1,540 calls. 378 conversations. Fourteen checks.
We've graded 1,848 recorded calls against the rubrics their own coaches use. This is what two campaigns at one company looked like when somebody finally counted all of them instead of the ones that were easy to find.
Most of it isn't a call.
Two campaigns at one company, same nine-beat rubric. 1,540 recorded calls between them. We graded every one.
378 were conversations. The other 1,162 were voicemail, wrong numbers, and the eight-second version where somebody says no thanks before the rep finishes the sentence. Nothing to coach in those, and no way to know which is which without listening.
Calls that became an actual conversation — roughly one in four. This is the number that makes call coaching expensive: three quarters of the listening buys you nothing, and you can't tell in advance which quarter matters.
This is also why sampling fails. Pull ten calls at random on a Friday and seven of them are a dial tone.
Nobody checks who they're talking to.
In those 378 conversations, the rubric asks a simple thing: did the rep establish that this person can actually buy, or get them to whoever can?
Calls where the rep confirmed they were talking to someone with authority to buy. Under four percent. On one campaign it was 2 out of 162.
They open well. Permission to keep talking gets asked on 81% of conversations, which is genuinely good and better than most teams manage. Then the call goes forward for several minutes without anybody establishing whether the person on the other end can say yes to anything.
That's not a motivation problem or a script problem. It's the single most expensive habit on the board, because a meeting booked with the wrong person costs an AE an hour and a pipeline entry that was never real.
Same company. Sixty times apart.
A failsafe is the thirty seconds at the end of a booked call where the rep makes the meeting survive contact with reality. A direct number to text. A named backup slot. Something more durable than an invite sitting in an inbox.
It only applies to calls where the rep asked for a meeting, so that's the population — 302 of the 378. Split by campaign, it stops being one number.
| Campaign A | Campaign B | |
|---|---|---|
| Calls graded | 1,220 | 320 |
| Rep asked for the meeting | 268 | 34 |
| Of those, set a failsafe | 2 | 15 |
| Rate | 0.7% | 44% |
Campaign A asks for the meeting eight times as often and protects it almost never. Campaign B asks less and locks down nearly half of what it gets. Neither team knew this about itself.
That's the part worth stealing if these were your numbers. Not the average across both — the sixty-fold gap between two groups running the same playbook, which is invisible until somebody counts every call instead of the ones that were easy to find.
The rubric comes from their coaches.
There's no generic call score here, and we didn't invent the standard. Before a campaign starts we ask whoever coaches that team what they already listen for — what the call has to accomplish, which beats have to land, what counts as landing one. Nine beats, in this case. Then every recorded call gets graded against those beats, one at a time.
Every call our clients have recorded and we have graded against their own coaches' rubric, as of August 2026. Not sampled. Not spot-checked. All of them.
- The two campaigns above share all nine beats, which is why their numbers can sit together. Our other clients have their own rubrics, so folding them in would have bought a bigger sample by comparing different questions.
- Every rate here is drawn from the calls where the behaviour was possible. A rep who never asked for a meeting can't fail to protect one.
- Nothing names a company, a rep, or a person who was called.
We'd rather publish a small number we can stand behind than a big one we can't. If you want to know how we got to a figure on this page, ask and we'll show you.
Go and count your own.
Take the last twenty conversations your team had. Not twenty calls — twenty where somebody actually talked back. Count how many established that the person on the other end could buy.
If it's two or three, you've found what we found, and it's the cheapest thing on your board to fix. If it's most of them, your team is doing something 96% of these 378 conversations didn't, and you should know which rep started it.
Twenty conversations is maybe an hour. Finding those twenty inside a week of dials is the part that isn't an hour, and it's the reason we exist.
We do this for every call.
Same idea, at the size your team actually runs at. Bring us a week of recordings and we'll tell you which beat yours is missing.