A tenth of the volume is not the same as a tenth of the value.

A channel producing twenty leads a month cannot be judged on a month. One large sale or one bad week moves the number further than any real change in performance would.

That is the honest reason small channels get cut: not that they were bad, but that nobody ever had enough data to say otherwise, and cutting is the default.

Taking six months or a year fixes it without changing anything else, and the report states how many leads each figure rests on so a number built on nine of them is visibly that.

Where the count is genuinely too small to read, the report says so rather than printing a ratio that looks like a finding. A figure built on nine leads is not a small finding; it is not a finding.

Search on the owned properties and search on partner sites are not the same inventory, and in some accounts the difference in lead quality is large.

Because both report inside one campaign structure by default, the blended figure can hide a partner network producing volume that never closes.

Where the export distinguishes them, the report splits revenue on it. Where it does not, the difference often shows up as a bimodal pattern in the same campaign.

This is the specific check worth running before concluding the channel does not work, because it is the one most likely to be the actual explanation.

Set beside a search channel ten times its size, this one loses every absolute number: fewer impressions, fewer clicks, fewer leads, less revenue.

None of those answer whether the next thousand dollars is better spent here or there, which is the only question a budget meeting is actually asking.

Revenue per lead and revenue per dollar spent do answer it, and in a fair number of accounts this channel wins on both while losing on all the absolute measures.

Nothing about this is a claim that it will. It is a claim that the comparison normally made cannot tell you either way.

It will not assign a sale to a campaign where nothing in the file connects them. Unmatched revenue goes to Direct / Unknown and stays visible.

It will not manufacture significance from a handful of leads. A figure resting on a small count is labelled as such.

It will not connect to your account or write anything back. Two exported files are the whole mechanism.

What you get is a defensible number, built from rows you can open, for a channel that is usually cut on a number nobody checked.