Commentary

Getting To The World Series Of Digital Media Buying

Media buyers and baseball coaches are a lot alike. The media buyer's team is, of course, the client or brand and it is the media buyer's job to make the most out of all the assets at his/her disposal.

The media buyer's players are the ad units or creative elements for a given campaign. Like players, each of those ad units has its relative strengths and weaknesses and some cost a lot more than others. Sometimes you, as the media buyer, would not have picked some of the creative for your team, but you still have to find a way to work with them. Your media budget is essentially the salary cap.

Successful baseball coaching and media buying are both largely numbers games. Both professions are flooded with statistics and data of all kinds, but the best coaches and media buyers know how to focus on the relevant data and actively apply it while ignoring all the noise and second guessing.

In order to take your digital media campaign to the World Series, it is key to determine the right data upon which to act. Billy Beane, the legendary general manager of the Oakland As, has shown the world that he is able to field a competitive ball club for less money than almost any other team year after year. Billy understands what data is really predictive of success and does not let himself get caught up in the hype around data that everyone else had been using for decades. He is also religious about applying data and does not let his "gut" or anyone else's "gut" get in the way of making consistent, data-driven decisions that pay off over time.

Related to Billy Beane's philosophy, the hottest thing in baseball management right now is Sabermetrics, defined by Bill James as "the search for objective knowledge about baseball." Just like in baseball, those that master ad effectiveness data -- and, equally important, act on it -- will win more over the long term.

If you happen to be coaching a direct response (DR) campaign, click-through-rates (CTRs) aren't a bad measure by which to evaluate a creative's performance because CTRs tend to be correlated with achieving the desired outcome. Sometimes you can even measure other action rates (e.g. registrations, request for information, acquisition, etc.) that are even more predictive of the ultimate desired outcome.

It can be a little harder to coach online brand campaigns, especially if you grew up coaching DR ball and try to stick to the same data that made you successful doing that. The two are different because in a DR campaign you are generally trying to stimulate an immediate behavior that can be measured online. For brand campaigns you win by influencing opinions and attitudes that carry over to the offline world and people's everyday lives. The difference in objectives requires a different metric.

Let's first acknowledge that it is hard to break comfortable, old habits that may have worked for you in the farm league or still work in some contexts. The problem is that if you don't break those habits and find a more relevant measure of success for brand advertising, you are going to lose -- lose the opportunity to influence people as planned, and lose your team's money.

Numerous studies have shown that CTRs are generally not relevant to brand building success and are sometimes even anti-correlated -- meaning, that if you manage your campaign based on CTR data you will likely have no effect on brand building and might even have a negative effect. One way to think of this is that just because you know the batting averages of your pitching staff, that doesn't mean that you ought to pick your pitching rotation based on how well they hit. If you do, you'll be optimizing for the wrong objective.

Applying real baseball data to that analogy, if three years ago you could have picked any five pitchers in the National League and you picked them by batting average, you would have had an estimated win-loss percentage of .47 or a record of 76 wins to 85 losses. Conversely, if you used a more relevant metric like earned run average, you would have had an estimated record of 110 to 52 and been the odds-on favorite to win the World Series.

You can achieve similar league-leading results for your brand advertising campaigns as well by optimizing against the appropriate metric. The objective of most well-conceived brand advertising campaigns is to shift attitudes and opinions in a specific way that will ultimately increase the likelihood of purchase. As a result, using a brand lift metric to directly measure how effectively a campaign is accomplishing its objective forms the basis for productive coaching decisions.

Once you know the brand lift associated with each ad unit on your team, you can optimize your campaign by intelligently deciding which ads to play, how frequently to play them, and where to play them. In fact, even very simple optimization schemes can pay big dividends.

For instance, we recently analyzed the brand lift of a half dozen online brand campaigns intended to drive purchase intent and modeled a couple of very simple optimization schemes. In one case, the impressions associated with the bottom quartile of the ad units were re-applied to the remaining ad units. In the other case, the impressions associated with bottom quartile of targets were reallocated to the remaining targets. Each optimization scheme improved brand lift by about 40% on average -- and if both optimizations are implemented, brand lift increased by an average of about 80%!

As Tommy Lasorda said, "No matter how good you are, you're going to lose one-third of your games. No matter how bad you are, you're going to win one-third of your games. It's the other third that makes the difference." Same goes for online media strategy and buying.

1 comment about "Getting To The World Series Of Digital Media Buying ".
Check to receive email when comments are posted.
  1. Brian Starks from OMD, November 10, 2010 at 6:23 p.m.

    “Numerous studies have shown that CTRs are generally not relevant to brand building success and are sometimes even anti-correlated”

    Dan, where can I find these numerous studies?

Next story loading loading..