Effective Frequency on Connected TV: How to Set and Measure It

"Effective frequency" is an old TV idea: people need to see an ad enough times to remember it, but not so many that you waste money. On CTV, the idea still holds. What changes is how hard it is to see real frequency across many apps and partners.

Set a range, not a number

Inputs that push frequency up:

• A complex message, such as a new mechanism of action or a new condition.
• A crowded category with heavy competitor TV.
• A short flight.
• Risk information that needs time to be understood.

Inputs that push it down:

• A simple reminder.
• An audience already familiar with the brand.
• Long flights with steady weight.

Write the range, for example "2 to 4 exposures per household per week," and say why.

Measure the distribution

Average frequency hides the problem. A plan averaging 3 per week might have:

Hypothetical numbers. In that case, 40% are below the range and a quarter above it. The average looked fine.

Only deduplicated measurement across partners shows this. See deduplication vs. frequency capping.

Find where response flattens

If you have outcome data by exposure level (site visits, condition searches, or Rx outcomes from a privacy-safe study), plot response against frequency. Typical findings:

• Response rises quickly from 1 to a few exposures.
• It flattens after a point.
• Beyond that, extra exposures add cost but little response.

Treat this curve carefully: households that see more ads may differ from those that see fewer, for example by viewing more TV. A holdout or randomized frequency test gives cleaner answers. The CTV incrementality testing article covers designs.

Adjusting the plan

• Too many households at high frequency: lower caps, consolidate partners, add reach-oriented publishers.
• Too many at 1: extend flight, raise weight on fewer partners, or accept that reach is the goal.
• Uneven by partner: shift budget toward partners delivering within range.

Planning tools

CTV frequency estimators and reach planners help set budgets. Their outputs depend on assumptions about audience overlap between partners. Ask what data the tool uses, then compare its forecast with deduplicated delivery after the first weeks.

Common mistakes

• One frequency number copied from another brand.
• Judging frequency by partner-reported averages.
• Ignoring the low-frequency tail.
• Reading a frequency-response curve as causal without a test.

Practical takeaway

After two weeks of any CTV flight, ask for the deduplicated frequency distribution. Move budget from partners driving the high-frequency tail to those adding new households within your range.

Editorial note. Analysis and frameworks are the author's own and do not represent Acxiom or any current or former employer, client, or named platform. Examples labeled hypothetical or illustrative are not results from real campaigns. Nothing here is legal, regulatory, or medical advice.