There’s a famous quote from the turn of the 20th century by an American merchant named John Wanamaker, that still rings true for many marketers today:
"Half the money I spend on advertising is wasted; the trouble is I don't know which half.“
A lot has changed in the last century, but the question of which advertising efforts actually contribute to the bottom line and which don’t still troubles many modern marketers. Fortunately, multi-touch attribution can be the answer.
The problems modern marketers face are quite a bit different than the ones Mr. Wanamaker faced. Today’s digital marketing offers a data-driven approach that wasn’t available a decade ago (let alone a century). Seemingly every week, channels and platforms develop more and more sophisticated ways to understand and improve performance.
Despite these advancements, it’s difficult to understand how these channels and platforms work together. Each platform has its own agenda (i.e., they want you to spend more money with them), so they can’t provide the full picture of marketing performance.
And that full picture is important because every touchpoint a customer has with your brand can influence their decision to convert. Multi-touch attribution is how you understand the role each touchpoint plays in creating a new customer and contributing to revenue.
What is multi-touch attribution?
Multi-touch attribution is the act of determining the value of each customer touchpoint that leads to a conversion. The goal is to figure out which marketing channels or campaigns should be credited with the conversion, with the ultimate intention of allocating future spend to acquire new customers more effectively.
With multi-touch attribution, you take a conversion event, like a customer signing up for a free trial, and look at the role each touchpoint played in creating that sign-up.
For instance, in the graph above, should the Facebook ad that kicked off the customer journey get more credit for the conversion than the Google paid-search ad?
In rough mathematical terms, the way multi-touch attribution answers this question is by taking into account the cost of each touchpoint and the weight (i.e., importance) you give its stage of the customer journey.
Then, it compares that relationship to the value of the conversion it contributes to (i.e., how much that free trial contributes to revenue). The formulas for this can get complicated and verge into true data science territory, but the basic relationship of these values is important to understand as we continue.
Perhaps the best way to understand multi-touch attribution is in contrast to other leading attribution models you may be familiar with: last-touch attribution and first-touch attribution.
Last-touch attribution vs. Multi-touch attribution
Last-touch attribution credits the last touchpoint of the customer journey for the end conversion. Many analytics and ad platforms prefer this model because it’s easier for them to track, and it often credits themselves with the final conversion.
Last-touch attribution is popular because, on the surface, it makes sense. What matters most is what finally caused the conversion, right?
But if you dig a little deeper, you start to notice that it doesn’t paint the full picture. Last-touch doesn’t take into account the full customer journey, so it can sway results in favor of certain platforms, like Google Ads, and lead you to make incorrect optimization decisions.
Say you have a YouTube pre-roll ad campaign running alongside a paid search campaign. After seeing your video ad, some customers decide to search for your product on Google. They decide to click your paid search ad and then buy your product that was advertised in the video ad.
Last-touch would only tell you that people searching for your product will convert. What about the YouTube ad that kicked everything off?
A metaphor we like to use is one where you’re a world-famous actor accepting an Oscar (congrats!), and you have the opportunity to thank the many people throughout your life who helped you reach this point. With last-touch attribution, you would only thank your director for giving the chance to play an award-winning role, and you’d leave out your parents, teachers, agent, and fellow actors, each of whom helped you reach this point in various ways.
First-touch attribution vs. multi-touch attribution
First-touch attribution credits the first touchpoint of the customer journey for the end conversion. It’s preferred by some marketers because it shows how top-funnel efforts can lead to bottom-funnel conversions, but it still only shows a small part of the picture.
In the YouTube-to-paid-search example above, first-touch attribution would overemphasize the importance of the YouTube ad, making YouTube look good. But the paid search ad still played an important role in that end conversion.
As an Oscar-winning actor, if you were to use first-touch attribution in your thank-you speech, you would thank only your parents, leaving out your teachers, agent, fellow actors, and director.
Multi-touch attribution models
Overall, there are four standard models for multi-touch attribution and a fifth way of modeling that is totally customized to your business. Each model emphasizes (or weighs) different stages of the customer journey and provides differing levels of insight.
Linear multi-touch attribution model
The linear multi-touch attribution model gives all interactions the same credit in the lead's conversion. This is more sophisticated than first- and last-touch attribution, but, yet again, it doesn’t tell the whole story.
Linear multi-touch attribution doesn’t emphasize any tactic over another, but instead focuses on each touchpoint throughout the customer journey equally. This is good for gaining a strong understanding of the overall customer journey, but it doesn’t give any insight into how effective each stage is at driving a conversion.
All you can glean from this type of modeling is that these stages helped, but you don’t know to what extent. In the Oscar-speech example, everyone you mention will get equal airtime and, therefore, equal credit. You may even say this in a robot voice to be sure no single shout-out is getting special treatment.
The audience would understand who helped you earn the Oscar, but they wouldn’t know who had the most impact or who you were mentioning just to be polite.
Time-decay multi-touch attribution model
The time-decay multi-touch attribution model gives most of the conversion credit to interactions that happen closer to the conversion event. It focuses on bottom-funnel touchpoints but doesn’t totally discount upper- and mid-funnel tactics like last-touch attribution does.
Analytics expert Avinash Kaushik advocates for this model, saying that the earlier touchpoints are weighted less because "if [those] were magnificent, why did they not convert?" Time-decay modeling provides a fuller picture of the customer journey, while providing insight into which tactics more directly lead to conversions.
In this case, during your Oscar thank-you speech, you’d thank people who led to your success linearly, in chronological order. However, toward the end, each individual you mentioned would receive a longer, more deliberate recognition.
The audience would realize that those you mentioned toward the end of your speech contributed more to your success than those you mentioned earlier.
U-shape multi-touch attribution model
The U-Shape multi-touch attribution model gives most of the conversion credit to the first and last touchpoints. The focus here is on the top- and bottom-funnel touchpoints, with less emphasis on mid-funnel engagement tactics.
This model provides an understanding of which tactics are good at kicking off the customer journey and which ones are good at creating a conversion. In this case, your Oscar speech would go through everyone chronologically, but most of your time would be spent thanking your parents and your director.
During your speech, the audience would see that your parents played a big role in developing a love for acting at a young age, and that your director took a big chance on you that paid off.
W-shape multi-touch attribution model
The W-Shape multi-touch attribution model spreads most of the conversion credit across the first, middle, and last touchpoints. This model is best suited for complex cross-channel campaigns with many touchpoints.
In this case, your Oscar speech would give slight emphasis to your parents, your agent, and your director while giving small mentions to your fellow actors and your agent. Here, the audience would get a wider picture of your journey to becoming a successful actor, realizing that your parents got you started, your agent kept you motivated during hard times, and your director helped bring out your best performance.
Custom multi-touch attribution model
None of the models so far are necessarily good or bad; they just have different use cases. And if none of their use cases fits your business, you can create a custom model. This is fairly complex and requires an intimate knowledge of which models have worked well for you, which didn’t, and why (i.e., long-term experimentation).
For your Oscar speech, you’d thank everyone at varying degrees based on who you’ve determined to have the greatest impact on your career.
How to do multi-touch attribution
Multi-touch attribution is not the easiest thing to set up, because the customer journey is becoming more and more complex by the day. For example “90% of multiple device owners switch between screens to complete a task” according to Google.
In the B2B world, the complexity of the customer journey is even more pronounced. The B2B customer journey typically consists of multiple stakeholders each with multiple touches over a prolonged period.
Getting all these touchpoints across devices, channels, and platforms to talk to each other is the main hurdle in implementing multi-touch attribution. The good news is that you’re not the first to do it, so you can utilize the infrastructure and technology set up by others. And once it’s up and running, you can start finding answers to the big question: what parts of my advertising strategy are actually working?
The overall steps
1. Collect
First, you need to collect data on who is visiting your site, how they got there, and whether they convert. There are three ways you can do this that are usually used in tandem:
JavaScript, where you plug code into your web pages to understand who is interacting with your site and how. This includes calls like
UTMs, which are snippets that populate at the end of URLs. They provide data about where the customer comes from, including information like the source (e.g., “Facebook”), the campaign (e.g., “fourth-of-july”), and the creative type (e.g., “fireworks-background”). You can integrate these with the JavaScript calls to get a clearer, more complete picture of your user.
APIs: integrations you have with your CRM, advertising vendors, and others that have proprietary ways of identifying your customers.
2. Combine
Next, you need to make sense of all this data by combining it in one place. You can pipe it all into a data warehouse, like Snowflake, for secure storage. But there are other solutions, too. Gijs Nelissen, cofounder of Prezly, found webhooks useful for this stage, but it depends on your capacity and technical ability.
Here it’s also important to bear in mind the nature of your customer journey. B2B businesses, for instance, will need to cater for the multiple stakeholders involved in the buying process, as well as the months-long cycle. In this case, it’s vital to transform the data through account-based modelling, or what is sometimes called B2B attribution.
3. Visualize
Finally, you need a way to query and report on all this data to turn it into graphs and charts you can understand. A solution like Segment’s Marketing Attribution Looker Block can help you do exactly that by using Segment, Snowflake, and Looker. Doing this yourself may require some SQL knowledge, but there are many great vendors that can take this on for you.
Option 1: Use the right software
You don’t need to reinvent the wheel to do multi-touch attribution; there are many vendors that have looked at this challenge and said, “Yeah, we can do that.” We did a rundown of great attribution tools that can help you get multi-touch attribution up and running, each of which has its own strengths.
The upside of using existing software is that it’s easy to implement compared with doing it all yourself. The downside is that it costs more money, and it’s not going to be specifically tailored to your business.
Option 2: Do it yourself
If you have technical expertise, like Gijs Nelissen, it’s not impossible to build your own solution. There are guides, like this one from Rittman Analytics, that can walk you through the process.
The upside of setting up multi-touch attribution yourself is that you know this it is specifically tailored to your business. But the downside is how labor-intensive it is.
The goal is agility
Overall, multi-touch attribution gives you a more complete understanding of the customer journey, no matter how complex or fragmented it is. The result of this understanding is the ability to adapt your strategy and optimize your ad spend in tandem with larger market shifts.
You’re no longer hopping from platform to platform trying to connect the dots on your own. You have a model that shows how effective each channel is at producing conversions, and you have the agility to act on those insights. If you think that sounds pretty great, you’re not alone. Martech Today reports that companies’ investment in multi-touch attribution solutions is at an all-time high and is set to continue to increase.
Multi-touch attribution has never been more prevalent or accessible. You can even get started today with your customer data platform — feel free to give Segment a try if you’re in the market — by collecting all your data, funneling it into one place, and then sending it to data modeling and visualization software.
Advanced analytics