And so I was interested to see how SAY Media was going to go about trying to measure the branding effect of its interactive rich-media ad units. The company just rolled out a method of measuring a sample of users who are interacting with their ad units to see how engagement with the ad lifts the standard brand metrics, as well as other consumer attitudes and intentions toward a brand. SAY deploys what it calls an AdFrames unit that expands to fill the screen when the user's mouse hovers over it for a requisite countdown. The company works on a cost per engagement rather than a standard CPM model. If the user lets the ad expand and enters the ad experience, they are considered engaged. Until now the main metric used to determine the effectiveness of the engagement has been time spent in the ad. "We wanted to understand beyond the proxy measure of time spent or complete views and doing things in the ad," says Matt Rosenberg, VP Solutions, SAY Media. "There isn't a clear line to the brand metrics. Everything in between are proxy measures that advertiser have to assume they know the meaning [of]."
While post-campaign brand effects studies can be done, the results often come weeks and months after a campaign has finished. SAY was looking for a way to make the results available in a timely enough way to perhaps affect the optimization of the current campaign. And the problem with measuring an ad unit that works by engagement instead of exposure is having to harvest enough of the engagers through the usual retargeting methods. In the end, Rosenberg says these campaigns really were being tested on their exposure value rather than their engagement value. In order to find that direct line between engagement and changed attitudes towards a brand, the method of measurement would have to be immediate, and successfully capture the people who were actively interacting with the ad unit.
Working with measurement firm KN Dimestore, SAY came up with a way to intercept users at the end of their engagement. When I walked through a demo, the sponsor's ad message popped up to fill the screen with multiple video clips to view. When I closed the ad, the original un-expanded ad unit changed to an invitation to take a survey. "The survey is served in the banner so it doesn't take the user away from the page," says Jordan Schlachter, SAY's director of research. "We ask between six and eight questions, since we found that responses dwindle drastically after that." Generally there are four standard branding questions and then other customized queries. A control group sees only the survey on the page.
"The first time we ran this survey we were hoping to be able to get an engaged cell of 100 people to get a minimum of what we considered valuable," says Rosenberg. "We got 350. In a recent campaign about halfway through we have 1200 completed surveys from engagers. It is getting large enough to cut fairly granular data." And ultimately it is scale that will make this approach to real-time research on engagement interesting. Over time they will be able to build a database of campaign results across categories to better understand how engagement with an ad's creative translates differently to brand metrics based on product type or perhaps even the medium applied. One can imagine an even more sophisticated metric with a large enough sample base that ties specific activities in an ad (playing video, answering a poll, etc.) with particular branding effects. Ultimately, the method could render a new kind of optimization on rich media that allows multivariate testing on various elements and messaging in real time.
But these are the sorts of next-gen metrics that need to await scale. While the product is too new to render hard results, Schlachter says that the branding lifts from engagement are not entirely unexpected. But what is interesting is being able to corral effectively the people who have engaged in these richer experiences immediately after the event, and create rich-media focus groups on the fly. The method speaks to the direction metrics should take, from measuring whether engagement happens to better understanding what it does.