Omnichannel Attribution: One Sale, Every Channel Claiming It

Run five channels and you have five systems each reporting on the same orders under its own rule, its own attribution window and its own definition of a conversion. Add up what they claim and the total exceeds what your store shipped. That excess is not a bug in any one platform. It is the predictable result of asking five self-interested bookkeepers to each count the same money, and its size is the single most useful number in omnichannel reporting.

The arithmetic that makes it visible

Take one month. Sum the conversions claimed by every paid platform, plus email and SMS. Divide by the orders your store actually shipped in the same window.

That is the claim ratio. At 1.0 the claims exactly equal the orders. At 1.6, the platforms collectively claim 1.6 orders for every one that happened, and at least 37% of those claims cannot each correspond to a distinct sale. The excess has to be double counting, orders claimed outside the window in which they occurred, or claims on sales that never existed.

The one-hour claim ratio audit is the procedure. It uses data you already hold and needs no new tooling.

Why each platform is individually defensible

None of them is lying. Meta counts orders it can associate with an ad impression or click inside its window. Google counts orders it can associate with its own inventory inside a different window. Your email tool counts orders from recipients who opened or clicked within its own lookback. Each is answering the question "could this order be associated with me", and for a customer who met several channels the honest answer is yes for several of them.

The error appears when someone sums the answers and treats the result as a description of distinct sales. Nobody built the numbers for that, and the platforms are not obliged to reconcile with each other.

What omnichannel makes worse

Two things specifically. Windows differ, so the same order can fall inside one platform's lookback and outside another's, which means the overlap changes when you change the date range rather than when the business changes. And offline channels, retail, wholesale and marketplace, are usually absent from the digital claims entirely while their orders sit in the store total, which pushes the ratio down and can mask overlap that is genuinely there.

If you sell through several routes, compute the ratio against the orders those channels could plausibly have influenced, rather than against every order in the business.

The question none of the five answers

Which channels caused the revenue. Every platform reports association, and association is what a channel can demonstrate about itself. The counterfactual, what would have happened without the channel, requires a design that observes the absence, and no platform's reporting contains the journeys in which that platform did not appear.

For the channel carrying the most budget, a holdout answers it. For everything between holdouts, a causal read estimates it from the variation already in a 40 to 90 day export and reports an interval on each estimate, with the coverage of the data stated. Incremental ROAS per channel without a geo test sets out what that can and cannot say.

What to do this week

• If you own the budget: compute the claim ratio for last month before you sign the next plan. If it exceeds one, the channel table you are allocating from is arguing with itself.
• If you have to defend the number: anchor every channel figure to store orders and show the claim ratio beside the table. It converts an argument about whose dashboard is right into an arithmetic everyone can check.

Omnichannel attribution with Klaviyo in the stack covers the email and SMS side, which is usually the largest single omission. The interactive demo shows a causal read on a sample store, no signup.

Product facts as stated on causalityengine.ai on 9 September 2026. The three-systems identity and the claim ratio are from The Price of Being Found (Edition 2.10), Chapter 9, with the book's caveats.