What Is Web Attribution? How It Works and Limits

Which campaign actually brought the customer who bought last Tuesday?

Your sessions report cannot say. Say 4k people visited, most came from "social" and "direct", and a few hundred signed up. It does not say which ad, which creative or which channel earned the sign-up. Budget gets moved on the first number while revenue lives in the second.

That gap is what web attribution closes. Here is how it works, where it stops working, and how to start.

What is web attribution?

Web attribution is the practice of linking website visits, sign-ups and purchases back to the campaigns and sources that brought each visitor, using UTMs, ad click IDs and a visitor identity.

It joins three things that normally live in separate places:

A visit: a person lands on your site, with a page, a time and a device
A conversion: that person signs up, starts a trial or buys, often with a value
A source: the campaign, ad or channel that sent them there

A hypothetical journey makes it concrete. A visitor taps a paid social ad on Monday and browses. They return on Thursday from a search result and buy. Web attribution is the system that records both visits, ties them to one person, and decides which source gets credit for the purchase.

How does web attribution work?

It runs in four steps.

1. Capture the source on landing

UTM parameters such as utm_source and utm_campaign are tags you add to the link. Consistent naming decides whether the data is usable, so a naming standard like UTM naming conventions that scale is worth agreeing before launch.
Ad click IDs (gclid for Google, fbclid for Meta, ttclid for TikTok) are added by the ad platform itself. They let the platform match a visit back to the exact ad click. Visits from AI assistants can arrive with no UTM or click ID, so see how to track AI assistant traffic to your website.

2. Recognise the visitor

A visitor ID is stored in the browser on first visit, so repeat visits join up.
At sign-up or login, you pass a user ID, so the anonymous history and the known customer become one record.

3. Record conversion events

Send named events such as sign-up and purchase, with a value where relevant.
Page views alone cannot be credited with revenue. Events are what make the report about outcomes.

4. Decide where the credit goes

First touch credits the source that first brought the visitor. Last-touch attribution credits the most recent tagged source before the conversion. Multi-touch attribution splits credit across several.
Neither is correct in all cases.
Our guide to first touch vs last touch attribution for websites shows how to keep both.
Our comparison of multi-touch against single-touch models covers the trade-off.
There is also a fair case that last-click is enough for many teams.
Any model also needs an attribution window: how long after a touch a conversion can still be credited to it.

Tech Explainer: how long does the data last? Storage rules decide what the model can see. In Linkrunner's Web SDK (beta), first-touch UTMs and click IDs are kept until browser storage is cleared. Last-touch click IDs are kept for 90 days. Last-touch UTMs are kept for 24 hours from the campaign click. A reload, a back button press or an internal link does not count as a new touch, and first touch never changes. Other tools choose different windows, so check what yours does.

Web attribution vs web analytics: what is the difference?

Web analytics answers "what happened on the site". Web attribution answers "which source earned the conversion".

Analytics reports sessions, pages, bounce and flows, grouped by channel
Attribution follows a person across visits and assigns a conversion to a source, with a model and a window
Analytics is about traffic, attribution is about outcomes. You need both

Why do the two numbers disagree? Analytics tools usually count sessions and credit by their own channel rules. Attribution counts people and credits by an explicit model. Ad platforms report their own conversions using their own rules, so their totals are often higher. Three honest reports with three definitions will not match, which is fine if you know which question each answers.

What does web attribution not solve?

Be sceptical of any page that claims perfect tracking. Real limits:

Cleared storage and private browsing. A visitor who clears their browser data looks like a new person on return. A payment gateway return in a new tab can lose the source too, as covered in tracking UTMs through payment gateway redirects.
Ad blockers and tracking prevention. Browser and extension rules can block third-party tracking scripts, so some visits are never recorded. Treat web numbers as a floor, not a census.
Cross-device journeys. Someone who clicks on a phone and buys on a laptop is two visitors until they sign in on both.
View-through. A person who saw an ad but never clicked leaves no UTM or click ID to capture, so click-based web attribution cannot credit it.

First-party collection is a common mitigation for blocked scripts. Instead of loading from a third-party domain, the tracker is served from your own site or a domain you control, which blockers may treat more leniently. It reduces loss but does not remove it, and nothing fixes cross-device journeys without a login.

Web attribution for teams with an app

Many teams run one campaign that sends people to both a website and an app. Measuring them separately produces two stories about the same spend.

Linkrunner is a mobile and web attribution platform. It started with apps (installs, deep links, ad spend) and now also measures website visits, sign-ups and purchases through its Web SDK (beta), so web and app results can be read in one platform.
Seeing web visits and app installs side by side shows which campaigns lead to a visit, an install or both.
The Linkrunner web attribution page explains how the two fit together.

Web-to-app is a different problem: moving a web visitor into an installed app with context intact is a job for deferred deep links, covered in deep linking for web-to-app. Web attribution tells you where the visitor came from, while web-to-app attribution tells you what happened after they crossed over. Web attribution vs web-to-app attribution compares the two.

How do you start measuring web attribution?

A checklist you can run this week:

Tag every campaign link with UTMs, using one agreed naming pattern.
Capture click IDs on landing, so Google, Meta and TikTok visits can be matched.
Identify users at sign-up or login, so history joins the customer record.
Send sign-up and purchase events, with a value on purchases.
Choose a default model. First touch is a reasonable start for awareness-led teams, last touch for performance-led ones. Keep both visible if your tool allows.
Test end to end. Click a tagged link, sign up, buy, and confirm the right source appears on the right conversion.

One way to do this is Linkrunner's Web SDK, currently in beta. It is an SDK that captures UTMs, ad click IDs, page views, sign-up and purchase events, and classifies traffic as paid, organic, social, AI assistant, referral or direct. Its reporting API returns last-touch credit by default, or first-touch on request. Setup details are in the Web SDK docs.

FAQ

What is web attribution? Web attribution links website visits, sign-ups and purchases back to the campaigns and sources that brought each visitor. It uses UTMs, ad click IDs and a visitor identity to decide which source earns credit for a conversion.

How is web attribution different from web analytics? Web analytics describes what happened on your site, such as sessions, pages and channels. Web attribution assigns each conversion to a source using a model and a window. Analytics measures traffic, attribution measures outcomes.

Do I need cookies for web attribution? It needs some way to remember a visitor between visits, and that is usually browser storage such as a cookie or local storage. Without any storage, you can still read UTMs on a single visit, but you cannot link a later purchase to an earlier ad click.

Can I use web attribution without an app? Yes. Web attribution works for any website with campaigns and conversions, such as a SaaS site or an online store. An app is not required. If you do have one, measuring both in one place avoids two conflicting reports.

What should I track first? Start with consistent UTMs on every paid link, then the two conversions that matter most, usually sign-up and purchase. Add a user ID at login so anonymous visits join the customer. Refine the model after the basics are clean.

Where to go from here

Web attribution is simple in principle: capture the source, recognise the person, record the conversion, credit one source by a stated rule. Its value comes from the discipline of consistent tags and tested events, and from being honest about the limits.

If you run campaigns to a website, an app or both, request a demo from Linkrunner to see mobile and web attribution in one place, with the Web SDK in beta. Or run the six-step checklist above on your own site first.