Every Platform Took Credit for the Same Sale

Picture a Monday morning review.

Meta says last week's campaigns drove 120 sales. Google says 90. TikTok says 40. Snapchat would like some of the credit too.

The shop sold 150 things.

Nobody in the room is lying, exactly. Every platform is reporting what it believes it caused. The trouble is that they all believe they caused the same sales.

I spent years with that problem, running paid campaigns across Google, Meta, TikTok and Snapchat on budgets in the hundreds of thousands of riyals a month. Sooner or later every review came down to the same argument: whose number do we believe?

Four reports, one sale

Each platform counts a sale as its own if its ad touched the customer somewhere along the way. A click, sometimes just a view, inside a window the platform chose for itself.

Follow one customer. They see a TikTok video on Saturday. On Sunday they search for the product on Google and click an ad. On Monday an Instagram ad reminds them, and they buy.

TikTok counts the sale. Google counts it. Meta counts it. One sale, three reports, three teams feeling good about their week.

None of them is technically wrong. Each one is answering a narrower question than the one you have. They're telling you "my ad was somewhere near this sale". You want to know "what would I lose if I switched this off?"

The part nobody says out loud

The platform measuring your results is also the one selling you the space. I don't think that makes anyone dishonest. But it does mean every default leans the same way: generous windows, generous definitions, generous credit.

You'd never let a supplier audit their own invoices. With ad platforms we do it every week and call it reporting.

So I stopped asking them

Eventually I got tired of arguing with dashboards and built my own attribution. Tracking at the code level, on the site itself, following a visitor across sessions and channels instead of trusting whatever the last platform reported.

Most of that kind of work is plumbing. Every visit, every step and every order recorded once, with a time, and tied to the same person even when they come back on a different device. The modelling on top is the easy part. If the plumbing is wrong, no model will save you.

It worked better than anything I'd been using. And because every order is counted once, the credit finally adds up to the sales you actually made.

What changes when the numbers add up

The conversation changes. You stop arguing about whose dashboard is right and start arguing about decisions. Which channel brings people in? Which one only turns up at the end to collect the credit? What happens to sales if this budget is halved for two weeks?

That last question matters most. The only honest test of whether a channel works is to change it and watch what happens to the total. Everything else is inference.

What it taught me

Building it changed how I think about marketing in general. Once you've watched four confident reports describe one sale, you get suspicious of any number that arrives already interpreted. You start asking who produced it, and what they'd lose if it were smaller.

It's the same gap I keep running into, in marketing and outside it: the distance between what you can measure and what you get to claim.

A platform can measure that its ad was shown before a sale. It can't claim it caused the sale.

Neither can you, until you've checked.