With a few notable exceptions, the best predictor of the future is the past. That may not help with events like volcano eruptions or, perhaps, when wishful thinking blinds us, but when it comes to marketing results, the more you know about what’s worked in the past, the better you can predict the future.

The challenge is to create a rational media plan that does more than copy last year’s program, but doesn’t test new merely for the sake of novelty.

Start your analysis by breaking down the full marketing budget and results by discrete channel. An overview like this lets you know whether TV is working better or worse than email, and whether direct mail is working better or worse than online display. This also offers the opportunity to refine the results within each medium – for example, analyzing DRTV response by geography, daypart, etc.

In effect, you now have benchmarks by which you can measure your next campaign’s results. And now the whole media budget can be allocated by real-world intelligence.

In other words, start from scratch. Don’t spend $10 million on DRTV just because that’s what you spent last year. Make each allocation based on what you know from the results. Rethink everything.

Think about the strengths of each medium you use and how those strengths might complement each other. For example, we’ve found that DRTV is better at seeding the market, while direct mail and email are better at harvesting response. Those media used in tandem are more effective than any one used on its own.

Now is also the time to think about new media. Is there something you should be trying? If you need better buzz in the market, consider putting some money into social media. Is your target audience starting to look for you (or your competitors) in mobile? Put a portion of your budget into testing. It’s the best way to learn.

Then, when you run your campaign, it’s possible to optimize on the fly. For example, as calls from a DRTV campaign come into the call center, you may see that a particular geography is responding better than another and you can flow more dollars into that area. Or, if you have staggered the sends for an email campaign, you can test and adjust the creative and send the best-performing subject lines or offers.

A test-and-learn, bottom-up approach to media selection is almost always appropriate and will pay off immediately and far into the future.