Blog What is multi-touch attribution?

Multi-touch attribution has never been more in-demand, with increasing demand for in-depth data from all corners of the mobile marketing ecosystem.

Marketers are becoming aware of the importance of touch points beyond the last, and want to better understand what kind of impact all of their campaigns have on a single conversion.

But in order to understand the multi-touch attribution definition, you also need to understand the first-touch attribution model and how it differs. In this article you will discover more about how multi-touch attribution works, and how it compares to other models. Scroll on to learn more.

Defining common attribution modeling versus multi-touch attribution

These days, most businesses utilize a few common types of modeling, or their own bespoke model if they’re working with more complicated campaigns. The two most common models are first touch and last touch attribution. Let’s start off by looking at how they work, to help contextualize multi-touch when we come to define it.

First touch attribution model vs. last touch attribution explained

In first touch attribution, the first impression that leads to a conversion receives the entire conversion credit.

Last-touch attribution is used by many attribution providers (including Adjust) as the default means of tracking conversions. Here, the last campaign (or impression) that leads to an install gets all of the conversion credit. Therefore, even if a user views multiple different ads — such as a Google campaign followed by an Apple Search Ad, and finally a Facebook post — Facebook would receive 100% of the credit.

Multi-touch attribution defined — and what makes it different

By contrast, multi-touch attribution takes all touch points into account when weighing credit for an install. For instance, if a user installs an app after seeing three advertisements from three different sources (such as Google Search, Facebook and Apple Search Ads), those publishers would all be given some credit for the install.

How the credit is attributed depends on the model the advertiser works with, and it’s something we’ll discuss later in the article.

Exploring the benefits of multi-touch attribution

Now that you understand the multi-touch attribution definition, let’s explore a little deeper. Multi-touch attribution gives mobile marketers a much better understanding of exactly what led to an app install, by providing more data on the entire chain that led to a conversion.

Marketers know that it’s not always just one touch that leads to an install, but instead that users will often view many ads before they hit “download.” Knowing which ads appear to users, and which help lead to a successful conversion (and which don’t), helps inform decisions, and also overall performance.

Yury Bolotkin, UA Manager at Wooga, knows this all too well: “Having the full picture in hand helps improve marketing strategies and allocate budgets towards legitimate sources. In fact, there’s often too much focus on the last touch, which means that marketers have to completely ignore all other interactions the user has before turning into a customer.”

“From my perspective, multi-touch attribution is an essential piece in the understanding of the actual journey of a user on the way to the conversion point.”

Paula Neves, CMO at Gazeus Games, tells us that, “Multi-touch attribution helps marketers understand what their true CPI, revenue and — therefore — ROI and ROAS are for their campaigns.”

Furthermore, with multi-touch, “you can find out that certain partners aren't delivering as many installs as you'd think, while other partners are delivering more than you’d expect. This can change the game completely in terms of your choice of investment, and also which partners to consider at each stage of your marketing mix."

How multi-touch attribution works

Understanding how multi-touch attribution works can be challenging, in part because there isn’t just one type: in fact, it’s an umbrella term for a variety of different methods, or models.

This is because marketers have to weigh credit differently depending on vertical, how they monetize, and so on. As such, the split is rarely even, and so models have been developed to attribute credit depending on the needs of the business.

Let’s look at some models individually. For simplicity, we’ll focus on the following five:

  • Linear
  • Time decay
  • U-shaped
  • W-shaped
  • Custom

The linear model

Linear is the simplest way of applying multi-touch attribution. This model gives all interactions the exact same credit for the conversion. There is no difference between the assigned weights, just even totals calculated by dividing the whole value by the total number of touch points along the path to purchase.

Why would a marketer choose this model?

For simplicity’s sake, the linear attribution model is useful for those who don’t necessarily view one touchpoint as any more important than any other.

However, some may consider this model too simplistic, for a few reasons. One, in particular, is this: linear multi-touch attribution doesn’t take into account the importance of key touch points like first and last touch, which might be viewed as having more importance than the many impressions in the middle of the funnel.

The time decay model

This model gives more conversion credit to interactions that happen closer to the conversion event.

Why would a marketer choose this model?

If a conversion takes some time, early touchpoints may not be the most important. Once the prospect is in the pipeline, concern is then given to nurturing them, as inciting a down-funnel conversion would likely take longer to complete.

The disadvantage to some may be that this model covers the entire cycle, where some marketers would prefer a model that gives more credit to fewer key milestones. One of these types is the U-shaped multi-touch attribution model.

The U-shaped model

This type focuses on two key milestones, while also taking consideration of the middle touchpoints between them.

In this scenario, the first touch is given 40% of all credit, while the last touch is also given 40%. The remaining 20% of the credit is divided across the middle touchpoints that occur between those two key stages.

Paula points out that models such as these “give you more insight into which partners are better at bringing awareness (at the top of the funnel), and which are good at bringing installs (at the bottom of the funnel). This level of information can lead to a huge gain of efficiency when putting together one's UA strategy.”

Why would a marketer choose this model?

The U-shaped model is preferred by marketers who see the entry and exit points as the most influential.

Paula’s one of them, and gave us some real insight into the U-shaped model: “Back in my e-commerce days, a U shape was good enough. It’s not that you completely disregard the middle of the path. Instead, you give the source that first makes the user aware of your app and the source that generates the install, more weight than the parts in between.”

However, for some marketers the opposite might be true: if you view that middle touch points are just as influential as first and last touch, then you’ll be providing less credit to them than you’d like. The W-shaped model addresses this issue.

The W-shaped model

W-shaped multi-touch attribution modeling is similar to U-shaped, but the model covers additional key touch points evenly, distributing the rest of the credit to between-stage moments.

The first, last, and another milestones such as lead-create touch, or opportunity-create touch, receive 90% of the credit (30% each), while the remaining middle touch points receive the remaining 10%.

Why would a marketer choose this model?

For some longer cycles, it makes sense to weigh in an extra keystone and even out the credit if it’s significant to the overall model. There are even possibilities to change the credit levels, for example with first touch or last touch (weighing 25% and 35% respectively).

However, Paula succinctly pointed out the issue with the W shape: “It's not that the middle isn't important. But, as I’ve experienced, the more complicated the attribution model, the more errors you can generate. It's about finding a model that both suits your business and needs and isn't too overcomplicated.”

The model isn’t necessarily an issue of balance, but of potential mistakes on what’s reported.

We’ll skip over the “full path” model, as that model versus the W-shape is largely similar, moving on to explain the custom model.

The custom model

If a business has already created extensive touchpoint tracking, then they will also have the capability (and, likely, the need) to adjust the weighting of their attribution model to fit their individual reporting needs.

As such, in custom models, the advertiser will (as above) set the various weighting of each touchpoint, as seen below.

Why would a marketer choose this model?

Because no two companies are the same, whether it be business model, vertical, or monetization strategy, it stands to reason that everyone would be better served by forming their own attribution model.

However, the model gets complicated, and the likelihood of errors does too. Furthermore, such models need to be optimized extensively as new data comes in. Essentially, such a model would require a lot of hard work, time and money to get right, something not many app businesses can spare.

Picking your model requires the right balance between the needs of the business, and what sort of resources can be dedicated towards it, and is something we’ll address in the next section.

6 best practices for B2B multi-touch attribution

If you’re getting started with multi-touch attribution, there are some basic requirements that you have to think about. From data issues to project management, here’s a quick checklist of things to consider before you get started.

Change the way you think

Multi-touch attribution is quite a big step up from traditional performance data, and so it’s important to manage expectations.

You’ll need to revise how you interpret your results, as the impact that one channel has on another is often underestimated, particularly as the results have often been looked at in isolation to each other.

Therefore, it’s critical to make sure that your team knows that the parts are no longer separate. To do that, you need to teach them.

Educate key stakeholders

An early step to making sure implementation is a success is ensuring that all stakeholders understand what multi-touch attribution is, what it will mean to the business, why it’s being implemented, and what a good outcome will look like.

Create new, shared KPIs

Speaking of success, it’s critical to apply a few KPIs to ensure you’re heading towards your goal. Since multi-touch is so broad, it will likely tie stakeholder targets together, and so your team needs to agree on a shared set of KPIs. This will allow an integrated view of performance.

Plan how to handle your data

Your data will now span across multiple campaigns and touch points, and will require a lot of extra handling when you’re getting set up.

Creating a blueprint that lists the types of data, the sources, methods of collection, format, and so on, will help get your head around the task, and help mitigate some early mistakes.

Start small

It’s important to get up to speed with repeated testing of smaller campaigns. You shouldn’t rely on multi-touch from the start, but slowly build campaigns, test, and make sure implementation is right.

If everything works out, you’ll also have the perfect internal case study to help convert more skeptical team members.


The essence of multi-touch is an understanding of many touch points driving performance together. That means there’s much more scope to tweak between each and improve your conversion funnel on a microscopic level.

Your team will need to look at ways to improve the effectiveness of different elements of your campaigns, from ad sizes, to copy and creative executions on a granular level that wasn’t perhaps quite so necessary as before. However, small tweaks can have a big impact.

For more on how multi-touch attribution works, and the potential pitfalls of the approach, take a look here for what Paul H. Müller, our CTO, has to say.

Want to get the latest from Adjust?