Every store owner runs the same quiet experiment: try a bit of everything — posts, emails, a reel, an influencer, some ads — and hope the orders that follow explain themselves. They never do. Orders arrive as orders, unlabelled, and at month's end the only honest answer to "what actually made us money?" is a shrug. That shrug is expensive. This post is about replacing it with a number — attribution for stores that do not have a data team.
Why "I think it's working" costs real money
Without attribution, budgets follow feelings, and feelings follow whatever was most recently visible. The reel with the most likes gets more effort than the unglamorous email flow quietly producing a third of revenue. Worse, every tool in your stack claims credit generously for the same orders — the email app, the ads platform and the review widget will happily report, between them, 250% of your actual sales. Marking their own homework is the business model. Problem 27 on our checklist — "I can't tell what's bringing in sales" — is not a data problem; it is a decision problem wearing a data costume.
The 20% of attribution that answers 80% of the question
You do not need enterprise modelling. You need three habits, applied consistently.
Tag every link you control. Every email, every social post, every DM link, every blog article carries a small tracking tag saying where it came from. Costs nothing, changes everything — this is how an order from your Tuesday newsletter identifies itself as one.
Read at order level, not platform level. The question is never "how did the campaign do in the tool's dashboard?" but "which orders in the store began at which touch?" When credit is anchored to actual orders, the double-counting problem largely dissolves — an order counted once cannot flatter three tools.
One report, not five. The numbers must land in a single view: revenue by channel, by flow, by post, side by side (problem 28 — "five apps that don't agree"). Not because dashboards are exciting, but because comparison is the entire point of measuring.
For the claims that matter, demand the holdout
Tagging shows correlation; for expensive questions — does the AI assistant actually lift sales? does that subscription tool pay for itself? — the honest instrument is a holdout: give the feature to most visitors, withhold it from a random slice, compare revenue. It is the same logic as a clinical trial, it runs quietly in the background, and it is the difference between believing and knowing. Any vendor unwilling to be measured this way is telling you something.
What changes when you can see
Stores that get attribution working have the same experience: one channel is quietly carrying far more than anyone credited (embarrassingly often it is the lifecycle emails), one beloved activity is producing almost nothing, and the next month's effort reallocates itself without an argument. The numbers do the managing.
We feel strongly about this because we priced our own service on it: a share of attributed sales, so we only earn when the tracking proves we did — every automated touch tagged, every order traced, one dashboard, holdouts on the big claims. The full architecture is in the complete guide, and the dashboard itself is on the service page. However you build yours: stop taking your own marketing's word for it. Make it show receipts.