Marketing Attribution: Who Gets Credit for a Sale?

Why it exists

Customers rarely buy on their first visit. They might see a social media ad, read a blog post, get an email, and then finally click a search ad to purchase. Without attribution, you can't know which of these efforts actually worked or how they worked together. It solves the problem of figuring out which half of your advertising budget is being wasted.

The mental model

Think of attribution like analyzing a championship sports team. The final goal wins the game, but what about the key pass that set it up, or the defensive play that started the counter-attack? Attribution is the process of reviewing the whole game tape to decide how much credit each player (marketing channel) deserves for the final win (the conversion). Giving all credit to the goal-scorer (the last touch) misses the bigger picture of how the win was created.

How it works

Attribution systems track a user's interactions, or "touches," across marketing channels leading up to a desired outcome, like a purchase. A model is then applied to assign value to each touch. Common models include: First-Touch (100% credit to the first interaction), Last-Touch (100% to the last), Linear (equal credit to all touches), and Time-Decay (more credit to recent touches). More advanced data-driven models use algorithms to assign credit based on statistical analysis of which touches most influence conversions.

When to use it

Use attribution when you have multiple marketing channels running and need to justify spend or optimize performance. It's essential for any business where customers interact with your brand multiple times before converting, which is common in e-commerce, B2B software, and most online businesses with a considered purchase.

When not to use it

For businesses with a single marketing channel or an extremely short, simple sales cycle (e.g., a user clicks one ad and buys immediately), complex multi-touch attribution is often overkill. If you can't reliably track users across devices, the data can be incomplete, making any model's conclusions unreliable.

One canonical example

A user sees a Facebook ad for sneakers (touch 1), reads a blog post review on your site a week later (touch 2), and then clicks a promotional email to complete the purchase (touch 3). A last-touch model gives 100% credit to the email. A first-touch model gives 100% credit to the Facebook ad. A linear model would assign 33.3% credit to each of the three touches, acknowledging that they all played a part in the final conversion.