From what I’ve heard, only about 18 percent of web users ever click on a display ad. But studies have clearly shown that display ads affect people’s behavior. Sometimes people see the ad and enter the URL directly. Sometimes it leads them to search on a brand-related term.

For example, a web user might see an ad for a Dell Lattitude on the side of the page and then type “Dell Lattitude” into his handy little google search box at the top of his browser. At that point he might click on an organic link or a paid link.

What advertising campaign is going to get credit for any resultant conversion? If you’re measuring by clicks, it won’t be the display ad.

Which leads us into “view-through conversions,” where the display network takes credit for some percentage of conversions based on the fact that the ad was displayed to the user some time before conversion. (The time window can vary.)

But is that fair? Maybe the person was going to convert anyway. Maybe he never even saw your ad.

Another option is re-marketing, where a visitor comes to your site, gets a cookie, then goes out into the world and sees ads to draw them back to your site. It’s effective, but how do you measure how effective it is? What are you measuring against what?

It seems to me that the logical way to do this is to use a combination of re-marketing and an A-B split.

IOW, some number of people come to your site and get a cookie. That population is split into two groups. Group A goes off into the world and isn’t exposed to any of your ads. Group B sees your ads. Compare the behavior of Groups A and B. Any difference can be safely attributed to the ads on that network.

For some odd reason, display ad networks don’t seem to be able to do this, and I’m not sure why. It doesn’t sound technologically difficult, and it would convincingly prove the effect of the ads.