Marketing attribution is the method you use to decide which ads, emails or posts get the credit for a sale. It matters because the answer decides where next month’s budget goes. Most founders run into it when two ad platforms both claim the same customer and the totals add up to more sales than the business made. This guide explains how attribution works, the main models, and a practical setup for a small team.
Why attribution is hard
Customers rarely buy after a single touch. Take a hypothetical customer, Priya. She sees an Instagram ad for a Mumbai yoga studio and scrolls past. A week later she searches the studio’s name on Google, reads a blog post, joins the email list, and books a class after the third email.
Which channel earned that booking? Each platform sees only its own part of the journey, and each will count the booking as its own if its rules allow.
The main marketing attribution models
An attribution model is the rule for sharing out credit between the touches before a sale.
- Last click: the final click before the conversion gets all the credit. For Priya, that is the email. Simple, but it undervalues whatever created the interest in the first place.
- First click: the first touch gets all the credit, here the Instagram ad. It favours channels that introduce people and ignores what closed the sale.
- Linear: credit is split equally across every touch.
- Time decay and position-based: rules that give more weight to recent touches, or to the first and last.
- Data-driven: the platform uses its own conversion data to estimate how much each touch contributed.
In Google’s tools the choice is now narrower. Google removed first click, linear, time decay and position-based attribution from Google Ads and Google Analytics 4 in 2023, leaving data-driven and last click as the options (see Google’s announcement). You will still see the older models discussed, and they remain a useful way to think about the problem.
Attribution windows and double counting
Each platform also uses a window: how long after a click or view it will still claim a conversion. Meta’s default has been 7 days after a click or 1 day after a view, and you can change it at ad set level. See Meta’s help page on attribution settings and check the options in your own account, as they do change.
A view-through conversion means someone saw the ad, did not click, and bought later. The ad may have helped, or the person may have bought anyway. Windows explain much of the double counting founders see. If a customer clicked a Google ad on Monday and scrolled past a Meta ad on Tuesday before buying, both reports can count the same sale.
A practical setup for a small business
- Pick one source of truth for sales: your store, CRM or accounts. Platform dashboards are useful signals, and your own records are the final count.
- Tag every link you control with UTM parameters so your analytics can see source and campaign.
- Ask customers. A “How did you hear about us?” field on your form or at checkout, with a free-text option, picks up word of mouth, podcasts and offline sources that no tracking code sees.
- Compare three figures monthly: total new customers, the sum of what each platform claims, and what customers told you. Large gaps show you where reports overlap.
- Use each platform’s own model for day-to-day tuning inside that platform, and your own numbers for deciding how to split budget between platforms.
Testing whether a channel adds sales
Attribution cannot fully answer one question: would this sale have happened without the ad? That is called incrementality, and you test it by switching off or sharply cutting a channel in one city or for a few weeks while watching total sales, not platform-reported sales.
Say a Delhi furniture store pauses its branded search ads, the ones that show when people search the store’s own name, for four weeks. If enquiries hold steady, those ads were mostly collecting credit for customers who would have found the store anyway. The store can move that budget to campaigns that reach new people.
Run one test at a time, keep each long enough to cover normal week-to-week swings, and write down what you expect before you start, so the result cannot be explained away afterwards.
Living with imperfect data
You will never get perfect attribution, and chasing it can cost more than it saves. Aim for numbers good enough to make the next decision, such as which channel to grow and which to cut, and review them every quarter.
Setting up tracking and a monthly report your team can trust is part of our digital marketing work. Connecting your CRM, store and ad platforms so the data flows without manual copying is the kind of job we handle under automation.
If your ad reports never match your sales, we can look at why on a short call.
Frequently asked questions
Which attribution model should a small business use?
Inside Google Ads or GA4, data-driven attribution is fine if you have enough conversions for it to learn from. Otherwise last click is easier to understand. More important than the model is a single source of truth for actual sales and a “How did you hear about us?” question. Use platform models to tune campaigns and your own records to set budgets.
Why do Google and Meta both claim the same sale?
Each platform counts conversions it can link to its own ads within its attribution window, and neither can see the other’s clicks or views. If a customer clicked a Google ad and also saw a Meta ad before buying, both may report the sale. That is why the sum of platform-reported conversions is often higher than your real total.
What is incrementality in marketing?
Incrementality is the extra sales a channel causes that would not have happened without it. Attribution shares credit among the touches before a sale, while incrementality asks whether a touch made any difference at all. You measure it with tests, such as pausing ads in one region and comparing total sales with a similar region where the ads kept running.