Getting as Complex as Necessary: Attribution Modelling

There is a design principle known as KISS: Keep it simple, stupid: We should aim to keep systems, models, and solutions as simple as possible, and unnecessary complexity should be avoided. When starting a project, we should ask ourselves how complex we need to get to answer the question at hand accurately, and go no more complex than necessary.

For example, take attribution. There are reasons that data-driven attribution can be such an expensive and time-consuming product to implement. One is that it can be a very difficult problem to solve for – how much credit should each marketing channel get so that they’re all getting credit fairly and accurately? The other is that it’s an area that (hopefully) leads to high impact: How should we distribute marketing budget to increase return on investment?

These two prongs, problem difficulty and estimated impact, can help determine how complex you should get in solving for a problem.

Problem difficulty is something we don’t really have the ability to control for: If something is hard to answer, that’s the way it is and if we want to answer the question we will have to get pretty complicated.

However, we can sometimes choose how complex to get.

If a problem’s solution is likely to be low impact, then it’s better to err on the side of keeping it simple. On the other hand, if the problem’s solution is likely to lead to high impact, such as increased return on investment through an accurate attribution model, then it’s better to err on the side of getting more complex to ensure accuracy.

For attribution, this means that if you don’t have a large media budget or that if you don’t have the ability to drastically alter your media spend between channels, maybe the cost of a large and complex attribution model build is not worth it – yet. On the other hand, if you are spending a lot it is worth it to get a high quality data driven attribution model as any redistribution of budget, if done accurately as a recommendation from a good attribution model, has high potential to really impact return on investment.

To that end, if you are just starting out in web analytics, have lower ad budget, or low ability to move advertising budget around, starting out with a more basic, though faulty, model like custom rules built attribution model may suffice. Then, once ad budget is up and your analytics maturity is higher, it will be worth it to spend the time and effort on the more complex solution that is a custom built attribution model.

When stating a problem, it is important to look at how much an impact the results of solving it will have and devoting resources to it proportional to the estimated level of impact. That way, we can keep it simple where it matters and dedicate resources to solving the high impact problems that matter.

So what does getting as complicated as necessary look like for you? That depends on where you are on the path to attribution. For example, for a Google Analytics user it might look something like this:

• Starting out in attribution? Use the rules-based attribution reports available in the free version Google Analytics.
• Explored rules-based and need to get a bit deeper but have low budget? Use customization of these rules based approaches. Further, Google’s announcement on free Google Attribution can also help get you started.
• Ready to start getting serious about attribution? Google Analytics 360 will allow you to use Data Driven attribution (on top of all the other great reasons to use Google Analytics 360.)
• Explored all the previous options and ready to get more serious about attribution in particular, and have larger budget? Time for an attribution specific vendor (for example Google Attribution 360.)