The chat widget produced four hundred conversations. How many produced money?

A chat widget collects a name and a number from somebody already on your site, which makes it one of the cleanest lead files a business has.

It is also routinely excluded from attribution because it is not a form and not a call, so it appears in neither the web analytics funnel nor the call tracking report.

Exporting conversations and matching them against closed revenue puts the widget in the same comparison as every paid channel, usually for the first time.

The result is frequently that the widget is doing more work than anything paid, because it is catching visitors the paid channels already bought.

Somebody messaging from a map result is choosing between you and the two businesses beside you. Somebody chatting on your site has already arrived.

Pooled into one messaging number, those two intents cancel each other out and the report says nothing useful.

Kept apart, the map channel becomes measurable on revenue, which matters because it is usually the one nobody has ever costed.

It is also the channel most affected by review volume and listing accuracy, which is where the rest of the product earns its keep.

A strong rating changes how many people choose you from a map result, and that effect is real, large and slow.

It is also untraceable at the individual level: no customer in your file can be linked to the review that influenced them, because they never interacted with it in a way anything recorded.

The honest treatment is to measure the map and directory channel accurately on revenue, and read it against review activity over months rather than manufacturing a per-customer link.

Any tool claiming to attribute revenue to a specific review is guessing, and a guess presented as attribution is worse than a stated gap because nothing warns you when it is wrong.

Review counts, response times and conversation volume are reported per location because that is how the software is organised, but revenue rarely is.

Without the revenue side, a location with excellent engagement metrics and poor conversion looks like a success, and the group average hides both.

Where the export carries a location, the report splits revenue on it, and the locations separate immediately on the only measure that matters.

That split also identifies the operational problem this category most often has, which is conversations nobody answered quickly enough at one site while the group average looked fine. That is a staffing decision rather than a marketing one, and it is invisible in every report that stops at the conversation.