Ask where a customer came from and every channel claims them. The ad platform says ads. Analytics says organic search. The customer says “a friend mentioned you.” All three are telling a version of the truth, because real customers touch several channels before buying — and ‘attribution’ is just the rule you choose for sharing the credit.
The rule you choose matters because it steers budget. Credit everything to the last click and you’ll overfund the channels that close (brand search, email) and starve the ones that introduce (SEO content, prospecting ads). This guide explains the main models, the honest limits of all of them, and a practical setup that’s good enough for real decisions.
It assumes you’ve met the six owner\u2019s numbers — attribution is what makes number 5, revenue by channel, honest rather than flattering.
Quick FactsQuick Facts: Attribution
| Detail | Information |
|---|---|
| Provider | Fredeveloper — full-service digital marketing agency |
| Topic | Marketing attribution — how credit for a sale gets assigned to channels |
| Best for | Owners splitting budget across two or more channels |
| What this guide covers | Why attribution is genuinely hard, the main models compared, what each one over- and under-credits, and a pragmatic small-business setup |
| Relevant services | Analytics · Paid ads · SEO |
| Engagement path | Free quote → audit & plan → done-for-you delivery → monthly reporting you can actually read |
| Contact | info@fredeveloper.com · Contact page |
| Last updated | 2 May 2026 |
The ProblemWhy ‘Where Did They Come From?’ Is Hard
A typical buying journey for anything considered — a renovation, an accountant, a £2,000 service — looks like this:
Infographic 01 · One customer, five touches
A realistic journey to a single enquiry
Add the messy reality — people switch from phone to laptop, decline cookies, hear about you offline — and perfect attribution is impossible for any business, at any budget. The goal isn’t perfection; it’s being less wrong than “whatever the ad platform claims.”
The ModelsThe Main Attribution Models, Compared
| Model | Who gets credit | Over-credits | Under-credits |
|---|---|---|---|
| Last click (most common default) | The final touch before converting | Brand search, email, ‘direct’ | Everything that created the demand |
| First click | The touch that introduced you | Awareness channels | Everything that nurtured and closed |
| Linear | Equal share to every touch | Trivial touches | Decisive ones |
| Position-based | Most to first and last, rest shared | — a reasonable compromise — | Middle touches, slightly |
| Data-driven (GA4 default) | Algorithmic share based on patterns | Opaque — hard to audit | Channels with sparse data |
Every ad platform grades its own homework: it claims any sale its ad touched, on its own generous rules, so platform numbers added together routinely exceed your actual revenue. Treat platform-reported conversions as that channel’s best case, and your own analytics as the referee — the ads-report guide shows the gap in practice.
Pragmatic SetupA Good-Enough Setup for a Small Business
You don’t need enterprise attribution software. You need four habits:
- One referee: pick GA4 (or your CRM) as the single source of truth for channel credit — platforms inform, the referee decides.
- Tag everything: every ad, email, and campaign link carries UTM tags so clicks are labelled at the door — ten minutes of discipline that fixes half of all attribution mess.
- Track calls: a tracking number per channel, because untracked phone calls all land in ‘direct’ and silently steal credit from every channel.
- Just ask: add “How did you hear about us?” to every form and sales call, and log it — self-reported attribution is imprecise but catches what software structurally can’t (word of mouth, podcasts, drive-bys).
Infographic 02 · The referee model
Platforms claim — your analytics decides
DecisionsWhat Attribution Should (and Shouldn’t) Decide
- Should: rough budget shifts — if a channel gets meaningful credit under any reasonable model, it’s working; if it gets none under every model, question it.
- Should: defending awareness channels — first-click and position-based views protect the channels that introduce you from last-click’s bias.
- Shouldn’t: precision arguments — debating whether a channel deserves 22% or 26% of credit is false accuracy; no model knows.
- Shouldn’t: killing a channel on one model’s verdict — check two models and the “how did you hear about us” log before any execution.
How We WorkHow Fredeveloper Handles Attribution
We set up the boring parts that make attribution work — GA4 conversion events, UTM discipline, call tracking — and report channel credit under more than one model when the models disagree, telling you so plainly. Budget recommendations come with the confidence level attached: “strong signal, shift money” versus “noisy, hold and watch.” The free quote includes an attribution health-check: where your tracking leaks credit today, and the shortest path to trustworthy numbers.
FAQFrequently Asked Questions
What attribution model should a small business use?
Start with GA4’s default (data-driven) for trends, sanity-check big decisions against last-click and first-click views, and keep a “how did you hear about us” log. Agreement across those three is as close to truth as small-business attribution gets.
Why do my ad platforms report more sales than I actually made?
Each platform claims every sale its ad touched, on rules it sets itself, so overlapping claims double-count. It’s structural, not fraud — but it’s why platform numbers should never be summed, and why your ads report should reconcile against your own analytics.
What does ‘direct’ traffic actually mean?
Honestly: “we don’t know.” Typed-in URLs, untagged links, many phone calls, some app and privacy-stripped clicks. A large ‘direct’ bucket usually signals missing UTM tags and absent call tracking rather than an army of URL-typers.
Does word of mouth show up in attribution?
Not in software — a friend’s recommendation leaves no click. It shows up as brand searches and ‘direct’, which then steal the credit. The “how did you hear about us” question is the only honest detector, which is why we treat it as a tracking tool, not politeness.
Is attribution still possible with cookie restrictions and iOS privacy changes?
Partially — journeys fragment, and models lean more on modelling than observation each year. The practical response is the setup above: first-party data (CRM), tagged links, call tracking, and self-reported source, all of which survive cookie loss.
How often should I revisit channel budgets based on attribution?
Quarterly for real shifts, monthly for monitoring. Channel performance moves slower than dashboards imply, and the six monthly numbers will flag anything urgent in between.
Keep ReadingRelated Guides for Owners
The 6 numbers for your monthly meeting · GA4 for business owners · Google Ads vs SEO
Know Which Pound Is Working
Half of all marketing budget debates are really attribution problems: nobody trusts the numbers, so the loudest opinion wins.
We’ll audit your tracking free — where credit leaks today, which channels are over- and under-counted, and how to fix it.