How Attribution Models Assign Credit Across the Customer Journey
First-touch, last-touch, linear and time-decay models each split credit for a sale differently. How they work, their tradeoffs, and how to choose one.
Before buying, people typically encounter a brand several times: a social media post, a display ad, a paid search ad. For marketers, knowing which of those touchpoints actually moved someone toward a purchase is essential to spending well.
Attribution models are the rules that answer that question. They decide which channels, posts and assets get credit for a sale, and so they shape where the next budget goes.
The common attribution models
Take a simple journey: someone sees a social media post, later sees a display ad, and finally clicks a pay-per-click (PPC) search ad and buys. Each model credits that journey differently.
First-touch
All the credit goes to the first interaction that brought the person to you, here the social post. This model highlights top-of-funnel marketing, but it ignores everything that happened afterward.
Last-touch
All the credit goes to the last interaction before the sale, here the PPC ad. It suits businesses with short buying cycles and assumes the touchpoints closest to the sale have the most influence. It is also the default in many tools, which is why paid search so often looks like the best channel.
Linear
Every touchpoint gets equal credit. The social post, the display ad and the PPC ad each receive a third. That is fairer to early touches, but it assumes every interaction mattered equally, which is rarely true.
Time-decay
Credit is weighted by how close each touchpoint was to the conversion, so recent interactions get the most. It reflects the assumption that touches nearer the sale had the greatest effect, while still giving earlier ones some credit.
Position-based
A common variation gives most of the credit to the first and last touches and splits the remainder among the ones in between, recognizing both the introduction and the close.
Choosing a model
Even if your tools default to last-touch, it's worth examining how each channel, post and asset contributed along the way. Looking at the same data through several models shows which conclusions hold up and which depend on the rule you chose.
The right model depends on:
- What you sell. Impulse purchases and considered purchases follow very different paths.
- How long people take to buy. The longer the cycle, the more early touchpoints matter and the more a last-touch view distorts.
- Which channels you use. If much of your spend goes to channels people see but don't click, click-based rules will undercount them.
Our guide to multi-touch attribution covers these models and their tradeoffs in more depth.
What attribution models can't see
Every rules-based model shares one limit: it can only credit touchpoints it can track. That has become a bigger problem since Apple's App Tracking Transparency and browser tracking protections cut the user-level data these models rely on. Channels such as TV, connected TV, podcasts and out-of-home are mostly invisible to them.
Attribution models also assign credit; they don't measure whether a touchpoint caused a sale that wouldn't otherwise have happened. For that, marketers turn to incrementality testing and to marketing mix modeling, which estimates each channel's effect on total sales from aggregate data.
Putting it to work
Attribution models remain useful for understanding the digital touchpoints you can track and for optimizing within those channels. Choose a model that fits your product and buying cycle, compare it with others, and pair it with a method that measures all channels' incremental impact before you make large budget moves.
Frequently asked questions
What is the purpose of an attribution model?
An attribution model is a set of rules for identifying which channels, posts and ads contributed to a conversion. It helps marketers understand the value of each touchpoint and decide where to spend.
Why look beyond the default last-touch model?
Last-touch gives all the credit to the final interaction, often a paid search click. That ignores the ads, posts and content that created interest earlier, so channels working higher in the funnel look less valuable than they are.
Which attribution model should I use?
It depends on what you sell and how long people take to buy. Short, simple buying cycles suit models that weight recent touches; longer considered purchases benefit from models that also credit early touchpoints. Comparing models side by side shows how sensitive your conclusions are to the choice.
What can't multi-touch attribution measure?
It only sees touchpoints it can track, mainly clicks and some digital views. Channels like TV, podcasts and out-of-home, and journeys lost to privacy restrictions, fall outside it. Marketing mix modeling measures those from aggregate data.