Choosing a marketing attribution model is like getting a new pair of glasses: The right ones can bring your whole world into focus, letting you see things much more clearly. Get it wrong, and everything’s going to look a little blurry.
Unfortunately, there are so many different attribution models available that it can be tricky to select a methodology that makes sense for your organization. In this post, we’ll highlight the most common attribution models, their pros and cons, and when it makes sense to employ them.
What Is a Marketing Attribution Model?
Put simply, a marketing attribution model is a set of rules or a process used to determine which marketing touchpoints helped generate a conversion and, if so, how much credit they deserve.
Attribution has never been more important than it is today. As marketing campaigns have grown more sophisticated, encompassing a wider number of platforms and channels, it’s essential for marketers to figure out which ones are creating the desired results and which ones are just wasting budget.
Attribution models fall into one of three categories: single-touch, multi-touch and data-driven.
Single-Touch Attribution Models
A single-touch model looks at a single marketing touchpoint to award all the credit for a conversion. These types of models are easy to understand and implement, but they have a higher risk of inaccuracy because they ignore all other touchpoints. In the right situation, though, a single-touch approach can be useful.
First touch (or first interaction) gives all the credit for a conversion to — you guessed it! — the first marketing message that a contact encountered. A first-touch model can be useful for marketers who are focused on awareness, demand generation and the top of the funnel. It shows you what’s bringing new people into your company’s orbit.
Last touch (or last interaction) uses the last marketing touchpoint before conversion to award credit for a conversion. If you need to know what is causing contacts to convert, then a last-touch model can be pretty helpful. It’s also handy for products with a very short sales cycle where the buyer only has a few touchpoints.
Last non-direct click is like last-touch attribution, but with one key difference: Credit is given to the last touchpoint or channel not counting times when a website visitor typed your URL into the browser or used a bookmark to visit the site. Last non-direct click helps keep the focus on the touchpoints or channels that led visitors to your site.
Multi-Touch Attribution Models
Instead of awarding all credit to a single message or channel, multi-touch attribution (aka fractional attribution) recognizes that it takes more than one interaction to generate a conversion. So, like the name implies, multi-touch divides the credit among multiple touchpoints in the buyer’s journey.
This method is more sophisticated than single-touch attribution, but it’s still relatively simple to implement because multi-touch follows clearly understood rules.
Linear attribution gives each touchpoint in a conversion path equal credit for that conversion — first, last and all the touches in between. Linear attribution fixes a big shortcoming in single-touch attribution, but there are two caveats. One, linear attribution assumes all touchpoints are created equal, but what if one specific touchpoint, like attending a webinar, was the tipping point? Shouldn’t it receive more credit than other touchpoints? And two, if the data from every touchpoint isn’t included, a linear model might overlook valuable engagements with offline media like direct mail or TV.
U-shaped attribution (also called position-based attribution) awards most of the credit for conversion to the first and last touchpoints. In Google Analytics, each touch receives 40 percent, while any other touchpoints share the 20 percent that’s left over.
W-shaped attribution is a lot like U-shaped attribution, except there are three main touchpoints: the first touch; a mid-funnel touch where a contact develops into a lead or a marketing qualified lead; and a late-stage touch where that lead grows into an opportunity. Those three touchpoints each get 30 percent of the conversion credit. The remaining 10 percent is shared by any other touches.
Time-decay attribution works on the principle that more recent touchpoints were more influential than ones earlier in the customer’s journey. So those late-stage touches receive more credit for the conversion. Google Analytics employs a “half-life” of seven days in its standard time-decay model. Under those rules, a touchpoint from seven days before the conversion only gets half the credit of a touchpoint that happened on the day of conversion.
Custom attribution is when you create your own rules or modify an existing model. Maybe you know from past experience — or maybe you just have a gut feeling — that booth visits during a conference are more valuable than other kinds of touchpoints. You could create an attribution model that gives booth visits 1.5 times more credit than other kinds of touchpoints. Like other rule-based models, a custom approach could be running on incorrect assumptions, but it gives you a starting point for testing your hypotheses’ validity against other models.
Data-driven attribution (also called algorithmic attribution) looks at all converting and non-converting paths in a customer journey to see which touchpoints are more likely to lead to a conversion and determine how much credit they should receive. Unlike other multi-touch models, it avoids giving credit to non-converting touchpoints that aren’t contributing to conversion.
Of all the attribution models, a data-driven approach is the most complex and the most difficult to implement.
For example, Google’s data-driven attribution option is available only to users of its enterprise-class Google Analytics 360 offering. To use this option on Google’s platform, the account must have generated a certain number of clicks and conversions — the model needs a critical mass of data to work.
If you develop a data-driven model outside of Google Analytics, you’ll need the expertise of a data scientist to create, monitor and refine your model.
Tips for Using Marketing Attribution Models
It’s usually better to use simpler single-touch and multi-touch models when you’re getting started with marketing attribution. For one, they’re easier to implement and understand. You may find that these models do a good job of awarding credit, especially if your customers only have a few touchpoints before converting. As a result, you won’t need to invest in more complex modeling.
It’s also a good idea to consult multiple models. A first-touch model will tell you things about your marketing’s impact that a last-touch model couldn’t, and vice versa. By using both of them, you get a more complete view of your funnel.
Remember to choose models that make sense for your particular role, too — a lead gen marketer who wants to know when contacts become leads might need a W-shaped model.
And finally, look at attribution modeling and reporting as a long-term strategy, not a short-term project. As your campaigns evolve, or as your level of insight becomes greater, you might need to use a more advanced approach like custom or data-driven models.
Ultimately, the more clearly you can see what’s really happening in your marketing, you’ll be able to make better choices about marketing strategy faster.
Multi-Channel Attribution Made Simple
Alight’s end-to-end analytics solutions feature multi-channel attribution modeling, built right into Power BI or Tableau. Request a free 30-day solution evaluation using your data in our dashboards!