Commentary

Calculating Your True Cost Per Acquisition

I had a conversation with a travel CMO excited at finally having cracked the code on her Display retargeting campaigns (prior to Google offering its own version of this link). Her cost per acquisition/booking (CPA) from this campaign was, let's say, $18. And, let's say she makes an average of $36 on each booking. So, great -- she's got it figured out, with a 100% return on ad spend (ROAS).

Well, not exactly. While retargeting is a great tactic to get lost leads back to your site, what the vendors won't tell you is that it's greatly biased to overemphasize conversions if evaluated on a post-impression (a/k/a view-through) attribution model.

Don't get me wrong -- I, for one, am a believer in the post-impression conversion. While Display ads have suffered from declining click-through rates since they debuted in the 1990s, consumers are still measurably affected by them and clearly respond to them to some degree. They may see that great deal to Cancun and go to your site on their own, and a great many consumers do this (there is even evidence that good Display drives search). So, clearly, I believe there is some effect that isn't related to clicks.

advertisement

advertisement

But, what I see happening with marketers like the CMO in my story is they attribute retargeting (and Display in general) on an all-or-nothing basis. In her case, it was fully attributed, regardless of whether the ad actually made the booking happen. Consumers come to book on her site all the time, but as a part of that process they might do research, confer with a family member, and then book later.

In the meantime, her retargeting campaign bombards these consumers with Display ads reminding them to come back. A great strategy, but one that undoubtedly shows an ad to someone who was going to book regardless of being shown an ad.

This is called cannibalization. Some percentage of bookings attributed to the retargeting campaign was going to happen independent of the retargeting campaign. If we fully attribute these bookings, we inflate the contribution of the retargeting campaign and underestimate the true CPA.

Let's say her cannibalization rate was 50%, or that half the bookings were going to occur with or without the ad campaign. This doubles her CPA, making it $36 and just at break even with her revenue per booking (a 0% ROAS).

Here's the solution: perform a test. Separate part of your retargeting population into a control group that gets 'retargeted' with an Ad Council ad, and note the conversion rate for this population. Subtract this from your actual retargeted ad conversion rate to reveal your true conversion rate with cannibalization taken into account. From there, you can calculate your true CPA.

Use this true CPA in evaluating your campaign's profitability, and you'll have a much better handle on what's working and what's not.

1 comment about "Calculating Your True Cost Per Acquisition".
Check to receive email when comments are posted.
  1. Albert Lin from Bigcommerce, April 13, 2010 at 7:55 p.m.

    Last-impression attribution will never be a good way to fairly value your advertising, and that skews the incentives for your vendors. The CMO who pays on the last impression will successfully get his/her vendors to fight harder for the people further down the "funnel", thereby bidding up prices for that deep-funnel inventory, increasing cannibalization and costs.

    It should come as no surprise that search and retargeting are usually the deepest into the funnel you can get and you're probably overpaying.

    Paul is absolutely right in advocating well-controlled tests to calculate a lift over control. Other, more holistic analytics may work, like marketing mix modeling.

Next story loading loading..