First-touch vs last-touch attribution: one journey, two answers
See how first-touch and last-touch rules assign credit, why lookback and direct-traffic rules matter, and what a model comparison cannot prove.
By Mika Garcia · Published
In this article
The difference is which eligible interaction gets credit
First-touch attribution assigns credit to the first eligible recorded interaction. Last-touch attribution assigns it to the last eligible recorded interaction before conversion. Both are single-touch rules: one interaction receives the credit in each view.
The word ‘eligible’ matters. A model also depends on which events were observed, the lookback window, the identity links, and whether direct visits count. The first recorded interaction is not necessarily the first time someone encountered your business.
This DATALYR guide compares the concepts using an illustrative journey. The examples are not customer results and should not be read as the exact behavior of every reporting platform.
Apply both rules to the same journey
Suppose a buyer clicks a paid social ad on day 1, follows an email link on day 4, and clicks a paid search ad on day 6 before buying for $200. Assume all three interactions are linked to the same buyer and are eligible under the selected window.
First-touch gives paid social all $200 of credit. Last-touch gives paid search all $200. The email receives no credit in either single-touch view, even though it is present in the observed journey.
Neither view creates or removes a payment. The $200 in the first-touch report and the $200 in the last-touch report are two allocations of the same amount. Do not add them together as though the campaign generated $400.
The useful question is why the allocation changes. In this example, one channel starts the recorded path and another appears near the purchase. That is a reason to inspect how the channels interact, not proof that either one is dispensable.
- Day 1: paid social click.
- Day 4: email click.
- Day 6: paid search click and $200 purchase.
- First-touch view: paid social $200; email $0; paid search $0.
- Last-touch view: paid social $0; email $0; paid search $200.
Which interaction receives the $200?
First-touch
Paid social: $200
Day 1 social → day 4 email → day 6 paid search and purchase.
Last-touch
Paid search: $200
The same path and payment, allocated under another rule.
Changing eligibility can change both answers
Now assume the reporting rule considers only interactions in the three days before the day-6 purchase. The day-1 paid social click falls outside that illustrative window. Email becomes the first eligible interaction, while paid search remains the last.
A report that says first-touch without specifying the window leaves out an important part of the rule. A short window may omit earlier research; a longer window may include interactions from a different buying episode. Select a window for a reason and record it with the report.
Missing identity links can have a similar effect. If the first click occurred on another device and no permitted link connects it to the purchase, the available journey begins later. Changing the model cannot recover an interaction the system never observed or connected.
Eligibility changes the first touch
All three clicks eligible
First: social · Last: search
Day 1 social, day 4 email and day 6 search are included.
Three-day lookback
First: email · Last: search
For the day-6 purchase, day-1 social falls outside the window.
Specify how direct visits are handled
Suppose the buyer returns directly on day 7 and purchases then instead. A literal last-interaction rule could credit that direct visit. A last-non-direct rule could retain the paid search interaction, assuming it remains eligible.
Those are different rules, even if both reports use the shorthand ‘last touch.’ Before comparing tools, write down whether direct sessions, impressions, email opens, and clicks are included. A click-only path and a path containing ad impressions are not equivalent inputs.
Keep the conversion itself consistent as well. Assigning credit for a trial and assigning credit for the first paid invoice answer different questions. A model comparison is most useful when only the model changes and the rest of the definition stays fixed.
Distinguish a reporting model from a delivery lookup
A conceptual first-touch comparison should not be substituted for a product’s actual conversion-delivery rules. DATALYR’s attribution documentation describes an ordered lookup for a conversion’s click context, including event, session, and linked-user information.
That documentation says the default is first click, a rule can select last click, and a same-session match takes precedence. This is more specific than simply taking the earliest or latest row across an unrestricted journey. Check the current documented rule and the actual conversion record before interpreting a delivery result.
Keep that lookup separate from whatever reporting model and date filters you are reviewing. A conversion’s delivery destination and an analyst’s allocation view may serve different purposes. Naming both prevents a generic attribution diagram from being mistaken for proof of implementation behavior.
Use disagreement to choose the next investigation
First-touch can help identify where recorded customer journeys begin. Last-touch can help identify which eligible interactions appear near conversion. Neither assigns value to all the assistance in between, and neither measures the counterfactual of removing a channel.
If a channel looks strong only in first-touch, inspect whether it regularly introduces buyers who later convert elsewhere. If it looks strong only in last-touch, inspect whether it reaches people already close to buying. These are hypotheses to investigate, not automatic instructions to move budget.
Keep a short review sheet for a few real transactions: conversion amount, eligible path, identity gaps, window, direct-traffic policy, and each model’s assigned source. Explain those records before relying on the aggregate difference.