Reach and frequency are the two numbers a media plan uses to describe an advertising schedule in terms of people rather than impressions. Reach counts the unduplicated audience exposed at least once during a defined period. Frequency counts how many times, on average, each of those people was exposed. The pair exists because impressions alone cannot distinguish a campaign that touched a million people once from one that touched a hundred thousand people ten times, even though both deliver the same volume and cost the same money.

The arithmetic linking them is fixed. Total impressions divided by reach gives average frequency. Reach expressed as a percentage of a target population, multiplied by average frequency, gives gross rating points, or GRPs, the currency television has used for decades. The Media Rating Council states the reach calculation as unduplicated audience divided by universe or target, multiplied by 100, and notes that reach can equally be reported as a whole number of unique users. Any two of the three figures determine the third, which is why planners describe the relationship as a trade-off: for a fixed budget, buying more frequency means buying less reach.

What the numbers actually count

The hard part is the denominator. Reach requires deciding that two impressions belong to the same person, and no advertising system observes people directly. It observes cookies, mobile advertising identifiers, connected television identifiers, hashed emails and logged-in accounts, then resolves those into an estimate of individuals.

Google Ads exposes the pair through the API as metrics.unique_users and metrics.average_impression_frequency_per_user. Both carry constraints that reveal how they are produced: neither can be aggregated across queries, and both are restricted to date ranges of 92 days or less, according to Google Ads API support responses. The second metric is a plain division of impressions by the first.

Average frequency is the least informative part of the pair, because it collapses a distribution into a single value. A campaign averaging three exposures might have given nearly everyone three, or given most people one and a small group twenty. Platforms therefore publish frequency distributions alongside the average. Google's reach forecasting service returns an EffectiveFrequencyBreakdown object whose effective_frequency field takes integer values from 1 to 10, each row reporting how many unique people were reached at least that many times. Display and Video 360 reports the same shape in buckets, and on July 7, 2025 added 8+ and 9+ tiers to fill the gap between its existing 1+ through 7+ rows and its 10+ row. Amazon DSP reports frequency bins running from a single view to ten or more per unique user.

Quality thresholds sit underneath. The MRC's Digital Audience-Based Measurement Standards specify that while served impressions may be counted in aggregate, viewable impressions are the minimum qualifying unit for digital and cross-media reach, frequency and GRP. An impression that never entered the viewport should not create reach.

Buying against the pair

Two purchasing models attach to these metrics. The first is reservation: a fixed price per thousand impressions agreed in advance against a stated reach and frequency plan. Meta operated exactly such a buying type under the name Reach and Frequency until October 2023, when it was renamed Reservation. According to Jon Loomer, who documented the rollout, the renamed type carries a minimum reach of 200,000, is limited to awareness and engagement objectives, supports reach, ad recall lift and ThruPlay performance goals, and makes target frequency rather than a frequency cap the default control.

The second model steers an auction toward a frequency target. Google's Fixed CPM bidding strategy accepts a FixedCpmTargetFrequencyGoalInfo object containing target_count, the number of exposures wanted per user, and time_unit, the window over which to hit it. Display and Video 360 introduced monthly target frequency for YouTube on December 2, 2024, optimising delivery toward a chosen level rather than halting at a ceiling. The distinction matters: a cap refuses impressions above a threshold, while a target actively buys toward one.

Forecasting sits ahead of both. Google's ReachPlanService, restricted to allowlisted partners under a data licensing agreement, returns on_target_reach and total_reach against a requested min_effective_frequency, along with viewable impressions and co-view figures. Reach Planner in the Google Ads interface exposes much of the same. Google has expanded the television data feeding those forecasts steadily, reaching 18 countries by July 2021 and adding Vietnam, France, Germany, Sweden and Spain in March 2022.

Origin and evolution

The pair predates digital advertising by decades. Herbert Krugman, then at General Electric, published "Why Three Exposures May Be Enough" in the Journal of Advertising Research in 1972. Michael Naples assembled the surrounding evidence in Effective Frequency for the Association of National Advertisers in 1979, and three exposures per purchase cycle hardened into the planning default. Erwin Ephron inverted the logic in the same journal in 1997, arguing that because the moment of purchase readiness cannot be predicted, continuous broad reach beats accumulated exposures. The dispute was never settled, and both positions remain visible in platform defaults.

Digital measurement had to rebuild the concepts from a different substrate. The MRC published its Digital Audience-Based Measurement Standards in December 2017 and Cross-Media Audience Measurement Standards, Phase I Video, in September 2019, the latter superseding parts of the former. Platform tooling followed: Amazon launched Frequency Groups in October 2023 to apply one cap across separate orders, added deduplicated cross-order reporting on December 23, 2024, and in October 2024 released an Optimal Frequency solution in Amazon Marketing Cloud plotting conversion rate across frequency bins. Google Ads introduced Cross-Media Reach in June 2024 for deduplicated figures across YouTube and, in some markets, linear television. LinkedIn added frequency capping on July 2, 2025, allowing limits of three to thirty impressions per member account across seven days.

Why it matters for marketers

Budget allocation depends on the split. Amazon internal data from September 2023, self-reported in its own announcement, put the average unique reach gain from Frequency Groups at 6 per cent, and a consumer packaged goods case in the same documentation found 34 per cent reach overlap across orders from three brands in one business line. Overlap that large is money spent on repetition nobody planned.

Pricing has followed. Display and Video 360 replaced a bid-request frequency metric with cost-based alternatives on August 26, 2025, introducing Cost per reach, calculated as total media cost divided by impression reach, and Savings Reinvested from Frequency Cap. That converts the pair from a planning abstraction into a line-item efficiency measure.

Limitations and disputes

The numbers move when definitions move, without any change in delivery. Google applied a Total Co-view definition to seven Google Ads reach and frequency metrics on June 2, 2026, five of which are frequency distribution thresholds. Google described it as a data definition change requiring no code migration. Reported unique users rose; the same impressions were served. Co-viewing is now counted three ways by three organisations, and Video Advertising Bureau research put the share at 60 per cent on premium connected television against 45 per cent on YouTube.

Deduplication is the second weakness, and it is structural rather than a vendor failing. Counters held by separate buying platforms cannot see each other, which is why Amazon extended reporting across advertiser accounts on April 2, 2026and why Display and Video 360 added Unique Reach Overlap dimensions in April 2026. Addressable reach, the share of a target the stack can identify at all, bounds what any of this can measure.

Whether frequency deserves its status is contested inside Google. Pablo Perez and Jesús Martín Calvo of the company's Marketing Research and Insights team argued on February 3, 2026 that first impacts carry the most value and that reach should be maximised over repetition, citing YouTube Brand Lift data published in 2017 in which second impacts indexed at 1.5 to 1.8 rather than the 2 that equal value would imply. They also treat S-shaped response curves in mix models as artefacts of weak measurement at low spend. The counter-position, that frequency correlates with outcomes in aggregate models, is accepted as an observation and rejected as an inference.

Not the same as

Frequency capping is a delivery rule, not a metric: it enforces a ceiling, while frequency reports what happened. Impressions count ad deliveries with no person attached. GRPs bundle reach and frequency into one number and cannot be unpacked without knowing one of the components. Effective frequency names a threshold judged sufficient for response, an editorial choice rather than a measurement, and appears in Google's forecasting as a request parameter rather than an output.

Recent developments

Cross-media currency work has accelerated. Nielsen launched four-screen deduplicated measurement in Italy on June 8, 2026, then removed the single-publisher restriction in Japan on August 3, 2026, with reporting covering unique audience, audience reach, average frequency and gross rating points. Nielsen and Adelaide combined attention scores with reach data on October 7, 2025. YouTube Studio began estimating people rather than devices in May 2026.

Timeline

  • 1972: Herbert Krugman publishes "Why Three Exposures May Be Enough" in the Journal of Advertising Research
  • 1979: Michael Naples publishes Effective Frequency for the Association of National Advertisers
  • 1997: Erwin Ephron publishes his recency planning argument, prioritising reach over accumulated exposures
  • August 2013: IAB updates its Audience Reach Measurement Guidelines to version 2.0
  • December 2017: MRC publishes Digital Audience-Based Measurement Standards version 1.0
  • September 2019: MRC publishes Cross-Media Audience Measurement Standards, Phase I Video
  • July 2021: Google expands TV Reach Planner to 18 countries
  • March 2022: TV data in Reach Planner reaches Vietnam, France, Germany, Sweden and Spain
  • October 2023: Amazon launches Frequency Groups; Meta renames its Reach and Frequency buying type Reservation
  • June 2024: Google Ads introduces Cross-Media Reach measurement
  • October 2024: Amazon Marketing Cloud releases Optimal Frequency
  • December 2, 2024: Display and Video 360 announces monthly target frequency for YouTube
  • December 23, 2024: Amazon adds deduplicated cross-order reach and frequency reporting
  • July 2, 2025: LinkedIn launches frequency capping for brand awareness campaigns
  • July 7, 2025: Display and Video 360 adds 8+ and 9+ reach buckets
  • August 26, 2025: Display and Video 360 introduces Cost per reach and frequency savings metrics
  • October 7, 2025: Nielsen and Adelaide integrate attention metrics with reach data
  • February 3, 2026: Google researchers publish an analysis challenging frequency orthodoxy
  • April 2, 2026: Amazon launches cross-account reach and frequency reporting
  • April 2026: Display and Video 360 adds Unique Reach Overlap dimensions
  • June 2, 2026: Google Ads applies a Total Co-view definition to seven reach and frequency metrics
  • June 8, 2026: Nielsen launches four-screen deduplicated measurement in Italy
  • August 3, 2026: Nielsen extends four-screen measurement in Japan beyond YouTube

Summary

Who: Media planners and buyers on the advertiser and agency side; platforms including Google Ads, Display and Video 360, Amazon DSP, Meta and LinkedIn that report and sell against the pair; measurement bodies including the Media Rating Council, the IAB and Nielsen that define and audit it.

What: Two linked metrics describing an advertising schedule in terms of people. Reach is the unduplicated audience exposed at least once in a period; frequency is the average number of exposures among that audience. Impressions divided by reach yields frequency, and reach percentage multiplied by frequency yields gross rating points.

When: The concepts were formalised in television planning research between 1972 and 1979 and contested from 1997. Digital standards arrived with the MRC in December 2017 and September 2019. Platform tooling for deduplication and frequency targeting has shipped continuously since October 2023, with cross-account and cross-media capability landing through 2026.

Where: Across television, connected television, video, display, social and retail media, inside buying platforms, reserved deals and third-party cross-media measurement products.

Why: Impression counts cannot distinguish broad exposure from repeated exposure, and the two produce different outcomes at the same cost. The pair separates them. Both figures depend on resolving identifiers into people, so they shift when identity coverage changes or when a platform revises a definition, and the question of how much frequency is worth buying remains unresolved after fifty years of research.