Reach and frequency: separating average exposure from useful pressure.
Key distinctions
Three conditions before calculating.
People, not impressions
Impressions measure opportunities to see. Deduplicated reach requires an identity or an overlap model between media. Adding up platform reach systematically overstates unique people.
Distribution, not average
Average frequency among the exposed ignores the unexposed and the dispersion. Thresholds of 1+, 3+ or 8+ make minimum reach, useful repetition and potential waste visible.
Window and creative
The useful number of contacts depends on duration, novelty, message complexity, format attention and creative rotation. A universal threshold is indefensible.
Method
Designing media pressure from the distribution
- 01Set population and window
Define the eligible segment, the area and the period; national monthly reach does not answer a weekly local decision.
- 02Deduplicate the media
Use consented identifiers, a panel or an overlap model, and publish the method and a range when deduplication is not directly observed.
- 03Build the classes
Split the population between 0, 1, 2, 3 to 5, 6 to 8 and 9 contacts or more, and connect each class to attention, memorisation or behaviour.
- 04Reallocate the duplicates
Reduce the inventories that concentrate over-exposure and move impressions towards media or segments that increase effective reach.
The original contact distribution
Two plans with the same impression volume are compared on total reach, 3+ reach and overexposure share, not only on average frequency.
reach · frequency 5.8
reach · frequency 5.3
Four million impressions, but two opposite distributions
Plan A reaches 690,000 people with 5.8 average contacts, of whom 410,000 at 3+ and 170,000 at 9+. Plan B keeps four million impressions, reaches 760,000 people and 450,000 at 3+.
The average frequency of plan B falls to 4,000,000 ÷ 760,000 = 5.3. Reach at 3+ increases by 40,000 people while over-exposure at 9+ falls by 55,000.
Plan B is preferable if measurement confirms that 3+ is a useful threshold. A drop in average frequency is not a loss when the distribution becomes more productive.
Acceptance conditions
What must be true to act.
- 01
Reach is deduplicated over an explicit population and window.
- 02
The frequency classes and the share of unexposed people are published.
- 03
The useful threshold is connected to evidence of attention or behaviour.
- 04
Creative rotation is built into the over-exposure analysis.
Limits
What this analysis does not prove.
- Reach modelled across platforms depends on overlap assumptions.
- A measured exposure guarantees neither visibility, nor attention, nor comprehension.
- Effective thresholds vary with creative, objective, category and the interval between contacts.
References
Works cited.
- Goodhardt, G. J., Ehrenberg, A. S. C. & Collins, M. A. (1975), The Television Audience.
- Naples, M. J. (1979), Effective Frequency: The Relationship Between Frequency and Advertising Effectiveness.
- Tellis, G. J. (1997), “Effective Frequency: One Exposure or Three Factors?”, Journal of Advertising Research.