The Klaviyo attribution window, and what it is actually counting

The window is a rule about credit

An attribution window is how long a message keeps the credit. Somebody opens or clicks, and any order they place inside that window counts as that message's work. Outside it, the same sale belongs to nobody in particular.

That's a rule about credit, not a discovery about what caused the sale. Klaviyo publishes its defaults and how to change them in its help centre, and the default is shorter for text messages than for email because the two behave differently.

One screen decides every number in the account

Those controls sit above the figures and govern all of them. Change one and every number underneath moves. Including the ones already sitting in a slide somebody sent you last quarter.

• The conversion metric, which decides what counts as a sale at all
• The window itself, set per channel, so email and text messages can differ
• Whether the change re-reads history, because most reporting recalculates the past instead of starting fresh
• The period on the report, which isn't the window and gets confused with it constantly

None of it is buried. Those dropdowns sit on the screen every report is run from, which is what makes asking an agency which window it used such a fair question. And such a telling pause.

The conversion metric matters as much as the window

Every account picks an event that counts as the sale. On a Shopify shop that's normally the placed order. It doesn't have to be, and an account measuring started checkouts will report numbers that look wonderful and mean something else entirely.

Four settings, and any one of them will make two honest reports disagree. None of them is hidden. None of them is a trick.

Why Klaviyo and Shopify never agree, and never will

Klaviyo credits a sale to a message somebody engaged with. Shopify and most analytics tools credit the last click before the order. Two different questions, so two different answers, and no amount of matching settings will make those lines meet.

So pick one as the number you run the business on and keep the other as a cross-check. We use the shop's own total as the denominator for exactly this reason: whatever a platform credits itself, the share of real revenue is a figure both tools can be held to.

What a report has to say before the number means anything

A revenue figure on its own is a screenshot. Put the settings and the store total beside it and you've got a report. One extra line.

• The window and the conversion metric the report was run on
• The exact dates, not a month name
• What the shop turned over in the same period
• Which channels are inside the total
• What else changed in the period that wasn't email

Write those five lines once and they turn into a habit. The first month it feels pedantic. The month somebody asks why a number moved, it's the reason you can answer in a sentence instead of a week.

How to compare two sets of numbers honestly

A longer window isn't cheating, as long as you say so. It's a fair choice for a considered purchase where people take a week to make their mind up. Changing it halfway through an engagement and then reporting growth is a different matter, and it's the one thing here worth being annoyed about.

So when two agencies put numbers in front of you, ask both for the window, the metric and the dates, then read the shares instead of the totals. Ours are on the Shadow Foam case study with the period printed beside them, and the terms are settled on the glossary if any of this is new.

One last habit worth building. When the attributed figure jumps and nothing else in the account changed, check the settings before you celebrate. Somebody widening a window is a far commoner explanation than a sudden improvement in the emails.