What is Multi-Touch Attribution (MTA)?

• Multi-touch attribution (MTA) assigns fractional credit to multiple marketing touchpoints along a customer’s path to conversion using user-level tracking data.
• MTA cannot measure non-addressable channels (such as print, radio, or linear TV) and is increasingly limited by privacy restrictions, walled gardens, and incomplete user-level data.
• Common models include rules-based (weighted or even), algorithmic/data-driven, first-touch, last-touch, and time-decay approaches.
• MTA estimates propensity to convert rather than true demand, so it is not reliable for forecasting or strategic budget decisions.
• Due to these limitations, many marketers have shifted toward incrementality testing and media mix modeling (often used together in a triangulated approach) as more robust alternatives.

Multi-Touch Attribution (MTA) is a method used in marketing that involves collecting user-level data for multiple media and conversion events to determine the impact and assign credit to each media event along a customer’s journey, from the initial engagement through to conversion.

Because MTA requires tracking and connecting all media at the user level, it does not account for non-addressable media such as print, radio, and traditional (linear) TV, which cannot be tracked to individuals.

MTA provides a holistic view by tracking each interaction across channels, enabling marketers to understand the combined impact of each touchpoint.  Implementing an effective MTA model is a complex and difficult process, but it can deliver effective results if the media mix is largely made up of addressable (trackable) media. 

User-level tracking is typically performed using Google Analytics, tools from data tracking vendors, or one of the many open-source tracking pixels available. Theoretically, the tracking data is then used to create detailed user click paths that map out the media touchpoints a customer encountered leading up to a conversion. 

Capturing impression-level data and piping it into attribution models can be a challenge because more and more publishers and platforms have become Walled Gardens and refuse to share user data. Impression views are an important part of the overall picture and this lack of visibility has been the biggest detractor to using MTA. Access to this critical data is still restricted even though Google cancelled its plan to retire third-party cookies in Chrome and dialed down the Privacy Sandbox initiative. Firefox and Safari continue to block third-party cookies, Apple’s App Tracking Transparency limits mobile identifiers, and walled gardens (like Google and Meta) are still refusing to share impression-level data. The user-level visibility that MTA relies on hasn’t returned and is likely not coming back. 

A wholesome attribution model assigns all the credit to the first touch or the last touch. A fractional attribution model spreads credit across all marketing touchpoints in the consumer journey leading to a conversion event.

The most common MTA models are:

• Rules-Based Weighted Distribution: Assigns weight percentages to first-touch and last-touch, then the third percentage to all the touchpoints in between. (Example: 60% first-touch, 30% last-touch, 10% other.) This model requires diligence, ongoing review, and frequent revisions to the weights to keep it close to a version of the truth.

• Algorithmic: Uses machine learning to objectively determine the impact of marketing events along the path to conversion. Building this type of model is extremely time-consuming and labor-intensive. It is also fraught with data breakage and lack of impression visibility in many major marketing channels.

• Rules-Based Even Distribution: Divides credit up equally across all touchpoints in the path to a conversion. While much simpler to calculate, this model is less common and less accurate than weighted or algorithmic models.
• Last-Touch Attribution Model: In the last-touch attribution model, the last touchpoint receives 100% of the credit for the sales conversion. 

• First-Touch Attribution Model: In the first-touch attribution model, the first touchpoint receives 100% of the credit for the sales conversion.

• Time-Decay Attribution Model: In the time-decay attribution model, the touchpoints closest in time to the sales conversion get the most credit. In this case, the last four touchpoints before the sales conversion receive the most credit, whereas the others receive significantly less.

No, MTA models estimate propensity to convert rather than demand and are therefore not directly applicable to forecasting. While demand curves can be inferred from MTA models, they typically do not have much validity at the campaign level and only inform sub-channel tactical decisions without forecasting or strategic decision-making support.

Many brands choose to implement an attribution platform. This is marketing technology software that captures user-level events across marketing channels and applies an algorithmic model to assign appropriate credit to the media touchpoints. This is an option to taking on the enormous task of building an MTA system in-house. There are also “full-service” MTA providers that instrument the tracking of the user-level events across media publishers and platforms, applying their own proprietary attribution models and delivering a bespoke reporting tool. 

MTA is an extremely difficult exercise to land. With each new channel added to the digital marketing mix comes another level of added complexity. MTA can take months to implement. It’s expensive and complicated. Additionally, without user-level tracking, it’s basically dead.

This is why Measured bet on incrementality testing and cohort-analytics as the future of measurement. It’s an effective solution to performance measurement because it’s not plagued by the user-level data challenges encountered by MTA.  

Deployed within the publisher platforms themselves, Measured experiments provide marketers with a true understanding of the incremental contribution of each marketing channel down to the most granular level. In addition, incrementality measurement is faster to set up. It can be used for scale testing and forecasting, and it measures the impact of both addressable and non-addressable media.

Download the guide to learn more about incrementality testing and experiments.