Marketing attribution models explained for small businesses

The short answer

Marketing attribution is the method used to share credit for a sale or enquiry among the marketing contacts that came before it. First-touch gives all credit to the first contact, last-touch to the final one, and other models split it. A small business usually does best comparing first-touch and last-touch, then asking customers how they found it.

What marketing attribution is

Marketing attribution is the rule you use to share credit for a sale or enquiry among the marketing contacts that came before it. A buyer rarely goes from one advert straight to checkout. They may watch a video, read an article some weeks later, open an email and then click a search advert before they buy. Attribution answers one question: which of those contacts gets the credit in your reports?

The answer matters because credit steers money. A channel that gets credit looks worth funding. A channel that gets none looks like waste, even when it started the journey. No rule is correct in an absolute sense. Each model is a choice about what to value, and each leaves something out.

Attribution is also separate from tracking. Tracking records what happened: visits, clicks, form submissions and orders. Attribution is the decision about how to divide the result across those records. If tracking is missing or broken, no model repairs it, so get that right first.

One customer journey, six ways to credit it

Take a business that sells handmade ceramic cookware online. One customer's path looks like this:

1. They watch a short cooking video from the brand on YouTube.
2. Weeks later they search for a style of pan and land on a blog post.
3. They join the email list to get a care guide.
4. They open a follow-up email but do not click anything.
5. Days later they click a search advert for the brand name and buy.

The channels in that path are covered in the guides to YouTube marketing for a small business and planning content marketing. Here the only question is who gets the credit. Each model answers differently.

Last-touch causes the most arguments. A search advert for your own brand name will nearly always appear near the end of a journey, because people search for a business they already know. It collects credit that other channels earned.

First-touch vs last-touch attribution

First-touch gives all the credit to the first recorded contact. Last-touch gives all of it to the final contact before the sale. They answer different questions. First-touch answers where new people come from. Last-touch answers what closes the sale.

Neither is wrong, and each hides what the other shows. A channel that opens many journeys but rarely closes them, such as videos or articles, looks strong in first-touch and empty in last-touch. A closing channel, such as brand search or a reminder email, looks the other way round.

Use the gap as information. Pull the same month of results under both views. If a channel scores well on one and poorly on the other, you have found its role: opener or closer. Where a purchase is quick and made in one visit, the two views will look much alike and the choice hardly matters. Where buyers take time, ask for quotes or talk to someone first, the views will differ a lot.

• To judge how new customers find you, look at first-touch.
• To judge which reminders and offers finish sales, look at last-touch.
• Before cutting any channel, look at both.

What Google Analytics 4 shows you

Check the current documentation before you rely on any detail here, because tools change their options. At the time of writing, Google's Analytics Help lists three attribution models for reporting: data-driven, paid and organic last click, and Google paid channels last click. The older rule-based models, including first click, linear, time decay and position-based, have been removed.

Google Analytics 4 also keeps two views of where people come from. The user acquisition report uses "first user" dimensions, which record what brought a person the first time. The traffic acquisition report uses "session" dimensions, which record what brought each visit. Read together they give a rough first-touch and last-touch comparison. Google advises against comparing numbers between the two reports, because they count different things.

Advertising platforms apply their own settings. A sale can be counted by an ad account and by your analytics under different rules, so the totals rarely match. Choose one source for decisions, usually your analytics, and treat the ad platform's figures as supporting evidence.

Do not mix attribution up with diagnostics inside an ad account. Quality Score in Google Ads is a diagnostic about your ads and keywords, not a way of sharing credit for sales. The guide to Google Ads Quality Score explains what it does measure.

Which attribution model suits a small business

For most small businesses the answer is not one model. It is two views compared each month, plus the customer's own account of how they found you. Small volumes make any model unstable: with only a handful of sales a month, one customer can swing the percentages. At that size, reading individual journeys tells you more than a model does.

A workable routine:

1. Choose the one action that counts as success, such as a purchase or a booking request.
2. Record that action as a conversion in your analytics, and tag campaign links so sources are not lost.
3. Each month, compare first-touch and last-touch results for the same period.
4. Add a question to the checkout or enquiry form: "How did you hear about us?" with a free text box.
5. Write down what the numbers cannot see, such as recommendations and visits to a shop.
6. Change budgets in small steps and note what you changed and when.

Self-reported answers catch what software cannot: a podcast mention, a friend's message, a conversation at an event. They are imperfect because people forget, but they often reveal channels that leave no click trail. Treat them as a cross-check on the model, not a replacement for it.

A data-driven model is a reasonable choice if your tool offers it and you have enough conversions for it to work. It is the default in Google's tools. Its weakness is that you cannot easily see why it credited a channel, so keep a simple first-touch and last-touch view beside it.

Where attribution runs out

Every model works only on the contacts it can see, and several things sit outside that:

• People who decline tracking or clear their cookies leave gaps or look like new visitors each time.
• A person who researches on a phone and buys on a laptop can look like two people.
• Phone calls, shop visits and word of mouth leave no click trail.
• Adverts that were seen but not clicked rarely appear in last-touch views.
• Open counts in email reports are only a rough signal, so judge email by clicks and replies.

An email list helps here for a plain reason. When someone signs up, you learn who they are and where they signed up, and you can connect later purchases to that moment if you stored the source and have consent. That is one reason to build a list, covered in the guide to email list building for a small business. Rules on consent differ by country, so check them with a lawyer or the official regulator where you sell.

A monthly attribution review

A short monthly review beats an elaborate model that nobody reads. Answer the same questions each time and write the answers down so you can see changes over time.

• Which channel introduced the most new customers?
• Which channel closed the most sales?
• Did any channel appear in many journeys but never at the end?
• What did customers say when asked how they found you?
• What will you change before next month, and what will you leave alone?

Be wary of two conclusions. The first is cutting a channel because it never gets last-touch credit. Test that by pausing it for a set period and watching new enquiries, remembering that seasons and other changes can blur the result. The second is crediting brand search with sales that other channels created.

If you are still deciding which channels belong in the mix, the guide to digital marketing channels for a small business sets them out. For help with tracking and reporting, the digital marketing service page explains what Liftup Technologies offers.

Related guides

• Digital marketing
• Youtube marketing basics for small business
• Content marketing plan for a small business
• Digital marketing basics for a small business channels explained
• Marketing funnel explained
• Transforming marketing with ai

Questions and answers

What is marketing attribution?

Marketing attribution is the method for sharing credit for a sale or enquiry among the marketing contacts that came before it, such as a video, an email and an advert. The rule you pick changes which channels look useful in your reports. It sits on top of tracking, so it only works on contacts your tools can record.

What is the difference between first-touch and last-touch attribution?

First-touch gives all the credit to the first recorded contact, so it shows where new people come from. Last-touch gives all of it to the final contact before the sale, so it shows what closes the deal. Comparing the two reveals which channels open journeys and which finish them.

Which attribution model suits a small business?

Most small businesses are better off comparing first-touch and last-touch results each month than trusting a single model. Add a "how did you hear about us" question to capture word of mouth and offline contacts. With very few sales, read individual customer journeys instead of percentages.

Is data-driven attribution better than last-touch?

Not automatically. Data-driven attribution estimates each contact's share from patterns across many journeys, so it can reflect real paths better than a fixed rule. It needs enough conversions, and its workings are hard to inspect. Many teams keep a simple last-touch view beside it as a check.

Why do my ad account and my analytics show different sales?

Each tool applies its own attribution settings, counting window and tracking method, so they rarely agree. Choose one source for decisions, usually your analytics, and use the other as a cross-check. Check each tool's current documentation for how it counts conversions.

Does attribution still work if visitors decline cookies?

Only partly. Visitors who decline tracking may be missing from your reports or may look like new visitors each time. Self-reported answers, direct enquiries and email sign-up sources help fill the gap. Rules on tracking consent differ by country, so check them where you sell.

Can I do attribution without paid software?

Yes. Free analytics, tagged campaign links and a "how did you hear about us" question are enough to start. Paid tools mostly add convenience and extra models, and they still depend on the same underlying tracking.