Like so many terms in digital today, “attribution” has been turned into just another buzzword. That’s a shame. Attribution modeling isn’t a perfect science, but it’s a fundamental and important concept in digital marketing. And, when talked about straightforwardly, the concept is not all that hard to understand.
Creating an attribution model is simply the process of figuring out which ads in a campaign are providing the most bang for your buck. In slightly more technical terms, it’s a way to understand how each dollar invested in a multi-channel campaign contributed to each conversion.
Why are some marketers so focused on attribution models? Because when you know which components of a campaign are the most cost-effective, you can invest more in those channels and make smarter decisions as you move ahead to your next spend.
In a perfect world, every marketer would be doing attribution modeling for every campaign. But, as you might have noticed by now, the world is pretty far from perfect. Surprisingly, some agencies still show little interest in attribution models. In fact, some agencies rely on their lack of a model as an excuse for bad planning. “We can’t measure the campaign fully so we aren’t even going to try,” the argument goes.
When agencies do use a good attribution model, they’re better able to say how one or two different channels in a campaign influence one another. In other words, if you’re investing in retargeting and you’re using a good attribution model, you’ll have a clear sense of how the investment is paying off. The model won’t necessarily give you one final and definitive answer, but it will generate lots of very useful data that will allow you to appreciate how all of the different parts of your campaign are working together and contributing (or not contributing) to success.
Attribution models, to be clear, can’t solve every challenge that marketers face or give you one simple answer about what to do next. Nor can an attribution model tell you exactly how every touch point in the campaign is working vis-à-vis the others. It should be considered a macro rather than a micro-analysis tool. You still need smart humans to look at the data and make decisions.
So, how do you determine which model is right for you? First click? Last click? Weighted attribution? Equal attribution? Cascading attribution? Different campaigns will require different approaches, though “last click” will almost always be the wrong approach, as it undermines the very point of attribution modeling: determining how all of the different channels contribute to a conversion.
If you’re new to attribution, remember: The output of a given model can look very different for each organization. I personally like a model developed by Forrester some years back that sets all of the trackable touch points against the time since the given touch point occurred. But it’s not necessarily the right approach for every campaign.
The one thing you don’t want to do is ignore attribution altogether or dismiss it as just another bit of jargon. An attribution model might not give you one final answer, but it can get you a lot closer to the truth.