I’m not sure how appropriate the title to an R.E.M. song is to this column, but I needed a title. It’s the first thing that came to mind. Now, I can’t get the song out of my head.

Actually, I should be asking, “What’s the frequency, Young-Bean Song?” Song is director of analytics for Atlas DMT. I spoke with Song, who believes his company has come up with a way to help online advertisers make the most of their impression delivery to maximize conversion rates and cost per acquisition.

Optimal frequency has been considered from a branding perspective by companies such as Dynamic Logic and Insight Express. However, there hasn’t been a lot of research on optimal frequency for online direct response campaigns.

Atlas looked at 38 advertiser campaigns across multiple categories. It found a large percentage of impressions go to users who are only exposed to the message a single time. A smaller percentage of impressions go to users exposed to the message two times, and so on. Finally, when you get to users who are exposed 11-plus times, there’s a significant increase.

Frequency Distribution Graph

I think we can agree users who see an ad once or twice are underexposed, users who see it nine or more times are overexposed. What if we could take those “wasted” impressions and either increase frequency against the underexposed users or reach completely new users?

Atlas did the math and found by implementing a “sensible” frequency cap, an advertiser can lower its cost per acquisition 10 to 30 percent.

Another issue the study considered was conversion rates at different frequency levels. Overall, Atlas found conversion rates are higher at lower frequency levels. In fact, the highest conversion rates occur on the first impression. The second and third impressions also garnered respectable conversion rates.

Could we cap our campaigns at one impression to achieve the highest conversion rates? Yes. Should we? I don’t think so. We must also consider volume. Capping frequency at one would result in fewer overall conversions.

The study also compared the most effective frequency with the most profitable frequency. We must define a cost per acquisition at which our clients still make a profit and raise the frequency cap to reach those users who won’t convert on the first impression. The right frequency cap for your client may be three or five impressions. You’ll have to determine that based on your strategies, target audiences, media costs, and cost-per-acquisition goals.

What does this mean for you? It’s a no-brainer. Conducting your own optimal frequency study is a way to reduce waste, reach more prospects, increase conversion rates, and maximize total acquisitions.

What does it mean for publishers? In talking with Song, my first thought was publishers would hate this. Implementation will wreak havoc on their inventory management. Users who are currently exposed to 11 or more impressions on a single campaign are the site’s lifeblood. Those are the users who enable publishers to promise you millions of page views and, thus, millions of impressions. If you cap the frequency at, say, three, publishers will have to find a few more advertisers to sell the same number of impressions they’re selling today.

Song suggests a frequency cap would allow publishers to charge a premium for those more valuable impressions. Though I agree premiums would be in order, it’s going to be a tough sell. The premiums publishers will want to charge to make up for unsold inventory could be unattractive to advertisers. The good thing is we now have a model to test against to find out if this works for our clients.