# It's Time To Toss Average Frequency Into The Bucket

You are building an ad campaign. You’ve completed vast amounts of research, worked hard on the creative, media plan and set your goals. Now it’s time to figure out frequency -- how many times a person should see each ad. This task is extremely important. If your ad gets either too many or too few views, the whole campaign can derail.

And once you’ve set the frequency, what is the actual probability of meeting it?  Most premium publishers and ad exchanges claim they’ll help advertisers achieve the perfect frequency target, by ensuring that the average frequency meets a set frequency cap. The key word here is average. But is a frequency cap what you need? If you advertise with different premium publishers, how do you choose the cap for each site? How do you avoid overexposure to people visiting more than one of those sites -- or underexposure to those visiting just one?

Achieving average frequency can mean you are not necessarily controlling ad exposure to those people reached. For example, a campaign reaching 1 million people with an average frequency of 10 impressions could mean either of the following:

• 1 million people were exposed to 10 impressions each (the desired result), with total impressions of 10 million.
• 800,000 people were exposed to one impression each (serious underexposure for a total of 800,000 impressions) and 200,000 others were exposed to 46 impressions each (serious overexposure for a totall of 9.2 million impressions).

In the second case, there are again 10 million total impressions, and again an average frequency of 10 impressions.  But while the first scenario is successful, the second one scenario misses the target frequency with both under and over exposure.

So what can you do to avoid such a scenario? Instead of looking at average frequency, you can look at frequency buckets: the number of people exposed to an exact number of impressions. In the example above, for example, you’d want to know how many people fall into frequency bucket 10 – that is, how many saw the ad the desired 10 times?  In the first scenario, the answer is 1 million; in the second scenario, zero – a total failure.  Some advertisers would also consider any user exposed to the ad nine or 11 times a success; they would look at a frequency bucket of 9, 10 or 11 impressions.  And again, in the example above, the first scenario would result in the desired 1 million, the second in zero.

The question advertisers should be asking is, “How many people actually had the proper amount of exposure?”  Not, “What is the average exposure?”

So avoid averages.  And raise the bucket!

4 comments about "It's Time To Toss Average Frequency Into The Bucket".
1. David Kissel from InStadium, October 23, 2012 at 3:22 p.m.

Ronit - Great post and you are trumpeting the problem we see across a multitude of brands and categories. It used to be average frequency was an acceptable metric for media planning, because the difference between the highest frequency and the lowest frequency for a campaign was fairly narrow. Today, due to audience fragmentation and lower TV ratings, we see a tremendous polarity between the frequency of the highest TV viewing quintile and the lowest TV quintile. Simulmedia's recent work in "The Book of Reach" lays this out in black and white for a number of brands, who are literally pounding a small segment of their intended audience with excessive frequency, while missing altogether the lighter viewers. The solution? Shift impressions from a portion of your high frequency cable weight to vehicles that both attract these harder to reach viewers, at scale, and deliver sight, sound and motion communications. Running your television advertising in sports stadiums and movie theaters offer advertisers this opportunity, and for many brands, are worth exploring.

2. John Grono from GAP Research, October 23, 2012 at 6:27 p.m.

Does that mean that in the US there are people who DON'T take into account frequency range buying? Downunder it became standard practice when TV optimisers became en-vogue in the mid '90s. You'd set the optimiser to get programme combinations that optimised on (say) 2-4 TV exposures in a week. The client goal would be to have (say) 85% of all exposures in that range.

3. Rebecca Caroe from Creative Agency Secrets, October 23, 2012 at 8:32 p.m.

I find clients often don't understand basic stats measures e.g. averages, medians.

All rather irritating to endlessly explain... or subscribe them to OccamsRazor blog for a crash course!

[I do wish I could do that sometimes!]

4. Pete Austin from Fresh Relevance, October 26, 2012 at 5:05 a.m.

Exactly, the question is “How many people actually had the proper amount of exposure?”. But the proper amount of exposure will vary for different people, so trying for a fixed number such as 10 for everyone is strange.