Marketing attribution: how to know which channels actually produce revenue

Most marketing dashboards give all the credit for a sale to whatever the buyer clicked last. In businesses where customers research for days or weeks, that picture is badly incomplete. A customer might discover you through a LinkedIn post, check your reputation on Google, read two articles over three weeks and finally type your web address directly. Last-click reporting credits only the direct visit, and the channels that actually won the customer look worthless.

This guide explains how attribution works, what has changed in Google’s tools, and how to connect marketing to the revenue your business actually collects.

What is marketing attribution?

Marketing attribution is the method of assigning credit for a conversion or sale to the marketing touchpoints that led to it, such as an ad, a search, an email or a social post. It matters because the channels that get credit get the budget. Wrong attribution moves money away from what works.

Which attribution models exist?

Why does last-click attribution mislead?

• It over-credits branded search and direct visits, which often happen after a customer has already decided.
• It under-credits awareness channels such as social media, video and content, which introduce customers who convert later.
• It ignores offline steps: phone calls, WhatsApp chats and sales meetings that happen after the click.
• It counts conversions, not revenue: a lead that never paid looks as valuable as a signed contract.

What is closed-loop attribution?

Closed-loop attribution follows a customer from the first marketing touch all the way to revenue collected, and sends that result back to the marketing platforms. It is the difference between knowing which ads produced form fills and knowing which ads produced paying customers.

1. Tag every campaign link with UTM parameters, and capture click IDs such as Google’s gclid and Meta’s fbclid.
2. Store those values with the lead in your CRM when someone submits a form, calls or messages on WhatsApp.
3. Record the outcome of each deal: qualified, won, lost and revenue collected.
4. Send won deals back to the ad platforms, using offline conversion imports in Google Ads and Meta’s Conversions API, so their bidding learns from real customers.
5. Report cost per customer and return on ad spend by channel and campaign, based on revenue rather than clicks.

How do you check attribution against reality?

Every model is an estimate. Two simple checks keep it honest:

• Ask customers. A "How did you hear about us?" question on forms and in sales calls captures influences tracking cannot see, such as word of mouth, podcasts or a billboard.
• Run a holdout test. Pause or reduce a channel in one region or time period while keeping everything else the same, and compare results with a similar region or period. If sales barely change, the channel was getting more credit than it earned.

Larger advertisers can add marketing mix modelling, which estimates each channel’s contribution from spend and sales over time. Open-source tools such as Meta’s Robyn and Google’s Meridian have made it more accessible, though it still needs a long, clean history of spend and sales.

What do you need to get started?

• Consistent UTM tagging on every campaign link.
• A CRM that stores the source of every lead and the outcome of every deal.
• Tracking for calls and WhatsApp, which are a large share of leads for many South African businesses. See our guide to WhatsApp marketing.
• Conversion tracking that counts only real leads, as covered in our Google Ads guide.
• A regular review that compares channel cost with revenue, not just leads.

Bringing marketing, CRM and financial data into one place is the core of what we describe in what a growth intelligence layer does. At Lulonke, this work starts with a diagnostic audit, and the initial architecture is typically R65,000 to R220,000. See marketing intelligence for how we approach it.