A customer sees an ad on Instagram, days later reads an article of yours, later searches for your brand on Google and only then buys. Which of these touchpoints deserves the credit for the sale? The answer you give — your attribution model — determines where you invest the next marketing euro. And most companies use, without knowing, the most misleading model of all.
The problem: the sale has many parents
Rarely does anyone buy on first contact. The typical journey crosses several channels and several days. But tools need to attribute the sale somewhere, and how they do it decides which channels look like heroes and which look useless — even if reality is otherwise. Attributing wrong is investing wrong.
Last click: simple and treacherous
The most common model gives all the credit to the last touchpoint before the purchase. It is simple to measure, and that is why it dominates. The problem: it overvalues bottom-of-funnel channels (like brand search) and ignores those that created the interest at the start (like that ad that planted the seed). Cutting top channels because they "do not convert" is the mistake this model causes.
First click: the opposite extreme
The first-click model gives all the credit to the initial touchpoint. It corrects the last-click blindness, but falls into the symmetric error: it ignores everything that happened after and closed the sale. Both "single-touch" models tell only half the story.
Multi-touch models: splitting the credit
- Linear: splits the credit equally across all touchpoints — fair, but it assumes all were worth the same.
- U-shaped: gives more weight to the first and last (create and close), less to the middle.
- Data-driven: uses your data to estimate each channel's real contribution, instead of a fixed rule.
A concrete case
A company spent a lot on top-of-funnel ads that, by last click, seemed to yield nothing — all sales appeared attributed to brand search. They almost cut that investment. By switching to a multi-touch model, they discovered it was exactly that "useless" channel that started most of the journeys ending in a purchase. Cutting it would have dried up the funnel months later. Correct attribution avoided a decision that seemed obvious and would have been disastrous.
The perfect model does not exist
No model captures the absolute truth — human behavior is too complex. The goal is not perfection, it is to stop blindly using last click and start seeing the full journey. Even an imperfect model that recognizes the whole funnel beats a simple one that always rewards the end.
In practice
Before cutting a channel because it "does not convert", ask which attribution model is behind that judgment. If it is last click, you are probably looking at half the story. Look at the full journey before moving budget. Do you know which of your channels start the sales that others close?
