is also widely known as the Baader-Meinhof phenomenon. In advertising it matters because it sits on top of a real delivery system: ads are repeated on purpose, retargeting follows people who showed interest, and a person cannot easily tell what a platform did from what they merely noticed.

How the illusion works

Two mechanisms are usually cited. The first is selective attention. Once a stimulus becomes relevant, the brain filters for it, so instances that would previously have gone unregistered now break through. The second is confirmation bias: each new sighting is taken as evidence that the thing is common, while the absence of sightings in the past is never tallied.

Arnold Zwicky, the Stanford linguist who named the effect, put it plainly in an August 7, 2005 post on the Language Log blog: "once you've noticed a phenomenon, you think it happens a whole lot, even 'all the time'." He also described the recency illusion, "the belief that things you have noticed only recently are in fact recent." The two often travel together.

A worked example shows the arithmetic. A commuter passes 40 cars of one model each week without registering any. After a test drive, they notice perhaps 10 a week. The true rate has not changed, but the perceived rate has gone from zero to 10, and with no memory of the unnoticed 40, the jump feels like a surge.

Research on frequency estimation supports the pattern. A 1986 study by Ian Begg and colleagues in the Journal of Experimental Psychology found that estimates of how often an item appeared tracked the variety of contexts in which it was seen, and that vivid or emotional items were overestimated. Christos Georganas, Paul Healy and Nan Li reported in the European Economic Review in 2014 that consumers' perceptions of inflation were skewed towards goods they bought frequently.

Origin of the two names

The folk name came first. On October 16, 1994, the Sunday Bulletin Board column of the St. Paul Pioneer Press in Minnesota printed a reader's letter describing how, after the Baader-Meinhof group came up in conversation, the West German militant organisation also known as the Red Army Faction, the writer kept encountering the name. The letter, signed with a pseudonym and later attributed to a reader named Terry Mullen, dubbed the experience "the Baader-Meinhof phenomenon". Other readers wrote in with similar stories, and the label stuck.

The scholarly name followed in 2005, when Zwicky coined "frequency illusion" on Language Log. He developed it in a 2006 abstract for the Linguistic Society of America, "Why are we so illuded?", on how linguists misjudge how common a usage is. The Oxford English Dictionary added "frequency illusion" in its March 2019 update, defining it as "a quirk of perception whereby a phenomenon to which one is newly alert suddenly seems ubiquitous", according to a March 19, 2019 Language Log post by Ben Zimmer. The OED cites Zwicky's 2005 post as the first use and records "Baader-Meinhof phenomenon" as an alternative name.

Where the illusion meets ad delivery

Elsewhere the illusion is a pure error: only attention has changed. In digital advertising, noticing a product and being shown more of it often happen together.

Consider the sequence. A person visits a retailer's product page. A tag such as the Meta Pixel records the visit, and the advertiser adds that person to an audience built for retargeting. Over the following days, that person genuinely does see more ads for the product than before. Having just researched it, they are also primed to notice every mention of it. Real frequency rises, and perceived frequency rises faster.

How much real repetition is permitted depends on settings the buyer controls. Frequency capping limits how many times one identifier can see the same ad in a window, such as three impressions per seven days. The headline number on a media plan, covered in the reach and frequency explainer, is an average: total impressions divided by unduplicated reach. An average frequency of 2.5 can hide a tail of people who saw an ad ten or more times, and those people are the most likely to feel followed.

Ads can also reflect the interests of friends or modelled lookalike audiences. Adam Mosseri, head of Instagram, cited these in a video reported on October 3, 2025, plus two explanations that are the frequency illusion in all but name: people see an ad, forget it, discuss the product and notice the ad only afterwards, and sometimes the timing is chance.

Why it matters for marketers

The most visible consequence is the enduring belief that phones listen to conversations. In a survey of 1,000 US adults run via Pollfish in September 2026, 70% said they believed their phone listens to conversations to influence ads, and 81% said they had seen an ad for something they had only discussed, according to Vibe, the CTV ad platform that commissioned it. Vibe sells targeted advertising.

The technical evidence points the other way. Researchers at Northeastern University, led by David Choffnes and Christo Wilson, tested more than 17,000 popular Android apps and found none that activated the microphone to send audio, according to a July 6, 2018 university announcement. They did find that about 9,000 apps could take screenshots of themselves. Facebook said in June 2016 that it "does not use your phone's microphone to inform ads or to change what you see in News Feed." Tracking, data matching and the frequency illusion explain most "uncanny" ads without eavesdropping.

That belief has commercial value to those willing to exploit it. Cox Media Group sold an "Active Listening" product to small businesses from 2023, claiming AI could detect purchase intent from conversations picked up by smartphones and smart speakers. The FTC found no such capability existed, and that buyers received email lists bought from data brokers. Cox Media Group agreed to pay $880,000, and the orders impose 20 years of FTC oversight.

The illusion also bears on frequency planning. Herbert Krugman's 1972 paper "Why Three Exposures May Be Enough" set a planning default that still shapes caps today. Two Google researchers, Pablo Perez and Jesus Martin Calvo, argued in February 2026 that a second exposure on YouTube delivered brand lift indexing at 1.5 to 1.8 against a first, below the 2.0 that would make it equally valuable. If people perceive repetition as heavier than it is, the cost of excess frequency includes irritation as well as wasted impressions. Excessive frequency was the top ad irritant for 35.2% of respondents in TiVo's Q4 2025 Video Trends Report.

Attention metrics measure whether a single impression was looked at, and cumulative attention, covered in its own explainer, sums attention across exposures. The frequency illusion is the perceptual flip side: a brand that has become salient to someone can feel omnipresent even when delivery is modest.

Limitations and disputes

Much of what is written about the Baader-Meinhof phenomenon in marketing is folk psychology. There is no large body of controlled research testing the named effect directly, and most of the supporting evidence comes from adjacent work on frequency estimation, attention and memory. Claims that the illusion can be "harnessed" to make a brand seem popular are rarely backed by published experiments.

The term can also be misused as a dismissal. Calling an uncanny ad "just Baader-Meinhof" ignores that it may have come from real tracking, such as a pixel firing or a list match, and a consumer cannot tell the two apart. The line is shifting, too. Meta began using conversations with its AI features, including voice chats, to personalise ads from December 16, 2025, so some ads now do follow from things people said, though to an AI assistant and with microphone permission rather than by covert listening. The eavesdropping belief is wrong on mechanics, but the sense of being watched is not baseless, as debates over ad personalisation show.

Not the same as

  • Mere exposure effect. Robert Zajonc showed in 1968 that repeated exposure to a stimulus tends to increase liking for it. That concerns liking, not counting.
  • Illusory truth effect. Lynn Hasher and colleagues reported in 1977 that repeated statements are rated as more likely to be true. It concerns belief, not counts.
  • Availability heuristic. Amos Tversky and Daniel Kahneman described in 1973 how people judge frequency by how easily examples come to mind. The frequency illusion is often treated as one instance of it.
  • Ad frequency. In media planning, frequency is a measured number of exposures per person. The illusion is a perception that may diverge sharply from that number.

Recent developments

The FTC announced the Cox Media Group, MindSift and 1010 Digital Works settlements on May 21, 2026, totalling $930,000, and finalised the orders in late August 2026. Christopher Mufarrige, director of the FTC's Bureau of Consumer Protection, said: "It is a basic rule of business that you need to be honest with your customers." Meanwhile, buyers have more tools to manage real repetition, including Amazon's Frequency Groups and cross-order reporting and the cost-based metrics Display & Video 360 introduced in August 2025. None of them reach the part of frequency that happens in a viewer's head.

Timeline

  • 1968: Robert Zajonc publishes research on the mere exposure effect.
  • 1972: Herbert Krugman publishes "Why Three Exposures May Be Enough" in the Journal of Advertising Research.
  • 1973: Amos Tversky and Daniel Kahneman describe the availability heuristic.
  • 1986: Ian Begg and colleagues link frequency estimates to contextual variety and vividness.
  • October 16, 1994: A reader letter in the St. Paul Pioneer Press coins "the Baader-Meinhof phenomenon".
  • August 7, 2005: Arnold Zwicky names the "frequency illusion" and "recency illusion" on Language Log.
  • 2006: Zwicky's abstract "Why are we so illuded?" for the Linguistic Society of America.
  • 2014: Georganas, Healy and Li report frequency bias in consumers' inflation perceptions.
  • June 2016: Facebook states it does not use phone microphones to inform ads.
  • July 6, 2018: Northeastern University reports no microphone activation across more than 17,000 Android apps.
  • March 2019: The Oxford English Dictionary adds "frequency illusion".
  • 2023: Cox Media Group begins selling "Active Listening" to small businesses.
  • Mid-2024: The Active Listening service is discontinued or wound down.
  • October 1, 2025: Meta announces it will use AI chat interactions for ad personalisation.
  • October 3, 2025: Adam Mosseri's denial that Instagram listens through microphones is reported.
  • December 16, 2025: Meta's AI-chat ad signals take effect.
  • February 3, 2026: Google researchers publish analysis questioning frequency orthodoxy.
  • May 21, 2026: FTC announces $930,000 in settlements over Active Listening.
  • August 27, 2026: FTC announces finalised Active Listening orders.
  • September 2026: Vibe survey finds 70% of US adults believe phones listen for ads.

Summary

Who: The term "frequency illusion" was coined by Stanford linguist Arnold Zwicky; the alternative name "Baader-Meinhof phenomenon" came from a 1994 newspaper reader. It affects consumers, and it matters to advertisers, platforms such as Meta and Google, and regulators such as the FTC.

What: A cognitive bias in which something newly noticed seems to appear far more often than it does, driven by selective attention and confirmation bias.

When: Named in 1994 and 2005, added to the Oxford English Dictionary in March 2019, and central to 2025 and 2026 disputes over whether phones listen for ads.

Where: In everyday perception generally, and in digital advertising specifically, where retargeting and repeated delivery produce real increases in exposure that the illusion amplifies.

Why: It explains much of the "my phone is listening" belief, shapes how repetition is experienced, and was exploited by a vendor that sold fake listening technology, making it relevant to frequency planning, privacy perception and trust.