How many times do you have to see an ad before it has its intended affect on you? One time? Two times? Ten times? Surely anything over 50 is just overkill – right? Before you go on reading – think of a number in your head that represents how many times you need to see an ad before it affects you. Got it? Ok, keep reading.

Optimal ad Frequency is probably one of the most highly ambiguous subjects in advertising. If you were to ask 10 different people their opinion on ideal ad frequency, you would be likely to get 10 different answers. At the core of the Ad Frequency issue, there is a simple principle: you want to make sure that people see your ad enough times for the message to stick.

In an attempt to quantify this principle, in your head you can picture an “S” curve (like the generic example I’ve copied below) where “ad effectiveness” is on the Y axis, and “number of exposures” is on the X axis. In this example, there is an “ideal” frequency of exposures that intersects the inflection point of the curve. Think of it like Goldilocks – not too few exposures (for fear your message will not stick) and not too many exposures (because you’ll be wasting your money) – you want to get it just right.

clip_image002

Most people will probably agree with the shape of the curve and the inflection point goal – however the major dispute is the scaling of the curve. At what frequency level does the inflection point exist?

Although this question seems simple – it is actually an impossible question to answer. There is no one “ideal” frequency – it varies from person to person. Each time a person sees an ad, it is a unique experience – it simply cannot be effectively modeled in aggregate.

We can make models, estimates, and hypotheses – but in real life, the way advertising affects each one of us is as unique as the number we thought of in the first stanza of this entry.

Frequency
  • I’m not sure I agree entirely.

    Actually, it’s possible that an S-curve could exist ONLY at the aggregate; e.g. that the individual response to another ad does not follow that distribution. It equally well could be that it’s only when the impact of the ad on multiple people is combined that an S-curve results.

    However, I agree with your major point; that is, aggregate-level statistics cannot tell you how many times you want to show an ad to each person. That is, it cannot (accurately) give you info on individuals; just on the average affect across individuals (and thus will vary by average, e.g. gender, education, location, profession, hobbies and so on).