• by Andrew Ettinger , May 1, 2009

I have a really nice blue pin-striped suit from barneys. It looks great and when I wear it I feel like a million bucks. Oddly, nobody ever compliments me on it. On the other hand, I have a cheap, very loud jacket that I bought on sale at Filene's Basement. Total strangers on the street tell me how much they like it. One just never knows what the public is going to like.

Consumers are elusive. No matter how much we research their likes and needs, they continue to change their minds. Luckily, the Internet allows us real insight into their behavior. Rather than guessing which ads provoke, finally we can know for sure. In today's information economy, the consumer controls the flow of information. The Web allows people to research products before they ever actually touch them. This is great for marketers because this can eliminate a lot of guesswork. Now we can track consumers online and determine exactly what resonates, and what is important to them. We can use that knowledge to better market our products. First, though, we must listen to what they are saying.

Anytime I introduce a client to Paid Search I advocate one basic principle - let the consumer decide. Rather than choosing only a few words, I recommend bidding a long list of terms. Because search is pay-per-click, bidding on multiple keywords costs you nothing but gains you everything. Search is a real-time focus group. We can test different combinations with no budgetary risk. This alone is a sea change in marketing. When people click on an ad, people are telling us what is important, rather than what we think is. At last we can determine which ads specifically motivate consumers, and the results can be surprising. Like they used to say on MTV: You think you know, but you have no idea.

When working with display (banner) advertising, the same principle applies. We can tease out the best tactical solutions by using many creative executions. Third-party ad servers allow advertisers to change online creative without even telling the publisher. I suggest that advertisers use vastly different ads when starting out. Do not bother measuring a red background versus green (that can come later, if it must). Instead, start with dramatically different executions. By doing so, we gain the most insight in the least amount of time. Very soon we see which ads generate clicks (or interactions or sales) and which do not. This is not A/B testing - this is A/Z testing.

The fundamental principle is that consumers' actions are a feedback mechanism. When they choose to click, they are providing feedback that we can use to tailor our marketing messages. These ideas can be applied offline, too. More specific commercial ratings will finally propel the TV community into the 21st century. Someday, TV buyers will use second-by-second ratings the same way online buyers use clicks. Only then will they be able to measure how engaging TV ads actually are. Individual commercial ratings will allow buyers to optimize in real time, just like we do with banner clicks. It will be up to agencies to measure and respond to how consumers react (or perhaps simply don't change the channel). Offline will eventually steal from the online playbook.

Some might argue that focus groups already provide a true consumer understanding. I think that is naïve. We have all seen ads that make us scratch our heads. How did those get past the focus groups, we ask ourselves. Results from focus groups are malleable; interpreting qualitative data is a skill, not a science. Experienced moderators may write impartial reports, but those recommendations can be easily ignored by people with other agendas. I have seen focus groups utterly reject a campaign only to see that data overridden by the account planner. I am sure the reverse happens equally as often. Even the worst ideas can pass muster if presented in the right way. That must be how NBC picks most of its primetime schedule.

I really don't understand why so many people like my coat. It's not that different or  special, but clearly it resonates. We can't 100-percent predict what will make other people react. Because consumers are so elusive, we have to hedge our bets. Practically speaking, this means we need more creative executions. The model calls for optimizing based on actual results, rather than guesswork.