One of the primary benefits of digital marketing over traditional media is the vast improvement in measurability of success. It’s easy to determine certain metrics such as impressions, clicks, and conversions. However, when it comes to determining proper attribution for online ads, particularly when you advertise via multiple online channels, the answer is often murky.
Though traditionally, display media has been sold on the basis of impressions, often at cost-per-thousand impressions (CPM), in the quest for proper attribution, many online marketers prefer to work on a cost-per-action or cost-per-acquisition (CPA) basis. Ostensibly, CPA is a fairer and more accurate pricing scheme than CPM, which does not guarantee that prospective customers will take the desired action. However, CPA schemes rarely rely on accurate attribution models and there are many instances where a CPM pricing model is actually better for the advertiser than a CPA model.
When CPA Gets It Wrong
Often, CPA deals are based on last-touch attribution, meaning the last banner or search ad a user clicks on or views before converting is the ad that receives all the credit. Last-touch attribution is incapable of telling the whole story, as each touch with the customer makes some incremental effect on the eventual purchase. This point becomes particularly salient with retargeting, where a consumer has often had over 15 touch points before converting. Relying on the last touch overvalues the final ad, and undervalues the initial and mid-level touch points.
Online marketing campaigns are often driven largely by performance data. Marketing should be data-driven, whenever possible, but when a marketing campaign is driven by the wrong data, you run into problems. This is how last-touch attribution gets you in trouble. If you optimize your ad spend based solely on the performance of the last touches, you are failing to incorporate key data points regarding the influence of the initial touch points on the purchase decision, resulting in you optimizing your campaign for sub-optimal results.
The Case of Double-Counted Attribution
If you are advertising online through multiple providers, using CPA for each provider is probably not an effective solution. As each advertiser anonymously tags users, they will be counting conversions in their own siloed data system, and if you’re using last-touch attribution, you could end up paying for the same conversion twice. We came across a client once who had previously used a CPA model with various advertisers, retargeters, and affiliate networks, and could not avoid the problem of double counting conversions. Even when they shortened the view-through conversion window for display to a single hour, they were still paying two, three or four times for the same conversion. (View-through conversion refers to a conversion where the user converts after seeing an ad, but did not click on any ad unit.)
The more complicated your network of advertisers, the more likely attribution double-counting is likely to plague you.
Limiting Your Potential
When you run on a CPM model, you have more room for improving your return on ad spend. Regardless of how many conversions you get, the media cost is constant. If you have a high-performing campaign, your effective cost per action becomes quite low, and as your number of conversions increase, your return on ad spend skyrockets. Particularly with high-return advertising techniques, like retargeting, you have the potentially to drastically increase the number of conversions without increasing your ad spend. When you pay on a CPA basis, the number of conversions you can see is strictly limited by your budget.
The Problems with CPM and How to Address Them
And yet there is another side to this story. The more conversions you see, the lower your effective CPA, so it follows that if you have a low number of conversions, you will have a high effective CPA. Low performing CPM campaigns will have incredibly high effective CPAs. The best way to address this problem is to make sure you and your advertiser are optimizing for high performance from the beginning of your campaign. If you have a high volume of low performing impressions initially, you could be wasting a significant portion of your ad spend. At ReTargeter, for example, we use site-based optimization to serve more ads on sites that demonstrate higher click-through and conversion rates, and fewer ads on sites that have lower click-through and conversion rates. Once we’ve collected enough data, the optimization process beings, which allows us to improve performance relatively early on in a retargeting campaign and help clients avoid wasted ad dollars.
If you continue using the same campaigns, your performance may decrease over time, and the value derived from CPM decreases as your effective CPA increases. This concern is relatively simple to address: update your campaigns by switching up your copy and rotating your creatives. If you vary your ad content, you will be able to avoid decreases in performance.
Your advertising provider always has an incentive to bring you conversions, even when operating on a CPM. If they don’t provide a high performing service, you’ll simply leave. That’s a better incentive than any other for an advertising provider to bring you as many conversions as possible, regardless of how they’re compensated.
Online attribution modeling is in fact a burgeoning industry, and companies like Adometry and Encore Metrics use sophisticated probabilistic models to determine how much credit each online touch point actually deserves. These services are complex and can be quite expensive. For many companies, this is unlikely to be necessary. If you are really interested in finding the precise due each online touch deserves, services like this are the only true way to do so. However, using an inaccurate attribution model like last-touch is often worse than using none at all. Until precise attribution modeling becomes commonplace, CPM pricing models will often be better for advertisers.