Closed Deal Attribution: Who Gets Credit When the Deal Closes

Two teams look at the same closed-won deal and reach different conclusions about who won it. Marketing says it sourced the deal. Sales says it closed one it already had. Nobody is lying, and both are reading the same CRM.

That argument is what closed deal attribution is for. It works backwards from revenue that has already landed, rather than forwards from a click, and it settles a question the touch-weighting models never quite answer: of the money we booked this quarter, how much can marketing put its name on?

Most writing on attribution covers the other half of the subject. Which ad, which email, which webinar earned credit for a conversion, and how to split it. Those are real questions and we have written a broader guide to marketing attribution that maps them out. This piece is about what happens after the deal is signed, when a finance person is in the room and next year's budget is being set.

What closed deal attribution answers

Touch attribution divides credit for a single conversion. Closed deal attribution starts at closed-won and asks which marketing activity can defensibly claim a share of that revenue. Same subject from a different starting point, and that difference matters more than it sounds.

The reversal changes what data you need. Forward-looking models care about the sequence of touches and the weighting between them. Working backwards, the sequence matters less than whether the deal record can be joined to a contact record at all, and whether that contact has any marketing history worth reading. Plenty of CRMs fail at the join before anyone gets to argue about weighting.

It also changes who is in the conversation. A first-touch-versus-last-touch debate stays inside marketing. Closed deal attribution gets discussed in front of the person deciding whether to fund the team next year, which is why the definitions need to be agreed in advance, before anyone has seen how the number lands.

Marketing-sourced and marketing-influenced

Two definitions do most of the work, and companies that never pin them down end up arguing about arithmetic when the real disagreement is about vocabulary.

Marketing-sourced means the deal exists because marketing produced the record. Someone filled in a form or arrived from a campaign, and that was the first the company had ever heard of them. Sales opened a deal against a contact that marketing put in the database.

Marketing-influenced means marketing touched the deal somewhere before it closed. The buyer might have been in the database for two years, or a rep found them at a conference, but on the way to signing they read a case study and sat through a webinar.

The two numbers will disagree, and the gap between them is usually large. That gap is what the two definitions are built to measure, so it is not a fault in the system. Sourced is small and difficult to argue with. Influenced is much bigger and easy to dismiss, because one opened email does not feel like it earned a share of a six-figure contract, and a sales director who has just closed that contract will say so.

Both numbers are worth having. They answer different questions. Sourced tells you whether marketing can originate demand on its own, which is the question that matters when you are deciding whether to hire another person into the team. Influenced tells you whether marketing is helping deals that sales originated, which is the question that matters when someone proposes cutting the content budget because "sales found all of these anyway."

Building a marketing contribution model

A marketing contribution model is the written agreement about which of those numbers you report, to whom, and what counts. It is not software. It is a handful of decisions someone has to make and everyone has to live with.

Four decisions carry it.

First, the window: how long before the deal opened does a marketing touch still count, and does a touch after it opened count at all? Ninety days is the common answer and there is nothing sacred about it. Pick a number that matches how long your buyers actually take. Most teams never pick one at all and end up with whatever the CRM shipped with.

Then what counts as a touch at all. An email open is not a demo request, and treating them as equivalent is how influenced numbers lose credibility with the people you are showing them to. Decide which activities are substantial enough to count, and expect the strict version to produce a smaller and more defensible figure.

Someone has to own the report and answer questions about it. When marketing produces one attribution figure and sales produces another, there is no agreed source to settle it against, so the argument runs every quarter. Same failure as any other system nobody owns.

And once the definitions are agreed they hold for the year. Change the window mid-year because a quarter looked thin and you lose the only thing the model was built to give you: a comparison you can trust over time.

We have watched this go wrong at group level, where two markets reported the same metric under slightly different rules. The consolidated number meant nothing, and nobody noticed until someone made a decision on it.

Where the model breaks

Some of this is fixable and some of it you plan around.

Deals that predate the CRM have no marketing history because there was nowhere to record it. They will show as sales-sourced forever. Exclude them from the comparison rather than letting them drag the ratio.

Reps who create contacts by hand break sourcing without anyone noticing. A rep who meets someone at an event, then types them in, has just recorded a sales-sourced contact even if that person turned up at the event because of a campaign. There is no clean technical fix. The realistic answer is a lead source field the rep is required to complete, and a manager who reads it.

Shared inboxes and generic addresses collapse several real people into one record, and the marketing history attaches to whichever of them happened to click. On a long buying committee this is common enough to distort the picture.

Reopened and re-won deals get counted twice unless the report handles them, which most default reports do not.

What you end up with is an argument backed by evidence, not a measurement, and it is worth saying so out loud when you present it. A contribution model that is honest about its limits survives a sceptical CFO. One presented as precise will not, because it only takes a single obviously wrong deal for the whole report to lose the room.

Common questions