Video attribution: which videos actually earned the money

Most stores measure video with plays, because plays are the number the player hands you. It is close to meaningless, and it is worth understanding why before building any opinion on top of it.

Why plays are not the number

Video autoplays. A play is recorded when a card scrolls into view, so the play count is a measure of how many people scrolled past the section. It goes up when your traffic goes up and down when your traffic goes down, and it is roughly independent of whether the video did anything.

Two videos with ten thousand plays each can have completely different commercial value, and the play count cannot tell them apart.

What attribution does instead

It follows the chain. A shopper opens a video, taps a tagged product, adds it to the cart, and eventually checks out. Each step is tied back to the video that started it, so the order can be credited to the specific thing that caused it.

The chain that matters is: views → interactions → add-to-carts → orders → revenue. Attribution is what connects the last two to the first.

Why the answer surprises people

Consistently, the video everybody in the business likes is not the one earning. The favourite is usually the most produced one — the brand film, the campaign hero — and it plays a lot and sells little.

The earner is frequently something unglamorous: a badly lit clip of someone demonstrating the thing that is hard to explain in writing. It plays less, it converts far better, and nobody would have picked it from a list.

That gap is the entire value of measuring. Without it you are merchandising on taste, and taste is systematically wrong about this.

The honest limits

Attribution windows are a choice, not a fact

A shopper who watches a video today and buys in three weeks may or may not be credited depending on the window. There is no correct answer — just be consistent, so your comparisons are comparing the same thing.

It is not the only cause

A shopper who saw an ad, read a review, and then watched your video did not buy because of the video alone. Attribution credits the last touch it can see, which overstates the video’s role, in the same way every other channel’s reporting overstates its own.

Cross-device is imperfect

Phone at lunchtime, laptop at night, is a genuinely hard case. Some of it is recoverable and some is not, on every platform.

How to use it without over-reading it

Use it comparatively rather than absolutely. Do not ask “did video generate exactly £4,200”; ask “which five videos generated the most, and what do they have in common”.

The first question has a precise-looking answer that is wrong at the edges. The second has a slightly fuzzy answer that tells you what to make next, which is the decision you actually have.