First Touch Attribution: How It Works, Worked Example and GA4 Setup

First touch attribution is a model that gives 100% of the credit for a conversion to the first recorded interaction a buyer had with your brand, such as the blog post they found on Google or the ad they first clicked. Everything that happened afterward gets nothing. It answers one question well, “which channel introduced this customer?”, and says nothing about what convinced them to buy.

This guide covers how the model works, when it is the right view and when it misleads, a worked example with real arithmetic, what Google Analytics 4 still offers after the first click model was removed, and how to capture the first touch yourself so it survives until the sale.

What Is First Touch Attribution?

First touch attribution, also called first click or first interaction attribution, is a single-touch model. When a purchase, sign-up or lead happens, it looks back to the earliest touchpoint it can find for that person and assigns the whole conversion, and its value, to that touchpoint’s channel, campaign or landing page.

The reasoning is simple: without the introduction, the rest of the journey never starts. That makes first touch a measure of demand creation. It is not a measure of what closed the sale, and it does not claim to be.

How first touch attribution works: one journey

Take one buyer of a project management tool:

1. Day 1: they search Google for “how to run a sprint retro” and read your guide.
2. Day 6: they click a LinkedIn ad and look at the pricing page.
3. Day 9: they download a template and join your email list.
4. Day 20: they click a trial reminder email and start a paid plan.

First touch gives the whole sale to organic search, and specifically to the retro guide. LinkedIn and email get 0%. Last touch would do the reverse and give it all to email. The journey is the same; only the question changes.

Pros and Cons of First Touch Attribution

First Touch vs Last Touch Attribution

Last touch attribution gives all the credit to the final interaction before the conversion. The two models are opposite ends of the same single-touch idea, and they disagree most about the channels that matter most to growth.

That last row matters more than most guides admit. Last touch only needs the converting session, so it is hard to break. First touch needs you to remember a visit that may be weeks old, which is where most first touch reports go wrong. For the full picture on the other side, see our guide to last click attribution.

First Touch vs Other Attribution Models

Outside attribution, two methods measure what a channel causes rather than what it gets credit for: incrementality tests (turn a channel off for some regions or weeks and compare) and marketing mix modeling (a statistical model of spend against sales over time). Our guide to data-driven attribution covers the algorithmic option.

First Touch Attribution Example, With the Math

The fastest way to understand first touch is to run the same deals through several rules. The six deals and the spend figures below are illustrative, not data from a real company. Total revenue is $6,000.

How each column is built:

• First touch: organic search opened A and D ($1,200 + $600 = $1,800). ChatGPT opened C ($1,500), paid search opened E ($900), paid social opened B ($800) and Direct opened F ($1,000).
• First non-direct: F started as Direct, so the rule moves to its next touch, paid search (brand). Paid search becomes $900 + $1,000 = $1,900.
• Last touch: B, C and D end on Direct ($800 + $1,500 + $600 = $2,900). Paid search closes A and F ($1,200 + $1,000 = $2,200). Email closes E ($900).
• Last non-direct: Direct is skipped when anything else exists. B goes to organic search (brand), D to organic search and C to ChatGPT, so organic search gets $800 + $600 = $1,400.

Return per dollar changes with the model

Now add illustrative spend for the same period and divide credited revenue by spend:

Paid social is the clearest case. Under last touch it returns nothing and looks like the first thing to cut. Under first touch it returns $2 per dollar, because it opened a $800 deal. Email is the reverse: a strong closer that introduced nobody. Neither number is “true.” A channel that moves a lot between the two views is doing a specific job, and cutting it based on the wrong view breaks the chain.

The same example by landing page

For SEO, the useful unit is the page, not the channel. In this example, /blog/crm-migration-checklist opened two deals worth $1,800 under first touch. Under pure last touch it gets $0, and under last non-direct it keeps only D’s $600. The /compare page that ChatGPT sent a visitor to opened a $1,500 deal. Reporting first touch revenue by landing page tells you which articles start valuable journeys, which a traffic report alone cannot. Our guide to SEO conversion tracking covers landing-page reporting in more depth.

The Direct Problem

Deal F shows a decision every first touch report has to make. A first visit recorded as Direct can mean someone typed your URL because they already knew you, an email or chat app removed the referrer, or tracking failed on that first visit. You have three options:

• Keep Direct as the first touch. Honest, but it hides the real source when the cause was a stripped referrer.
• Skip Direct and use the next touch. This is what the “first non-direct” column does. It gives F to paid search (brand), which arguably harvested demand rather than created it.
• Ask. A “how did you hear about us?” field catches podcasts, referrals from friends and AI recommendations that never produced a trackable first click.

Pick one rule, write it down and keep it. Changing the rule mid-year moves credit between channels and looks like a performance change when nothing happened.

Does GA4 Have First Touch Attribution?

The top results disagree here. Some say Google Analytics offers first touch by default; others say GA4 removed it. Both are half right.

• The first click model is gone. Google states that the first click, linear, time decay and position-based models are no longer available as of November 2023. You cannot pick first click as the reporting model for key events.
• First user dimensions still exist. First user source, First user medium and First user default channel group describe the source “by which the user was first acquired”. Put First user default channel group as the row in an exploration and key events or purchase revenue as the value, and you get a user-level first touch view. It is the source of the user’s first session, judged with GA4’s own rules, which skip Direct when another source is available.
• BigQuery keeps it too. In the GA4 BigQuery export, the traffic_source fields hold the source, medium and campaign that first acquired the user, and they do not change with later campaigns. That is the raw material for a first touch model in your own warehouse.

The catch: “first acquired” means first seen by that GA4 cookie on that device and browser. If the cookie is deleted or the person switches devices, GA4 acquires them again as a new user with a new first source. That leads to the main failure modes.

Why Your First Touch Is Often Not the First Touch

A first touch report is only as old as the memory behind it. These are the common reasons the touch it shows was really the second or third, and how to check each one.

The Safari behavior is documented by WebKit: Intelligent Tracking Prevention deletes all of a website’s script-writable storage after seven days of Safari use without user interaction on the site. For any tool that remembers the first visit in a browser cookie or local storage, a Safari buyer with a long gap between visits will look new. For timing choices, see attribution windows.

How to Set Up First Touch Attribution

You can read first touch from a CRM, from GA4’s First user dimensions, from a warehouse, or from a dedicated attribution tool. Whichever you use, these five steps decide whether the numbers hold up.

Step 1: Decide what counts as a touch and how far back to look

A touch can be a website visit, an ad click, an event badge scan or a sales call. Mixing them without a rule makes reports impossible to compare. Then pick a lookback that covers almost all of your time from first visit to conversion. A 30-day window on a 90-day sales cycle will mislabel many deals.

Step 2: Capture the first visit once

On the first page view, store the landing page, the referrer, any UTM parameters and the timestamp. Write them only if nothing is stored yet, so later visits never overwrite them. Keep a separate “latest touch” set if you also want last touch.

Step 3: Carry it into forms and the CRM

Copy the stored values into hidden fields on every lead and sign-up form, and save them on the contact as write-once properties. This moves the first touch out of the browser, where it can expire, and onto the record that will eventually hold the revenue. HubSpot does a version of this automatically with its Original Traffic Source property, which records the first known web source of a contact. Note that it can be edited manually, so lock down who can change it.

Step 4: Put a value on every conversion

Counting a newsletter sign-up and a demo request as one conversion each makes every model misleading. Give each conversion type a dollar value, using closed revenue when you have it and an expected value when you do not. Our guide on how to calculate conversion value walks through the formula.

Step 5: Report by channel and landing page, next to last touch

Sum value by first-touch channel and by first landing page, and put a last touch column next to it for the same period, like the example above. The gap between the two columns is the most useful number in the report: it tells you which channels open and which close.

When to Use First Touch Attribution, and When Not To

The B2B question splits the top results. Some call first touch ideal for long sales cycles, others a poor fit. Both are right about different jobs: it is a good lead source record and a poor explanation of what closed the deal.

First Touch Attribution for SEO and AI Assistant Traffic

Organic search and AI assistants tend to start journeys rather than finish them, which makes first touch the view where they show their real contribution.

Informational SEO pages are usually read early. The reader leaves, comes back days later through branded search or an email, and converts. Last touch gives the page nothing. First touch by landing page, as in the example, shows which articles start the journeys that end in revenue.

ChatGPT, Perplexity and other assistants are often a research step. When the assistant passes a referrer, the first visit is identifiable and first touch can credit it, like deal C above. When it passes none, the visit arrives as Direct, and no model can recover the source. Many people also read an AI answer, never click, and search your brand later; their first recorded touch is branded search. A self-reported source field is the only practical way to see that group. Our guide to AI conversion tracking covers what can and cannot be identified.

SEOConversion reports conversions and their value by landing page for organic search and AI assistants, so you can see which pages and assistants bring in visits that convert. When an assistant sends no referrer, the visit stays Direct rather than being guessed.

FAQ