Average frequency is impressions divided by people reached, but it says nothing about distribution.
Average frequency is impressions divided by people reached, but it says nothing about distribution. An average of five contacts can conceal a majority reached once and a small overexposed group. Publish 1+, 3+ and above-saturation reach by segment on a consistent window.
Decision and method.
Reallocate impressions across media to increase useful reach without funding excessive repetition. Define eligible population, geography and period; deduplicate media using consented identifiers, a panel or an overlap model, stating the method and a range if it is modelled. Split population into 0, 1, 2, 3–5, 6–8 and 9+ contacts, then reduce inventories that concentrate overexposure and move impressions to media or segments that raise effective reach.
Worked example.
Both plans use 4,000,000 impressions. Plan A reaches 690,000 people at 5.8 average contacts, including 410,000 at 3+ and 170,000 at 9+. Plan B reaches 760,000 people and 450,000 at 3+: 4,000,000 ÷ 760,000 = 5.3 average frequency. It gains 40,000 people at 3+ while reducing 9+ overexposure by 55,000. B is preferable only if measurement confirms 3+ is useful; lower average frequency is not a loss when distribution becomes more productive.
Checks and limits.
Deduplicate against an explicit population and window; publish frequency classes and non-exposed share; tie useful threshold to attention or behaviour; include creative rotation. Cross-platform reach relies on overlap assumptions, exposure does not guarantee visibility, attention or comprehension, and effective thresholds vary by creative, objective, category and spacing.
Resources and sources.
Download the contact distribution. Related: link exposure and creative quality, arbitrate communication horizons, reallocate by marginal return. Sources: Goodhardt, Ehrenberg & Collins (1975); Naples (1979); Tellis (1997).