An attribution model is the rule an analytics or advertising tool uses to decide which marketing touchpoints get credit for a conversion, such as a sale, a booking or an enquiry form. Change the model and the same customer journey produces different numbers for each channel, even though nothing about the customer has changed.
How an attribution model works
Picture someone in Leeds who needs an accountant. On Monday they click an Instagram ad, look around and leave. On Thursday they search for a small business accountant in Leeds and click your Google ad. On Saturday they type your web address straight in and fill in the contact form. That is three touchpoints and one enquiry, and the model decides how that one enquiry is divided between them.
- Last click gives all the credit to the final click before the conversion. GA4 ignores direct visits when a campaign came before them, so here the Google ad would get it.
- First click gives everything to the first interaction, here the Instagram ad.
- Linear, time decay and position-based models split the credit evenly, weight it towards the most recent touches, or give most of it to the first and last.
- Data-driven attribution compares the paths of people who converted with the paths of people who did not, and credits each touchpoint in proportion to how much it appears to change the odds of converting.
At the time of writing (October 2026), Google Ads and GA4 offer only data-driven and last-click models. Google retired first click, linear, time decay and position-based from both products during 2023, so you will now meet those older models in other tools and older articles rather than in Google’s own reports. Meta Ads does not offer a choice of model in the same way: it credits a conversion to a Meta ad the person clicked or viewed within the attribution window you set.
Why it matters
The model decides which channels look profitable, and so where the budget goes. Under last click, brand search and remarketing tend to look excellent because they sit at the end of journeys that other channels started. Cut the Instagram campaign on that evidence and the brand searches it was feeding may quietly fall away a few weeks later.
It also explains why your platforms never agree. Google Ads, Meta and GA4 each apply their own model to their own view of the journey, and each will claim the same enquiry. Add up the conversions every platform reports and the total is normally higher than the number of enquiries in your inbox. I go through the usual causes in why GA4 and Facebook conversion numbers do not match.
UK businesses have a further gap. Visitors who decline analytics cookies on your consent banner are not followed from one visit to the next, so their journeys are partly or wholly missing. Google fills some of this with modelled data, but no attribution model can credit a touchpoint it never saw.
Common mistakes
- Comparing figures from two reports that use different models, then treating the difference as a change in performance.
- Switching model in GA4 or Google Ads and reading the following month’s shift as a real rise or fall. A new model changes the arithmetic, not the customers.
- Treating data-driven attribution as proof of cause. It is a statistical estimate built only from the touchpoints the tool can see.
- Judging upper-funnel channels such as video or social prospecting on last-click figures alone.
- Adding up platform-reported conversions to work out cost per lead, instead of counting the leads you actually received.
How to act on it
Start by writing down which model each tool uses. In GA4 it sits under Admin, then Attribution settings; in Google Ads it is set on each conversion action. Keep them consistent where you can, and note the date whenever one changes so nobody misreads the next report.
Then choose one source of truth for results: the enquiries in your CRM, the orders in Shopify or the bookings in your system. Use GA4 with data-driven attribution for the cross-channel view, use each ad platform’s own reports to optimise inside that platform, and check both against the real total every month. When a channel looks weak under last-click attribution but useful under data-driven, treat that as a reason to test, not a reason to cut.
With a larger budget, the honest answer to what a channel is really worth comes from an incrementality test, such as pausing a channel in some regions and comparing them with the rest. Judging every channel against one cost-per-lead or return target, rather than against each platform’s own claims, is the core of performance marketing.