What is multi-touch attribution? An attribution guide.
May 25, 2021
As your business grows, you’ll want to “pull the lever” on different marketing channels. At its inception, you’ll likely start your marketing efforts with a limited number of marketing tactics, like paid ads on Google Search or social media platforms like Facebook and TikTok. As your business evolves, you’ll throw more and more digital marketing techniques into your marketing mix in order to attract new customers, but how will you know who’s coming from where?
The answer? Marketing attribution
Marketing attribution is an overarching term used to describe how you can attribute each conversion—or sale—to a specific marketing touchpoint in any given campaign. At its most complicated, marketing attribution can track every single marketing touchpoint in the customer journey. At its simplest, it can track the first or last marketing touchpoint in the customer journey, and attribute 100% credit to that touchpoint. We’ll dive into that later.
What is multi-touch attribution?
Multi-touch attribution is the most complex type of marketing attribution because it takes into account every single step of the customer journey. Let’s say you’re selling smartphone accessories. A potential customer sees an ad for your product while watching a YouTube video and skips the ad.
Later that day, that same potential customer searches for “iPhone case” on Google Search, sees one of your sponsored ads as a top result, and clicks on the link but doesn’t make a purchase.
A few days later, that potential customer gets one of your marketing emails but doesn’t even open it. Finally, they see a native Facebook ad, which leads them to make a purchase, thus converting.
To which of these digital marketing tactics can we attribute the conversion? With multi-touch attribution, some of the success of the campaign would be attributed to the Google Search ad and some would be attributed to the native Facebook ad. The skipped Youtube ad and unopened marketing email should, in theory, get little to no credit for the conversion because the potential customer didn’t interact with either. In reality, however, these touchpoints might get equal credit to the conversion event depending on which attribution model is used. We’ll cover that later on.
It can be a pretty complicated process to set up multi-touch attribution, especially because it involves a reliable technology stack in order to properly credit each customer touchpoint. One of the main benefits of using this more complicated attribution model is that it eliminates the bias in attribution. It’s no longer a debate about where the conversion came from, allowing you to better target and personalize content for your audience. Multi-touch attribution also lends itself to an iterative approach to digital marketing—you can quickly tweak your ad creative and copy to test their effectiveness.
Another benefit is that it gives you a more holistic view of your digital marketing efforts. If you notice a lack of engagement with your Youtube ads and emails, you might decide to cut back on—or even eliminate—these channels as a part of your marketing mix. It also allows you to reallocate funds to your more successful marketing channels, like Google Search and Facebook, ultimately helping you to get more bang for your buck by improving your ROAS.
What are the different multi-touch attribution models?
Marketing attribution isn’t a one-size-fits-all solution, and that rings true when it comes to multi-touch attribution. There are 4 or 5 common multi-touch attribution models, depending on how you classify them: linear, time-decay, U- and W-shaped, and custom attribution models.
The linear attribution model is one of the simplest of the multi-touch attribution models. It assigns equal credit for each marketing touchpoint. So if we look at our earlier example, our potential customer had 4 marketing touchpoints: a YouTube ad, a Google Search ad, an email, and a Facebook ad.
With a linear attribution model, each marketing touchpoint would get equal credit for the conversion, meaning each touchpoint would get 25% of the credit. If there were 3 marketing touchpoints, they’d each receive 33.33%, 5 would each get 20%, and so on.
As you can imagine, one of the pros of this attribution model is that it’s easy to understand and somewhat simple, once you’ve set up your tracking and analytics. As for cons, giving each marketing touchpoint equal credit isn’t necessarily accurate.
In our example, the skipped YouTube ad and unopened email both get equal credit as the lead creation and conversion events (the Google Search and Facebook ads, respectively). While this isn’t reason enough to not consider the linear attribution model, it doesn’t leave much room for comparing the performance of each of your digital marketing channels.
The time-decay attribution model is, as the name suggests, a model that assigns more credit to a touchpoint the closer it is to the conversion. This model assigns the majority of the credit to the touchpoint that leads to a conversion, and less and less credit the further away from the conversion that a touchpoint is.
Going back to our example, the conversion event was a Facebook ad, so we’d assign the majority of credit to that. Then, in descending order, we’d assign credit to the email, the Google Search ad, and the YouTube ad.
There’s no hard and fast rule for the amount of credit given to each touchpoint, but using our example, we could give 50% credit to the Facebook ad, 30% to the email, 15% to the Google Search ad, and the remaining 5% to the YouTube ad.
One of the pros of this attribution model is that it assigns the highest value to the conversion touchpoint, which leads into one of the main cons of this model: it doesn’t give much value to awareness touchpoints. Because this model assigns most credit to the conversion event, it can be seen as discounting the value of the marketing touchpoints that created brand awareness in the first place. Depending on the length and complexity of the sales cycle at your business or in your industry, this may or may not be the model for you.
The U-shaped attribution model, or position-based attribution model, is quite straightforward, assigning 40% credit to the first and last marketing touchpoints, and dividing the remaining 20% across the rest.
Looking at our example, this model would assign 40% of credit to both the YouTube and Facebook ads. The Google Search ad and email would each get 10% of credit. If you were to have 10 marketing touchpoints in the customer journey, you would give 40% credit to touchpoints 1 and 2, and 1.25% credit for the remaining 8 touchpoints (8 divided by 10 is 1.25).
The main benefit of this model is that it gives heavier weight to the first and last marketing touchpoints, covering both awareness and conversion. This model, however, falls short with its simplistic treatment of all the in-between touchpoints.
The W-shaped attribution model is one that assigns the majority of the credit to the first marketing touchpoint and the lead creation and conversion touchpoints. These 3 touchpoints typically each receive around 30% of credit, leaving roughly 10% to split between all other touchpoints in between.
If we bring it back to our earlier example, our YouTube ad, Google Search ad, and Facebook ad would each receive 30% credit, taking us to 90%. The remaining 10% would go to our email touchpoint. This is because our YouTube ad is the first touchpoint in the customer’s journey, while the Google Search ad is the lead creation event and the Facebook ad is the conversion event.
With our example, this would be the most accurate way to attribute credit because we’ve given heavier weight to awareness, lead creation, and conversion, while giving less credit to our unopened marketing email. Although this may seem like the perfect model, it may not be appropriate for those businesses that don’t have a clear conversion point.
For example, some B2B SaaS businesses have “conversions” that don’t lead to a sale because the prospective customer doesn’t qualify for their offering. For more straightforward businesses, like most e-commerce and B2C software organizations, this model often works.
Custom attribution models
Depending on the complexity of your digital marketing efforts, the attribution models we’ve covered may or may not work for your business. The good news is that custom attribution models exist, allowing you to fully customize an attribution model based on your needs.
Custom attribution models are great in that they offer unlimited flexibility in how you’re attributing credit. They can grow and evolve as your business does, but this also comes with the question—how do we pay for or maintain this?
This leads into the downside of a custom attribution model: they can be time consuming and expensive. You can go the route of hiring—or contracting—someone to build out your attribution modelling, which will involve marketing technology know-how and ideally past experience in building custom attribution models. If you don’t want to bring someone in house, however, there are agencies who can help guide you along the way, like our partners at Common Thread Collective (CTC).
“Both for clients as well as our in-house brands, marketing efficiency rating (MER) serves as the centre of attribution gravity: total ad revenue divided by total ad spend,” explains Aaron Orendorf, VP of Marketing at CTC.
“Sometimes you’ll hear it referred to as blended ROAS. Whatever the name, two elements are paramount. First, with Apple’s iOS 14.5 privacy updates—and with Android and Chrome following suit—you need an overarching metric to measure the effectiveness of your paid efforts that operates independently from Facebook attribution and other major platforms. Second, MER doesn’t negate channel-by-channel attribution; it does, however, give you a buck-stops-here gut check across all your digital marketing plans that ensures new spend, new channels, and new creative actually results in new revenue.”
What about single-touch attribution models?
If you want to keep things simple as you develop a marketing attribution model for your business, multi-touch attribution isn’t your only option. Two popular single-touch attribution models are first- and last-touch attribution models. Luckily, single-touch models are far less complex, making them easier to understand.
This model, as its name suggests, assigns 100% credit to the first marketing touchpoint—or first “touch”. Using our earlier example, the skipped YouTube ad would receive 100% of credit for the eventual conversion through a native Facebook ad.
As you might imagine, this is a far easier approach to attribution modelling, but because of its simplicity, it’s not very accurate. This approach tends to give more credit to awareness touchpoints, as it’s rare that a user will convert after just one marketing touchpoint.
As with first-touch attribution, the last-touch attribution model’s name is self explanatory—it assigns full credit to the last marketing touchpoint. Going back to our example, we’d assign 100% credit to the conversion event—the native Facebook ad. Last-touch attribution is the most commonly used single-touch attribution model.
Similar to the first-touch model, last-touch attribution is easier to set up and track, but is overly simplified, not assigning credit to any of the awareness or lead creation touchpoints. One main issue of this model is that if a customer becomes aware of your brand via a previous digital marketing campaign and goes directly to your website to convert, that “direct” visit will receive 100% of credit for the conversion. This becomes an issue because the digital marketing campaign gets no credit for the conversion despite doing its part in creating awareness.
So, how can we avoid this?
Last non-direct touch attribution
The last non-direct touch model is essentially an answer to the issue with the last-touch model. It assigns 100% credit to the last non-direct touchpoint. While this wouldn’t change anything in our example, it would impact cases where a customer converts after directly landing on your website.
To clarify, a direct visit includes when you type a URL directly into the search bar of your browser or reach a website through an organic search. It could also happen if a friend sends you a link to a website which you then visit. To put it simply, it’s any visit to a website that isn’t facilitated by paid marketing efforts.
So, let’s say a customer sees a Google Search ad, a YouTube ad, and a sponsored influencer post on Instagram for your product. They click on the Google Search ad and ignore the YouTube ad and sponsored Instagram post. Then, a week later they Google your company, land on your page without clicking on a paid ad, and make a purchase. Using the last non-direct touch model, we’d give 100% of credit to the sponsored Instagram post.
The issue with this model? In this example, the Google Search ad was the only marketing touchpoint that the user interacted with, but it gets zero credit. While it might be perfect for avoiding wasting attribution credit on direct website visits, it still has its flaws.
Which marketing attribution model is right for me?
While it’s safe to say that multi-touch attribution models are more accurate and all-encompassing than their single-touch counterparts, they’re not without their flaws. If you’re just starting to track attribution for your business, don’t fret over getting it right the first time. It may make sense to start with a more rudimentary attribution model until your marketing efforts become more complex.
Remember that marketing attribution is something that even some of the biggest companies struggle with. The fact that you’re already considering your attribution strategy is a great start. Take inventory of your current marketing strategy and consider looking at some of your customer data and comparing it against the different attribution models. There’s no right or wrong approach—but starting to attribute credit to your efforts is always a great first step.
The Complete Multi-Touch Attribution Guide
Know which marketing channels drive revenueGet the Guide
The Clearco Newsletter
Sign up to get fresh content straight to your inbox!