How can you accurately measure Display campaign ROI?

Edwin Dewez

Display campaigns are probably the most difficult of online channels to assess. The reason being that most of their impact resides with a user seeing and hopefully remembering an ad, rather than clicking on it. The industry average click through rate for display ads is 0.08%: of 10,000 banner views, only 8 will generate a click and a visit to the site (and some of these clicks are accidental but that’s another topic). 

There are typically 2 ways to measure the efficiency of your display banners: you either trust the provider that you use (Google Display Network, Double Click, Criteo, Quantcast…) or Google Analytics. Both are poor sources of information and here is why with a real life example from one of our clients. 

This Wizaly client in the travel vertical wanted to compare 2 different DSPs and put them to the test. In order to ensure a fair comparison, they split the US equal regions based on population, wealth, conversion rates and asked each DSP to target one of the 2 regions with an equal budget. The DSP with the best result was going to be kept as the retargeting provider of choice going forward. 

Geographic split for the test

Both providers used a 24 hour view-through window and a 30 day click-through window. In other words, they considered that a conversion meeting these view-through and click-through criteria was 100% attributable to them.

At the end of the test, the 2 vendors reported 1,583 and 1,645 conversions respectively. This seemed a huge number and a very positive outcome.

When looking at Google Analytics, the results were unsurprisingly very poor: 4 and 5 conversions respectively. This is easily understandable: Google Analytics does not record the view-through impact and measures a channel success on a last click (non direct) basis. In other words, for a sale to be attributed to one of the 2 vendors, someone would have had to click on the ad and purchased from the site immediately.

Wizaly gave them a very contrasted picture: our software deduplicated the vendor’s results and weighted their credit by taking a number of signals into account:

  • Real viewability (we counted a view only when 50% of the ad was displayed on the user’s screen resolution for at least a second according to IAB standards)
  • Engagement metrics of the person visiting the site after having seen an add (how many pages did the user see / how long did they stay on the site?)
  • The real added value of having one of the vendors in a user journey (did it make a statistically significant difference when one of the vendors’ ads was present in the user path within the relevant view-through window?)


The result? One of the vendors was credited with 148 conversions while the other one only got 34. In terms of revenue, it represented half a million $ over the test period alone. By trusting the individual vendors or Google Analytics, this contrast could not have been identified and a $10m yearly opportunity could have been missed.

DSP performance