The short answer
Attribution is deciding which piece of marketing gets the credit when someone becomes a customer. It matters because most people do not see one advert and buy. They notice you several times, in several places, before they get in touch.
There are different ways to share out the credit, and each tells a slightly different story. We deal with this in our tracking and analytics service, so this comes from doing the work. The aim is a consistent, useful answer, not a perfect one.
A simple example journey
Take a made-up customer, Sarah, who needs a new driveway.
- Week one: a neighbour mentions your firm.
- Week two: she sees one of your social media posts on her phone.
- Week three: she searches for driveway companies on her laptop, clicks your paid ad and reads your site.
- Week four: she searches for your company name, clicks the normal (unpaid) result and fills in your enquiry form.
Four steps, one customer. Which step earned the job? Your analytics will most likely say the last one, a search for your name. Sarah would probably say her neighbour.
Notice that only the last step produced something you can count: a form. If you do not yet track calls, forms and bookings, there is nothing to attach the credit to, so fix that first.
First touch, last touch and multi-touch
These are attribution models, which means rules for sharing out the credit.
| Model | Who gets the credit | Useful for | Weakness |
|---|---|---|---|
| First touch | The first step you know about | Seeing what introduces new people | Ignores what closed the sale |
| Last touch | The final step before the enquiry | Seeing what prompts action | Over-credits searches for your name |
| Multi-touch | Several steps, in shares | A fuller picture of long journeys | Needs more data than most small firms have |
In Sarah's case, first touch (as far as software can see) credits the social post. Last touch credits the search for your name. Multi-touch splits the credit between the post, the paid ad and the name search. None of them credits the neighbour, because no software saw that conversation.
Attribution is about customers, not visits. A channel can send plenty of visitors and few buyers, a gap we cover in website traffic vs conversions.
Why no model is right
Each model is a point of view, not a measurement. Last touch flatters the channels that catch people who had already decided. First touch flatters whatever starts conversations, even if those people rarely buy. Multi-touch looks more careful, but it still rests on a rule somebody chose.
The practical approach is to pick one model, use it consistently and know its bias. Then compare it with what customers tell you. To turn the result into pounds, our guide on how to measure marketing ROI shows the sum.
What software can and cannot see
Analytics tools such as GA4 record visits to your website and, when set up properly, the enquiries that follow. That is useful and worth doing. It is also incomplete. Tracking tends to miss:
- Offline conversations, such as a recommendation over the fence or at a networking event.
- Word of mouth in private messages and group chats.
- People switching device, such as seeing you on a phone and enquiring from a laptop.
- Visitors who decline cookies, and browsers that limit tracking.
- Vans, signs, print and anything else seen in the real world.
- Long gaps, where someone first visited many months before buying.
Rules on cookies and personal data apply to tracking, and what you need depends on your site and the tools on it. Check your own situation. This is not legal advice.

The low-tech fix: ask and record
The cheapest attribution tool is a question: how did you hear about us? Ask it on every phone call, add it to your enquiry and booking forms, and write the answer against the customer's record. It catches the neighbour, the van and the networking event that software cannot.
- Ask everyone at the first contact, not after the sale.
- Use a short fixed list of sources, plus 'other' with a note.
- If the answer is 'Google', ask what they searched for or whether they clicked an ad.
- Ask 'had you heard of us before that?' to catch the first touch.
- Keep the answer alongside what the software recorded, not instead of it.
Memories are imperfect, and many people will say 'online' when the real story is longer. That is fine. Two imperfect sources, the software and the customer's own account, give a better picture together than either does alone.
How much precision you actually need
Less than software vendors suggest. A small business is usually choosing between a handful of channels and needs to know which to keep, which to grow and which to stop. For that, knowing roughly where customers first heard of you and what finally prompted them is enough.
A fair test: would a more precise answer change a decision? If referrals and search clearly bring most of your paying customers, it does not matter whether search deserves a little more or a little less of the credit. Spend the effort on recording sources reliably, every time, before you spend it on a cleverer model.
Common mistakes
- Trusting last touch alone and cutting everything that appears not to convert.
- Crediting searches for your own name without asking what made people search.
- Never asking customers how they heard about you.
- Letting staff record 'internet' or 'website' as the source for everyone.
- Switching model every few months, so no two periods compare.
- Buying complex attribution software before basic enquiry tracking works.
Common questions
Which attribution model should a small business use?
Last touch from your analytics, plus the customer's own answer to 'how did you hear about us?', covers most needs. It gives you the end of the journey from software and the start from the person. Whatever you choose, keep it the same so periods can be compared.
Why does so much of our traffic show as 'direct'?
Direct usually means the tool could not tell where the visitor came from. It includes people who typed your address, but also visits where the source was lost along the way. Treat it as partly unknown, not as proof that people already know you.
Is word of mouth impossible to measure?
Software cannot see it, but you can count it by asking. Record referrals as a source like any other, and note who made the referral where you can. It can turn out to be a bigger share of your work than analytics reports suggest.
Do we need to pay for attribution software?
Usually not at first. Standard analytics, a record of enquiries and the habit of asking will answer most questions a small business has. Paid tools become worth considering when you run several paid channels with a meaningful budget and sales take a long time to close.
What to do next
Add 'how did you hear about us?' to your enquiry form and your phone script this week, with a short fixed list of answers. After a few months, compare the answers with what your analytics says. The differences show where your marketing is being under-credited or over-credited.
If you would like someone to look at your set-up, ask for a free 24-hour audit. A founder records a personal video walkthrough of your website within 24 hours, which can include what the site records about enquiries and where the gaps are. No call is needed.
Want this applied to your website?
George or Nicholas will record a personal walkthrough of your site within 24 hours.
Not ready for an audit? See Tracking and analytics.
