There's No Such Thing As Bad Data

By definition, programmatic buying allows advertisers to hit their desired audience with every impression while providing instant feedback as well as a strong foundation to make optimizations. Conceptually, it sounds as if every campaign should drive results through the roof. However, the reality of advertising is that campaigns sometimes miss the mark. But (in my best John Madden voice), that’s why you play the game (and don’t run ad campaigns on paper)!

While missed KPIs are never a goal, they should be far from a nightmare. It’s surprising that so many marketers throw the baby out with the bathwater when talking about why a campaign failed, mislabeling misses as “bad data.” Good, bad, or ugly, data received from a campaign is always a source of fuel that helps power optimizations for live campaigns and valuable knowledge for future initiatives.

So why does data get thrown out? Advertisers and agencies often operate on the mindset that failure is something to be avoided like the plague, rather than something that can be learned from. I’m not here to say that every caterpillar is a butterfly, but too many marketers pull the plug on poor performing programmatic campaigns without really giving them a chance to mature. Programmatic buying allows marketers to test campaigns, learn, and adjust. Marketers may even learn that their notions of their target audience are false.

Let’s look at an example using an American motorcycle brand. If you asked 20 random people on the street what a motorcycle owner looked like, they would probably think of a crowd of tatted-up, rough and tumble dudes who spend a lot of time maintaining their bikes. Think Jesse James or the types of guys you see on American choppers. A motorcycle brand may task its agency to build a campaign targeting individuals who fit this standard description, building profiles around online users who display affinity for stereotypically “macho” things.

Now, what if that campaign misses the KPIs, even though the desired audience is hit? Maybe traffic to the brand’s website rises slightly, while interactions and clicks are half of what was expected.

Most marketers would probably write this off as a failure. Is the Web just not an effective place for a brand like this to advertise?

Instead, marketers should try and see what insights they can glean from their “failed” campaign. Who actually did visit the site as a result of the ads? Who was clicking and engaging?

The brand may find that the specific audiences that responded to buying a new chopper are also concerned with the merits of vinyl versus aluminum siding for their garage, or paying for their kid’s college education. Perhaps the preconceived notion about the audience is incorrect. They shouldn’t be looking for hardcore bikers, but rather for men suffering from mid-life crises looking at a motorcycle as a way to rekindle their coolness. You know, the whole “can’t judge a book by its cover” theory. The next round of campaigns can optimize toward this mid-life crisis audience, and the brand can expect to see the original desired results.

How do we achieve the best outcome possible from a less-than-stellar campaign?  Without digging too deep, there are so many small things to look for in every campaign, whether the KPIs were low or off the charts:

Creative & messaging
Is there a noticeable difference in how audiences respond to different creative?  Consider how to change that on campaigns going forward, so that the pieces drawing a response are used more frequently.

Time of day
Is there a specific window during the day that performance is particularly strong? Analyzing something like time of day can tell us a lot about a brand. Is this something that people engage with at work? On weekends?

Sequence of messaging

Marketers using dynamic creative to build custom units may notice that one sequence or product set outperformed all others. Use that as building block going forward.

There’s plenty to glean from these three items alone, and every subsequent campaign results in more learning.

We can all agree that wasting money on poor-performing campaigns is foolish, and eventually marketers have to make tough business decisions. But if brands and agencies were willing to learn from what they consider “bad data,” then what looks like a loss turns into a new opportunity and a better marketing strategy. Remember: Chickens come from hatched eggs, not smashed ones.
Recommend (5)
2 comments about "There's No Such Thing As Bad Data".
  1. Paula Lynn from Who Else Unlimited , December 31, 2013 at 12:58 p.m.
    Garbage in, garbage out.
  2. Pete Austin from Triggered Messaging , January 3, 2014 at 1:19 p.m.
    Very good article. A corollary is that, if you could fully optimize your marketing and therefore only collect "good" data, you would lose the ability to keep learning when works and what doesn't. So some bad data is good and 100% good data is bad.