Impressions are a vanity unit. A million impressions can mean a million people saw your ad once, or a hundred thousand people saw it ten times. Same volume, same invoice, wildly different outcome.
That is why media plans still lean on the oldest pair of numbers in the business: reach and frequency. PPC Land has published a deep explainer on how the two work across today’s buying platforms, and it is worth every performance marketer’s attention.
The math is fixed, the denominator is not
Reach is the unduplicated audience exposed at least once in a defined period. Frequency is the average number of exposures among that audience. Divide total impressions by reach and you get average frequency. Multiply reach as a percentage of your target by average frequency and you get GRPs, television’s long-running currency.
Any two figures give you the third. Which is exactly why planners talk about a trade-off: at a fixed budget, more frequency means less reach.
The hard part is deciding that two impressions belong to the same person. No ad system sees people. It sees cookies, mobile ad IDs, connected TV identifiers, hashed emails and logged-in accounts, then estimates individuals from that mess. Every reach number you have ever read is an estimate built on identity resolution.
Average frequency hides the truth
Average frequency is the weakest half of the pair because it flattens a distribution into one number. A campaign averaging three exposures might have delivered three to nearly everyone, or one to most people and twenty to a small cluster. Those are different campaigns.
That is why platforms publish distributions. Google’s reach forecasting returns effective frequency rows from 1 to 10, each showing how many unique people were reached at least that many times. Display & Video 360 added 8+ and 9+ buckets on July 7, 2025 to close a gap in its reporting. Amazon DSP reports bins from a single view up to ten or more.
Practical rule: never accept an average frequency figure without asking for the distribution behind it.
Caps versus targets
A frequency cap refuses impressions above a ceiling. A frequency target actively buys toward a level. Not the same thing, and the difference shows up in your delivery curve.
Google’s Fixed CPM bidding accepts a target count of exposures per user plus a time window. DV360 introduced monthly target frequency for YouTube on December 2, 2024. On the reservation side, Meta renamed its Reach and Frequency buying type to Reservation in October 2023; per Jon Loomer’s documentation, it carries a 200,000 minimum reach, is limited to awareness and engagement objectives, and defaults to target frequency rather than a cap. LinkedIn joined the party on July 2, 2025 with caps of three to thirty impressions per member account over seven days.
Duplication is where the money leaks
Separate buying platforms cannot see each other’s counters. Amazon’s own September 2023 data put the average unique reach gain from Frequency Groups at 6 percent, and one CPG example in its documentation found 34 percent reach overlap across orders from three brands in a single business line. That is repetition nobody planned and nobody budgeted.
The industry is patching it. Amazon launched cross-account reach reporting on April 2, 2026. DV360 added Unique Reach Overlap dimensions in April 2026. Google Ads shipped Cross-Media Reach in June 2024. Nielsen rolled four-screen deduplicated measurement into Italy on June 8, 2026 and extended it in Japan on August 3, 2026.
Definitions move, delivery does not
Here is the trap. On June 2, 2026 Google applied a Total Co-view definition to seven reach and frequency metrics, five of them frequency thresholds. Reported unique users went up. The same impressions were served. Video Advertising Bureau research puts co-viewing at 60 percent on premium CTV versus 45 percent on YouTube, and three organisations count it three different ways.
DV360 also moved pricing into the frame, replacing a bid-request frequency metric with cost-based alternatives on August 26, 2025, including Cost per reach.
What to actually do
- Report reach in absolute unique users alongside percentages, and always pull the frequency distribution.
- Audit overlap across DSPs, retail media and social before adding budget.
- Decide explicitly whether you want a cap or a target, and document why.
- Log definition changes from platforms so a metric jump is not mistaken for performance.
- Insist on viewable impressions as the qualifying unit; MRC standards say an ad that never entered the viewport should not create reach.
Fifty years, still unsettled
Herbert Krugman argued three exposures may be enough in 1972. Michael Naples turned it into planning doctrine in 1979. Erwin Ephron flipped it in 1997, arguing continuous broad reach beats accumulated exposure because you cannot predict purchase readiness. On February 3, 2026 Google’s own researchers Pablo Perez and Jesus Martin Calvo pushed the reach side again, citing YouTube Brand Lift data where second impacts indexed at 1.5 to 1.8 rather than 2.
Translation for your next plan: treat effective frequency as an editorial choice you defend, not a fact you inherit.
Source: PPC Land
