In the recent Econsultancy Report Marketing Attribution: Valuing the Customer Journey, Bill Kee, Product Manager for Google Analytics points out that although only 14% of businesses think that last-click attribution is effective, more than half of those who are doing conversion attribution are still using last-click attribution*. They're doing it wrong, and they know they're doing it wrong.
However, getting it wrong isn't necessarily a bad thing. It all depends on how wrong you're getting it.
Although last-click attribution may be simplistic, may hugely overweight the contributions of some channels, and undervalue others, and may be misleading when it comes to analysing your marketing mix, it still does what attribution is meant to do. It provides you with information from which you can take action.
Information that doesn't lead to action is useless. It's useless and it's costing your business money. A theory that's wrong but testable, that's skewed but open to refinement is preferable to a perfect one that you can never implement because the tools are never there.
As technologies develop, so does the desire to fully understand your customers. We want to know every touchpoint on the customer journey: we want to see their thought-processes and their buying-processes.
It's an understandable desire, but it's one that can lead to paralysis if you stick to it slavishly. Your model will be incomplete. No matter how it is weighted, no matter how nuanced the weighting you give to each channel at each point in the funnel and after it, your customers are still more complicated than we can give them credit for.
We can't understand every interaction in all its depth and complexity. Certain parts of certain customer journeys will always remain mysterious.
However, what we can do is provide powerful tools for understanding the basic thrusts and trends that affect all of your customers. We can show the ways in which channels interact, we can divine the ways in which on- and offline marketing complement each other, we can certainly tell you a lot more about what is working and what isn't than last-click ever will.
Approach it like Michaelangelo did. He said “Every block of stone has a statue inside it and it is the task of the sculptor to find it.” You've got a big block of data, and your attribution weighting should help reveal what is inside.
But you have to start somewhere. You have to knock out a rough shape before you can start finding individual features in the stone. To begin with, keep your attribution modelling simple. Knock the corners off the block.
You will be able to think of exceptions for your weighting models. Don't give them too much credence. Don't let them stop you getting your attribution working, and then testing and refining it.
Keep it simple. Keep it testable. Keep refining. Keep working at it.
And remember: if it doesn't lead to action, it's useless.
The Econsultancy report highlights a lot of important trends, and it is great that the business world is realising the power of correct attribution. It will become one of our most powerful marketing tools in the next few years. But we have to walk before we can run.
It's great to see the ambition marketers have for their conversion attribution. However, more than a quarter of the companies Econsultancy spoke to have only started doing conversion attribution in the last 6 months, and 83% started in the last two years. Again, we have to be able to walk before we can run.
Your customers are unique individuals, all of whom has different experiences with all of your marketing channels. Every product you make is different, every project is different. Seeing all of the nuances of every customer journey is, at the moment, impossible.
The perfect weighting model does not exist. Don't let your search for it prevent you making easy wins.
You're getting attribution wrong because you have to. The impossibility of seeing inside another human's mind in all its working, the limitations of today's technology, customers' reluctance to be tracked all day every day: all of these combine to mean that your model is incomplete.
Your model has holes, inconsistencies, customers for whom it simply isn't relevant. But rather than denying it we should embrace it. Embrace it, test it, refine it, and have another go. Because it's an ongoing process, and the more you put into it, the more you'll get out.
Embrace the wrong! See each problem with your model as an opportunity to make it better.
You're getting attribution wrong because you have to. And next time you're going to do better.
But you'll still be wrong.
And that's great news!
*Econsultany Report, Marketing Attribution: Valuing The Customer Journey, pp. 1
