View-through conversion

Here is the sequence that causes most of the confusion, and it does not involve a click anywhere.

Someone is served your display ad and ignores it. Days later they search for you, arrive through organic search, and buy. Your analytics records an organic search conversion. The ad platform records a view-through conversion. One sale, two systems, two entirely different channels credited.

Why they disagree, and why neither is lying

The ad platform served the impression, so it knows about it. Your analytics never received that impression at all — no request was made to your site, nothing was recorded, the visit did not happen yet. From your analytics’ point of view the customer’s first contact with you was the organic search.

So the two tools are not applying different attribution models to the same data. They are working from different data. The ad platform holds an event your analytics has no way of knowing about, and it credits itself with it.

That framing matters because the usual response is to go looking for a tracking bug. There is no bug to find. The gap is structural, and no amount of tag debugging closes it.

What follows from that

The same sale is counted in two places. Add channel totals across your analytics and your ad platforms and you will exceed real revenue. This is the single most common cause, and it is not deduplication failing — dedup only works inside one receiving system. Two vendors both counting the same conversion never see each other, and no shared event ID fixes it.

Organic looks better than it is, and display looks better than it is, simultaneously. Both reports are internally consistent. Both are claiming the same revenue.

Buying more impressions raises view-through counts on its own. Show enough ads to enough people and some of them will convert for unrelated reasons. Every one of those becomes a view-through. The metric rises with spend whether or not the advertising changed anyone’s mind — which is exactly the property you do not want in a number you use to justify spend.

Reading them honestly

Check the window. View-through windows are usually much shorter than click windows for good reason. If yours is long, the channel is claiming credit for things that happened days after a glance.

Notice who is counting. The party crediting the view-through is the party selling the impressions.

Never sum across systems. Pick one system as the reporting source of truth and use the others for diagnosis. Reconciling them line by line is not a solvable problem.

Treat it as directional, never as a line item you would defend to finance. The way to find out whether impression-based advertising works is to stop serving it to a comparable group and see what happens — that is incrementality, and no view-through report answers it.