Both of them. The 400 counts records created, which is checkable. The 12 counts conversations that were worth having, which is a judgement. Nobody has written down the rule that turns one number into the other, so the two teams argue about facts when they are disagreeing about a definition.
This is the oldest argument in B2B and it is getting worse rather than better, because form volume is falling while the share of serious buyers who never fill anything in keeps rising. The two numbers are drifting apart faster than the conversation about them is improving.
What each number actually counts
| Marketing's 400 | Sales' 12 | |
|---|---|---|
| Counts | Records created in the period | Conversations judged worth having |
| Measured | Automatically, completely | By hand, after the fact, inconsistently |
| Verifiable | Yes | No |
| Includes students, competitors, existing customers, bad emails | Yes | No |
| Includes leads nobody called | Yes | No |
| Changes if you change the form | Immediately | Eventually |
The last two rows do the damage. Marketing's number includes every record a rep never opened, and sales' number excludes them, so a routing failure shows up in the report as a lead quality problem. That is the most common misdiagnosis in the whole handoff and it points the fix at the wrong team.
The two hour exercise that settles it
Do not debate the definition in the abstract. Debate it against a list.
Take one month. Put every record in a single spreadsheet with its source. Both teams sit with it and mark each row accepted or rejected, and every rejection carries a one word reason. Not a paragraph, a word.
The reasons will collapse into four or five categories, and in most teams they land roughly like this:
- Not a buyer. Student, researcher, competitor, job seeker. A form problem, fixable.
- Wrong fit. Real company, wrong size or sector or region. A targeting problem.
- Already ours. Existing customer or open deal. A matching problem, and an embarrassing one.
- Too late. Contacted on day four or not at all. A process problem, and not marketing's.
- Genuinely no reason. The rep did not like the look of it. Worth knowing about.
Once the rejections have names, the argument stops being about whose number is honest. It becomes a list of four specific fixes with an owner each, and two of them are usually a half day of work.
The rejection that is nobody's fault and everybody's problem
"Too late" deserves its own treatment, because it is the one category where both numbers are accurate and the lead was still wasted.
Response time research has been consistent for over a decade: the odds of a meaningful conversation fall steeply across the first day and keep falling after that. A record contacted in the first hour and the same record contacted on day four are not the same proposition, and when sales rejects the second one they are not wrong about what they found.
So the acceptance rate on its own is not enough. It has to sit next to the median time from record created to first human contact attempt, which is the subject of what is an MQL worth if sales never calls it. Without that second number you cannot tell a definition problem from a speed problem, and they have opposite remedies.
Why the gap is widening
Three things changed at once and all three push the two numbers further apart.
Form volume is falling. With zero-click search at 75.1% and most evaluation happening before contact, fewer serious buyers identify themselves. The records that still arrive skew towards the people who had a reason to fill in a form that was not buying.
Volume targets survived. The lead target did not fall with the form fills, so the threshold absorbed the difference.
The serious buyer became invisible. The company reading your pricing page three times produces no record at all, so it appears in neither number, which is covered in a company read three pricing pages and nobody called them.
That last one is why the argument is less useful than it used to be. Both teams are now arguing about a list that no longer contains the best opportunities of the month.
What to change
One list, not two systems. The reconciliation exercise is only painful because the campaign, the record, the first call attempt and the deal live in different places. When they sit on one account in Empiraa Signal, acceptance is a field rather than a spreadsheet and a meeting.
Acceptance as a shared number. Not a marketing metric reported to sales. One number both teams are measured on, which is the structural argument in should marketing carry a pipeline number.
A rejection reason that is compulsory. One click, five options. Without it you will run the two hour exercise once, fix four things, and be back here next year.
The uncomfortable version
Sometimes the 12 is also wrong.
If reps are rejecting records they never opened, or cherry-picking on company names they recognise, then their number is a judgement about their own time rather than about the leads. That is worth finding out before the definition gets rewritten around it, and the way to find out is the same exercise: a rejection with no reason attached is a data point about the rep, not about the lead.
Both teams should expect the reconciliation to be slightly unflattering. That is how you know it was honest.
Related reading: what is an MQL worth if sales never calls it, and should marketing carry a pipeline number.
Marketing qualified lead has a definition on this site. See Signal's marketing side, in the same system as the pipeline.
