Attribution Part 1: The Loch Ness Monster of the marketing world
Summary
Attribution is an essential part of a measurement framework where there is extensive research before purchase, with a long path length and time delay. Attribution is often pitched as a ‘silver’ bullet to improving marketing ROI but any attribution insights aren’t generally actions.
Whilst Media Mix Model provides a higher level overview of channel performance and allows you to forecast the impact of budgetary changes in overall , attribution is a more regular solution, used in daily and weekly optimisation of channels and campaigns.
In our three part series, we explore why attribution is so important, what action can be taken based on attribution insight, and the People, Process and Platform challenges preventing you from taking action on attribution today.
The attribution monster
Like a tale that’s lived on from the olden days, thus is the fable of the ‘attribution monster’. Many bravely tell of the time they got close to the ever-elusive legend: from your manager, to your team, from the publishers you work with to the articles that you read. It is somewhere out there, and if you hunt hard enough, on a clear day you might sometimes catch a shimmer in the distance.
One day, you decide to take the plunge and ask a deep-lake diving expert (your data consultant) to perform an in-depth analysis of your media attribution. A few weeks, and many thousands of dollars later, they come back with the big insight — your attribution model undervalues upper funnel channels. You think about the meaning of this, and how you can improve the catch (media return on investment, that is). You think some more. But like ol’ Lochie, just as soon as you think you’re close to the many spoils of the attribution monster, it disappears once again. Life goes on, and the monster is no closer to being caught.
In a world of aggressive targets, conflicting priorities and outdated platforms, it’s no wonder you have zero idea where to turn next. The promise of attribution as the golden ticket to making your organisation more ‘data-driven’ thus falls flat, lost along with your search for the legendary beast. You go on as you were, disheartened… but not beaten.
Taking the plunge
In this post I talk through the big, often overwhelming topic of attribution. What is it? Why do we even care? And how can we overcome the People, Process and Platforms challenges that are preventing us from catching the beast?
If you have spent more than five minutes working in a marketing role, then you would have read, in some form, whether it be a blog post, an article, or a research paper, about the increasing complexity of the customer journey and the need to move away from last click attribution. Often, one of the key actions from these articles is to “…implement data-driven attribution”.
Oh, if only it were so easy!
However, after this post you will see that maybe attribution isn’t the scary monster that myth would have you believe, and you should have the right tools to ‘take the plunge’.
Should I even worry about the attribution monster?
The longer the purchase cycle, the more need there is to build an attribution model. An impulse purchase, with a one day time to purchase (such as for groceries) will have fewer stages to purchase and so arguably, have less need for an attribution model as a priority. Any research-driven purchases (such as buying a new car) with longer path length are the perfect candidates for thinking about attribution. We can often get an idea about where we sit on this spectrum within our Analytics platform by looking at an attribution report. As an example, within Google Analytics, you can look at the ‘Path Length’ report and note the most common path lengths. If most conversions occur with a path length of less than 2–3, then your attention might be better spent elsewhere. However, where more than 50% of your revenue is coming from longer paths, then this is a hunt worth pursuing.
Figure 1: Path Length report in Google Analytics
How does attribution fit into my overall measurement strategy?
The goal of any measurement strategy is, ‘To understand the impact of media channels on your core KPI’. Or to put it simply: ‘To be less wrong with your measurement.’ A combination of media mis modeling, geo testing and attribution can help us be ‘less wrong’ when it comes to measuring our marketing performance.
How do they work together?
Let’s start with the media mix models, run a few times a year by specialised agencies. This statistical model gives us a higher level overview of how the various internal factors (e.g. channel spend, channel impressions) and external factors (such as the impact of COVID lockdown) influence the overall objective. They are used to forecast the impact of changing a variable (such as spending more on TV) assuming that everything else stays the same, and can be used to define the optimal allocation to drive maximum revenue in the annual budget setting process.
Next, we come to the world of geo-testing. These measure the incremental (read causal) impact of a media change on overall sales based on the difference between the ‘test’ region and the ‘control’ region. Does the extra $100k you spend on brand advertising drive additional sales or would the users have bought anyway? Geo tests can be run multiple times a year, with platforms such as Google Ads and Facebook Advertising making it even easier to run Geo Experiments through their platforms.
Finally, we move down to attribution. To understand attribution, assume that a salesperson has recently closed a big deal for your business. Hopefully, you do not give all the credit for the final sale to the salesperson only, but to all the people involved, including the product, content, analytics teams? Attribution allows us to understand how the various channels (and campaigns) interact on a path to purchase and drive the overall conversion. Attribution is easier to implement and scale so can be run on a more regular basis than the other options
Figure 2: MMM vs geo-test vs attribution
As you can see, attribution is important, and catching the attribution monster is an essential part to any measurement strategy.
Over the next few articles we will share more about the action that you can take once you catch the attribution monster (because, after all, data without any action is a lot of money wasted). We will discuss how the recent changes to the faithful cookie will have an impact on our attribution measurement. Finally we will discuss the most common People, Process and Platforms challenges to implementing attribution, gleaned through our experience working with some of the largest retailers within Australia.
As ever, do reach out if you wanted a free consultation on your measurement challenges.
