First touch and last touch, what each one answers

Most attribution arguments come down to one question: which model is correct. But first touch and last touch aren't measuring the same thing, so neither one is the answer to that question. A store reading only one of them is missing half the picture, in a direction you can predict.

What each one answers

Last touch answers: what closed the sale. It credits the final interaction before the order. It's the default in most tools because it's the cheapest to compute and the hardest to argue with, since the touch is sitting right next to the purchase.

First touch answers a different question: what started the relationship. It credits the interaction that produced the very first visit, however long ago, as long as it falls inside your attribution window.

Neither one measures a channel's value. Both measure its position in the journey.

Where each one is wrong

Last touch over-credits whatever sits closest to the purchase: retargeting, brand search, email, direct visits. Those channels mostly harvest demand rather than create it. A store optimising on last touch keeps adding budget to harvesting until there's nothing left to harvest. The decline doesn't show up in the report.

First touch over-credits the top of the funnel: broad prospecting, awareness video, creator content. It ignores that plenty of first touches lead nowhere, and it tells you nothing about what converted.

Both errors run in a known direction, which is what makes reading them together useful.

How to read the difference between the two

A campaign at 0.8× on last touch and 2.67× on first touch isn't a contradiction. Both numbers describe the same thing: 28 customers came in through that campaign and bought later, after a branded search or a direct visit. You only see it when you put the two columns side by side.

The third row is worth attention. In many stores a large share of orders have exactly one touch, and every model agrees on those.

Which one for which decision

• Growth budget. First touch. You're asking what brings new people in.
• Checkout or offer changes. Last touch. You're asking what converts.
• Cutting a channel. Both together. Cutting on last touch alone removes openers.
• Paying a creator or partner. First touch, since their role is usually to open.

The window comes before the model

With a 7-day window, a customer who takes 40 days to decide has their opening touch erased. Awareness campaigns, creator links and slow-working content all read as zero, not because they failed but because the tool can't see far enough back.

A model can only work with the touches inside your window. So before arguing about first versus last, check how far back your window reaches. Flowfy runs a 90-day window over stored journeys, so you can read the same period under either model without re-collecting anything. And if you want one number instead of two readings, there are multi-touch models.

Read your campaigns under both models over the same period and write down the difference per channel. Where first touch is much higher, don't cut on last-touch numbers.