Are you chasing the perfect attribution model? Stop. You’re not going to find one. Here’s why and how you can still find a meaningful (and profitable) way to measure your campaign.

• It’s not possible to attribute mixed media 100 percent. Even though every digital impression can be tracked back to a browser, perfect attribution breaks down as soon as someone goes into his/her car and listens to the radio or sits on the couch and flips on the TV. There’s no connection between that impression and a browser. And at that point, accuracy is lost. 
  
  Everything becomes an assumptive model of the importance of an impression during an ad’s lifecycle. It’s also an assumption that it even was an impression. Which is why:

• No digital data is perfectly clean. People use a mix of cell phones, laptops, tablets and other tech devices on any given day. Unfortunately, there is no linkage technology that’s 100 percent accurate and can span the gap between all these different models of technology.

• Attribution cannot ignore aggregate performance. The weight, repetition and frequency of impressions build up during the course of someone’s day or week or month—and this is important for response. For instance, say a person hears two radio ads and sees three online display ads. The last TV ad that person sees for the product or service could be the response trigger that makes him/her buy, but only because of the weight of all the other impressions that came before.

So the idea that you can use last-click analysis to truly understand the impact of media is deeply flawed.

Instead, make your creative and media optimization strategies a series of small horseraces. At the end of the day, every media has a scale, i.e., there’s only so much display you can buy before you saturate the market and don’t get a return, etc. So optimize each media channel and expand into other channels to grow.

Optimize creatives first. Each creative in each media channel needs to be a horserace; you’re always after the winning horse. Have at least three horses (creatives) in your race as a minimum and continue to get rid of the slowest. Then repeat.

Keep in mind, creative performance should be independent of the relative performance of media. Don’t worry if radio ads outperform TV ads because you’re going to be in both of them in a successfully scaled campaign. Just figure out the best creative to run in each media channel, run them and then optimize the media by looking at the relative performance of each intersection you’re buying.

When you’ve gotten your radio, TV and digital buying strategies optimized, try media mixed modeling. Because you can’t attribute cleanly, you have to model it by real world performance. For instance, what happens if you double radio spend? Or halve your TV spend? You need to watch the overall performance and then build a model based on the proven knowledge of your varying media levers. And this is how you build a successful multi-channel media buying strategy.