Frequency is the number of times a person is exposed to an ad or campaign within a given period. It is usually reported as average frequency: total impressions divided by reach, the number of unique people exposed. Frequency and reach together describe how a campaign's impressions are spread across its audience.
Formula and example
Average frequency = Impressions ÷ Reach, which is the same as Impressions = Reach × Average frequency.
In traditional media planning, the same relationship is expressed in rating points: Gross rating points (GRP) = Reach % × Average frequency, where reach % is reach as a share of the target population.
Example: a campaign delivers 1 200 000 impressions to 400 000 unique people in a target population of 2 000 000. Average frequency is 1 200 000 ÷ 400 000 = 3.0. Reach % is 20%, so the campaign delivered 20 × 3.0 = 60 GRPs.
Frequency works alongside reach: for a fixed number of impressions, more reach means lower frequency, and vice versa.
Ways to describe frequency
Average frequency: impressions ÷ reach. Easy to report, but it hides how exposure is distributed.
Frequency distribution: how many people were exposed once, twice, three times and so on. An average of 3.0 can mean most people saw the ad three times, or that a small group saw it fifteen times and most people saw it once. Google Ads, for example, reports average impression frequency per user over rolling 7- and 30-day windows (Google Ads Help).
Effective frequency: the minimum number of exposures assumed to be needed for an ad to work. People reached at or above that level make up effective reach.
Frequency cap: a limit set in the buying platform on how many times one user can be shown an ad in a period.
Where "three exposures" comes from
The idea that about three exposures are needed comes from Herbert Krugman's 1972 article "Why three exposures may be enough" (Krugman, 1972). In 1979 the Association of National Advertisers published a review of frequency research edited by Michael Naples, which became the standard reference for effective-frequency planning and supported a 3+ threshold (Naples, 1979).
The rule has been debated since; Krugman himself later revisited the data behind it (Krugman, 2000). Note also that a planned media exposure (an opportunity to see) is not the same as an exposure the person actually noticed. Tellis argued that neither "one exposure is enough" nor "constant repetition" holds as a universal rule, and that the effective level depends on factors such as brand familiarity, message complexity and the novelty of the ad (Tellis, 1997). A practical reading: use 3+ as a starting assumption, then test and measure it for your brand and channel.
What drives the right frequency
- Brand familiarity: well-known brands typically need fewer exposures to remind than new brands need to inform.
- Message complexity and creative novelty: complex messages may need more repetition; a new creative can reset wear-out.
- Purchase cycle: for frequently bought products, the timing of an exposure relative to purchase matters as much as the count.
- Channel and format: exposures differ in quality between, for example, a full-screen video and a feed impression scrolled past.
Measuring frequency across channels
For TV, radio and print, frequency is estimated from audience panels and expressed through GRPs. Digital platforms report frequency within their own ecosystem, usually based on modeled unique users (Google Ads Help, Measuring reach and frequency). Because each platform deduplicates only its own users, nobody sees the true cross-platform frequency: a person can be at frequency 2 on each of four platforms and at 8 overall.
Frequency and saturation in marketing mix modeling
High frequency is one of the main sources of diminishing returns. As spend on a channel grows, a larger share of impressions goes to people who have already seen the ad several times, and each extra exposure adds less. In marketing mix modeling (MMM), this appears as a saturating response curve, usually modeled with a Hill or similar function on spend or impressions (Jin et al., 2017).
When reach and frequency data are available for a channel, frequency can be modeled directly. Google researchers proposed a geo-level Bayesian hierarchical MMM that takes reach and frequency as inputs and can recommend an optimal frequency (Wurm, Liu et al., 2023). Google's Meridian applies a Hill function to average frequency, so the response to frequency can be S-shaped, and finds the optimal frequency numerically as the level that maximizes expected ROI (Google Meridian documentation). This turns the effective-frequency debate into an estimate from your own data rather than a rule of thumb.
Frequency targets should ultimately be judged by incremental sales, not exposure counts. MMM estimates this from historical variation; experiments that vary frequency or spend between regions or audience groups test it directly.
Key takeaways
- Average frequency = impressions ÷ reach; GRP = reach % × average frequency.
- The 3+ effective frequency rule comes from Krugman (1972) and Naples (1979) and is a planning convention, not a universal law.
- Rising frequency is a main driver of saturation; MMM with reach and frequency inputs can estimate the frequency that maximizes incremental return.
Related terms
Further reading on Sellforte
- Advertising response curves: what are they and why do you need them?
- Why diminishing return curves are crucial insights in marketing planning
- How to measure offline media with marketing mix modeling
References
- Krugman, H. E. (1972). Why three exposures may be enough. Journal of Advertising Research, 12(6), 11–14. https://cir.nii.ac.jp/crid/1574231874476450560
- Krugman, H. E. (2000). Memory without recall, exposure without perception. Journal of Advertising Research, 40(6), 49–54. https://ideas.repec.org/a/cup/jadres/v40y2000i06p49-54_00.html
- Naples, M. J. (1979). Effective frequency: The relationship between frequency and advertising effectiveness. Association of National Advertisers. https://worldradiohistory.com/Bookshelf/Business/Effective-Frequency-ANA-Naples-1979.pdf
- Tellis, G. J. (1997). Effective frequency: One exposure or three factors? Journal of Advertising Research. https://www.warc.com/content/article/jar/effective-frequency-one-exposure-or-three-factors/en-gb/9066
- Jin, Y., Wang, Y., Sun, Y., Chan, D., & Koehler, J. (2017). Bayesian methods for media mix modeling with carryover and shape effects. Google Research. https://research.google/pubs/bayesian-methods-for-media-mix-modeling-with-carryover-and-shape-effects/
- Wurm, M., Liu, Y., et al. (2023). Bayesian hierarchical media mix model incorporating reach and frequency data. Google Research. https://research.google/pubs/bayesian-hierarchical-media-mix-model-incorporating-reach-and-frequency-data/
- Google. (n.d.). Optimize frequency. Meridian documentation. Retrieved 2026-10-30, from https://developers.google.com/meridian/docs/user-guide/optimize-frequency
- Google. (n.d.). Measuring reach and frequency. Google Ads Help. Retrieved 2026-10-30, from https://support.google.com/google-ads/answer/2472714
- Google. (n.d.). About avg. impr. freq. per user (7 or 30 days). Google Ads Help. Retrieved 2026-10-30, from https://support.google.com/google-ads/answer/9507337
Authors
Lauri Potka is the Chief Operating Officer at Sellforte, with over 15 years of experience in Marketing Mix Modeling, marketing measurement, and media spend optimization. Before joining Sellforte, he worked as a management consultant at the Boston Consulting Group, advising some of the world’s largest advertisers on data-driven marketing optimization. Follow Lauri in LinkedIn, where he is one of the leading voices in MMM and marketing measurement.
