Marketing Attribution Models, Explained (for UK Small Businesses)

Marketing attribution is how you decide which clicks, ads and channels get the credit when someone buys from you or sends an enquiry. The practical headline for a UK small business: you can read about five attribution models, but in GA4 and Google Ads you can really only choose between two approaches now, data-driven and last click. This guide explains all five in plain English, says which to actually use, and is honest about where attribution stops telling the truth.

What marketing attribution actually means

Attribution is the rule that decides which touchpoints get credit for a conversion when a customer’s journey has more than one step. Almost nobody buys on the first visit. Someone might find you through a Google search, come back a week later from an Instagram ad, then type your name straight into the browser and convert. Three touchpoints, one sale. Attribution is the method you pick for splitting the credit between them, so you can see which marketing pulls its weight and which is along for the ride.

Get it wrong and you make expensive mistakes: cutting the channel that quietly starts journeys, or overspending on the one that only ever closes deals other channels warmed up. Get it roughly right and you spend where it counts.

The five attribution models, in plain English

There are five models worth knowing, and they differ only in how generously they hand out credit along the path. Four follow a fixed rule you could work out on paper. The fifth lets Google’s modelling do the sums.

Last-click attribution

Last click gives 100% of the credit to the final click before the conversion, and nothing to anything earlier. It is the default most people picture, because it is the easiest to explain and the easiest to measure. The catch is that it flatters whatever comes last, usually brand search, direct traffic or retargeting, while the blog post or organic result that started the journey gets nothing. Useful as a quick sense-check, risky as your only lens.

First-click attribution

First click does the opposite, handing all the credit to the very first interaction that brought someone in. It is handy when your main question is “what introduces new people to us?”, because it values discovery over closing. The flaw mirrors last click: it ignores everything after the first visit, including the ad or email that finally persuaded someone to act.

Linear attribution

Linear attribution splits the credit evenly across every touchpoint on the path. If there were four clicks, each gets 25%. It is the fairest-sounding model and the least decisive, because not every touch matters equally. A throwaway click and the one that genuinely changed someone’s mind are treated as identical, which is tidy but not true to life.

Position-based (U-shaped) attribution

Position-based attribution, also called U-shaped, gives 40% to the first click, 40% to the last, and shares the remaining 20% across everything in between. The logic is that the introduction and the close are the two moments that matter most, with the middle playing a supporting role. It is a sensible compromise for businesses that care about both finding new people and sealing the deal, which is most of them.

Data-driven attribution

Data-driven attribution lets Google’s machine learning share the credit based on your own account’s data, rather than a fixed rule. It compares paths that converted with paths that did not, then works out how much each touchpoint genuinely contributed, giving fractional credit that adds up to the whole. The order of interactions, device type and time from conversion all feed in. Because it is built from your real data, it tends to be the most realistic picture, and the one Google now steers you towards.

A quick comparison

Here is the whole set side by side, with what each is good for.

Note: in GA4 and Google Ads today you can only select data-driven or last click. First click, linear and position-based are included here to explain the thinking, not because you can still switch them on.

What you can actually use in GA4 and Google Ads

This is the part that catches people out: three of the five models above were retired, so your real choice now is data-driven or last click. Google removed the first-click, linear, time-decay (more credit to clicks nearer the sale) and position-based models from Analytics and Google Ads in November 2023, after finding that only a tiny share of conversions ever used them (Google Analytics Help, Get started with attribution).

In GA4 you now get three options: data-driven attribution, “paid and organic last click”, and “Google paid channels last click”, with data-driven set as the default for new properties (Google Analytics Help, Select attribution settings). Google Ads is simpler still, with two models, data-driven and last click, and data-driven as the default for most conversion actions (Google Ads Help, About attribution models).

So while it helps to understand first-click, linear and position-based thinking, you will not be toggling them on. The decision in front of you is narrower, and easier, than most articles make it sound.

Which model should a UK small business use

For most UK SMBs, use data-driven attribution as your main model and keep last click as a cross-check. Data-driven gives the fairest read on which channels combine to win business, and it is already the default, so many accounts use it without having chosen it. Last click alongside it is a useful reality check: if the two disagree wildly about a channel, that is a prompt to look closer, not a crisis.

There is one honest caveat. Data-driven modelling needs a reasonable flow of conversions to learn from. Google dropped the old minimum-data thresholds, so it is open to every account now, but on very low volumes it has less to work with and last click can be steadier. If you get a handful of enquiries a month, do not agonise over the model. Agonise over tracking those enquiries cleanly.

That last point is the one that actually moves the needle. A model only redistributes the conversions it can see, so it is worth making sure GA4 records the right actions. If you are unsure your enquiries and sales are counted properly, our guide to key events versus conversions in GA4 clears up what to mark, and the GA4 setup guide for UK websites covers the install from scratch. For paid campaigns, the same applies to Google Ads conversion tracking: no attribution model can fix a conversion that never fired.

The limits of attribution (the honest bit)

Whatever model you choose, attribution is an educated estimate, not a ledger, and treating it as gospel is how marketing budgets get misspent. Every model works only with the data it can see, and in the UK that data has real holes. Here is where it slips.

• Consent gaps. UK sites must get consent before analytics cookies fire, so a share of your visitors are never measured at all. Consent Mode models some of the gap, but your attribution is still built on the people who said yes, not everyone.
• Cross-device journeys. Someone discovers you on a phone at lunch and buys on a laptop that evening. Unless they are signed in, those often look like two different people, and the path breaks.
• Offline conversions. Plenty of SMB sales happen on the phone, in a showroom or after a quote. The click that triggered the call gets no credit unless you feed that outcome back in.
• The walled-garden problem. Google’s models see Google’s clicks best. Channels outside its view, from organic social to word of mouth, are routinely under-counted.
• Long research cycles. Lookback windows have an edge. A considered purchase researched over three months can start before the window even opens.

None of this means attribution is useless. It means you read it as a compass, not a map: decide on the trend over time, compare like with like month to month, and sanity-check the numbers against what you actually banked. The businesses that get value from it are the ones that hold it loosely.

A quick example (illustrative, not one client)

To make it concrete, here is a composite drawn from work we have done and anonymised, not a single client’s figures. A Leicester home-improvements firm judged every channel on last click and was about to cut its blog and paid social, because almost all conversions showed up under brand search and direct. Switching the reporting to data-driven told a different story: the blog and social ads were doing the introducing, and brand search was just where people landed once they already knew the name. The spend that looked wasteful was feeding the spend that looked brilliant. Nothing was cut, tracking was tightened so phone enquiries counted too, and the picture finally matched the sales pipeline.

Pick a model, then trust your tracking more than the model

The short version: use data-driven attribution, keep last click as a sense-check, and put more effort into clean measurement than into agonising over models. The model you pick changes how the credit is shared. It does not change what really happened. What changes your decisions for the better is trustworthy data underneath, counting the right conversions, the same way, every month.

If you want a straight look at whether your attribution and tracking are telling you the truth, that is exactly what our measurement and analytics work is for. Or book a free growth review and we will check how your conversions are being counted, which model your accounts are really using, and where your reporting and your sales have drifted apart. Prefer to talk it through first? Get in touch.