by: Roger Dooley
You never know who you’ll run into at a trade show. I stopped by Ad:Tech Chicago earlier this week, and in the exhibit hall reception spotted Bob Cringely, PC industry pundit extraordinaire.
I’d met him briefly at a previous WebmasterWorld Pubcon, and ambled over to say hello. I found that Cringely wasn’t wearing one of his usual hats (author, columnist, broadcaster), but was attending as the founder of TaguchiNow. (Cringely is the single reason I maintain my subscription to InfoWorld – his column always exhibits a dry, ironic sense of humor that is otherwise absent in the bland world of IT.) With him was Mario Fantoni, President and CEO of TaguchiNow. The premise of the firm is the perfect elevator pitch: they claim to have figured out how to apply Taguchi Method statistical analysis, widely accepted in quality management, to advertising; furthermore, they boldly guarantee that they will double your advertising results or you’ll pay nothing.
Their methods involve creating alternate ad vehicles, testing them, and using Taguchi math to analyze the results. As a direct marketer, I find this approach highly appealing. When I published catalogs, we’d periodically try split run tests on covers or other catalog elements. It wasn’t uncommon to find that just putting a different cover picture on the same catalog could cause a 10-20% difference in sales. Seeing results like this, I always wondered about the many mailings where we didn’t try to optimize the cover… were we passing up an easy double-digit increase in sales (and much bigger increase in profits)? Doubling ad results is quite a challenge, but Fantoni says the firm has yet to have to refund a client’s money. Taguchi method analysis goes way beyond A/B split run testing – rather, nine, or even eighteen, versions can be tested simultaneously, greatly compressing the time needed to identify the best performer.
How does this statistical ad optimization involve neuromarketing? It doesn’t, at least not yet. Indeed, TaguchiNow’s approach is the antithesis of the kind of data neuromarketers are collecting. Fantoni doesn’t really care why a particular ad outperforms another one; it’s sufficient to know that it does. Neuromarketing and neuroeconomics researchers, on the other hand, are trying to discover the neurological basis for why ads work or don’t work.
One of the big problems with the state of the art of neuromarketing today is that researchers are developing colorful brain scans showing how subjects react to ads, but there’s not much in the way of published work that connects the scan results to ultimate consumer behavior. As we pointed out our comments on the fMRI analysis of the SuperBowl ads (see Super Bowl Ads: GoDaddy Girl 1, Neuroscientists 0), one of the ads determined to be a “loser” was actually the leader in driving website traffic.
In chatting with Cringely and Fantoni, I learned that they had existing sets of ads that had already been compared in detail using actual customer purchase data. They know which ads in a given set sold best, and which were duds. It was immediately evident that it would be a no brainer (sorry about that) to backtest these ads using the same fMRI techniques now used to make vague comparisons between TV commercials. If a correlation between specific brain activity and actual ad performance could be found, then future fMRI studies could be used to evaluate new ads with some ability to predict performance and choose the most profitable ads. TaguchiNow, of course, is far from the only firm with sales performance statistics for sets of different ads. One wonders, though, if Taguchi math could be applied to the scan results themselves to tease out correlations that aren’t immediately obvious.
Sooner or later, an enterprising neuroscientist will launch a project to compare ads for which extensive performance statistics are already available. Will that lead to identifying the location of the hypothesized “buy button”? That’s doubtful, in my opinion. Consumer decision making is far too complex to be localized in a single brain location. Still, I think one could reasonably expect strongly performing ads to have some distinctive brain activity characteristics. These characteristics will likely NOT be the same for all ads, since one successful ad might work because it really grabs the attention of the viewer, another because it shows the viewer how to reduce “pain” (be it physical or emotional), and so on. A realistic goal for such research would be to identify a handful of brain activity centers that are important to different types of marketing pitches. For example, if the goal of an advertising campaign is to improve brand memorability, it seems likely that such areas could be identified by combining brain scan data with field test results from the actual ads.
Published research that ties together measured brain activity with real world consumer behavior, even if the data was case-specific and difficult to generalize, would go a long way to resolving the neuromarketing credibility issue that is so often raised.