First-touch Attribution

One customer doesn't arrive in one motion. They find a blog post through search, see an ad weeks later, search again, and sign up. With three touchpoints in play, every channel can claim it brought the person in unless the rule for assigning credit is settled in advance.

Ravi hit that while building a monthly report at Pagedeck. Adding up conversions by channel gave 2.4 times the actual number of signups. Each channel was measured with a different tool, and every tool counted its own click as the final touchpoint.

First-touch attribution is the version of that rule that picks the first touchpoint. The view behind it is that without the first meeting none of the rest happens, which makes it the model for judging work that brings new people in.

Pull the touchpoint history of everyone who converted, in time order, and take the earliest one. Four things get decided.

1. Put the recording in place first. Tag links with UTM parameters so the path in is identifiable, and make sure that value survives until signup. Without the record there's nothing to pick a first touch from.
2. Set the lookback window. Thirty or ninety days is typical, and touchpoints older than that don't count as the first. Set it short and the starting point of a customer who evaluated for two months gets cut off.
3. Pin down what counts as a conversion. A signup and a payment give the same channel very different scores.
4. Assign 100 percent to the earliest touchpoint and sum by channel. Read a blog post, see an ad, arrive through search and sign up, and the whole conversion lands on the blog. Nothing is left over for the ad or the second search.

The output is one number per channel, which can be used for budget allocation as it stands.

"The channel totals come to 2.4 times our signups. Let's fix one model and pull it again."

Without a single rule the numbers can't be added together.

"This is first-touch, so ads score low. Ads usually land in the middle."

The model structurally understates certain channels.

"A 30-day lookback cuts off the first touch of anyone who evaluated for two months."

The window setting changes the result.

Ravi pulled the same month under two models. Under first-touch, the blog came out as the channel bringing in the most signups; under the model that credits the last touchpoint, search ads took first place. Neither was wrong — they answered different questions. The budget for bringing in new people got set from the first number, and the budget for pushing known people to pay from the second.