Email is last to touch almost everybody, which is not the same as bringing them.

Marketing automation talks to people who have already given you an address, which means it touches nearly everybody on their way to buying.

Any model that credits the last interaction therefore ranks email first, in almost every account, regardless of whether it changed a single decision.

That is not the tool being dishonest; it is a last-touch model applied to a channel that is structurally last. The fix is to rank on where the contact came from originally.

Doing that usually moves a large share of revenue off email and onto whatever actually produced the contact — search, an event, a referral, a paid campaign.

It also makes email's real contribution measurable, which is a question about what it does to conversion rate and time to close rather than about who to credit.

Once email stops competing for acquisition credit, the interesting question becomes what the automations are doing to people who were already in the file.

That is a retention and velocity question: do contacts on a sequence close at a higher rate, and do they close faster, than comparable ones who are not.

Both are answerable from the same two exports by splitting contacts on whether they entered the sequence, and neither requires attributing revenue to a send.

The result is a defensible number for the programme without pretending an email produced a customer that search produced.

It is also more useful operationally, because it points at which sequence to change rather than at a budget line.

Holding contacts and deals in the same place means the export is clean and the match rate is high, which is a genuine advantage over most of this catalogue.

What it does not do is verify the source field, which is still populated by a form mapping written two years ago, an import, or whoever created the record.

Bringing in a separate lead file — call tracking, an advertising platform's lead export — gives a second, independent view, and the disagreement between them is the finding.

The report shows where the two differ rather than preferring one, because a source field that has quietly been wrong since a form was rebuilt is a common and invisible failure.

A won deal carries whatever amount was entered, at whatever moment somebody moved the card, and that is frequently the list price rather than what was paid.

Where the business invoices from another system, exporting revenue from there instead turns the deal value back into a forecast and the ranking into money.

The gap between the two is not spread evenly across sources: channels that produce price-sensitive buyers show the widest one.

Where only the CRM figure exists, the report uses it and labels it as such rather than implying a precision the data does not have.