What to judge a setter on, and what to leave alone
The first setter page we built for a phone team ranked its busiest dialer as "Unknown user", followed by a string of letters. That person had made more calls than anyone else, and the page said nobody had.
The cause was ordinary. The id belonged to a user the CRM had since deleted. The call log still held every dial they made, tagged with that id, but a lookup of it now answered "not found". Our page asked the CRM who each id belonged to at the moment it drew the chart. For the hardest worker on the team, the CRM no longer had an answer.
A setter page is read by people deciding who stays, who gets the better leads and who gets coached. If the top row is a ghost, nothing below it can be trusted, so this post covers the numbers we put on that page, what each one can tell you, and what it cannot.
Why a wrong setter number costs money
Setters feed the closers' calendar. If the number you use to judge them is off, you move the wrong people. A setter on a cold list can look weak for a month and be doing the job well. A setter with a warm list can look strong and be coasting. The first mistake loses a good setter. The second keeps a weak one and hands them the best leads.
The numbers also decide how the lead pool gets shared. An owner who ranks setters by one raw count will give the best leads to whoever tops that count.
What the tools give you
Most teams start from the call log in their CRM or dialer. HighLevel's help center is exact about what that log means: a call status is "a system-recorded result based on what happened technically during the call", while a disposition is "a user-defined classification selected after the call to describe its business outcome" (HighLevel help center). The status tells you the phone connected. It does not tell you a person talked to a person. In the logs we pulled, a completed call included voicemail, and the outbound call record had no field that marked a voicemail or a machine. We checked that field by field before we trusted anything built on it.
The booking side is better, if the setup is right. Calendly stores the five standard UTM values with the event, and you can read them on the meeting, in webhooks and in the API; it also says a value over 255 characters is not tracked (Calendly help center). So a booking link that names the setter in a UTM value can credit the setter automatically, as long as the setter used their link.
So the tools give you two things: a call that connected, and a booking that may name its setter. Everything between them is a definition you have to write down.
The four numbers
Dials. This is effort. It is cheap to count and it says nothing about quality, so compare it only between setters working the same kind of list.
Conversations. We define a conversation as a connected call of at least 60 seconds. The floor is a decision, and we set it by looking at how long the connected calls ran. More than half of them were over in under ten seconds, and another cluster ran 35 to 50 seconds, which looks like voicemail greetings. The floor is a setting in the model, not a fact about phones, and the dashboard says what it is. We also keep a second floor at five minutes for calls long enough to be a real pitch. In that data, nearly every booking came from calls that ran that long.
Bookings. A booking needs one setter's name on it, and the rules for that are where the mistakes hide. The setter named in the booking link gets the credit. A reschedule keeps the original setter, because the person who got the person on the calendar did the work once. When a booking carries no setter at all, the fallback is the setter whose completed call to that phone number came last before the booking. A booking with no setter in its link is common enough that we print, on the page, how many bookings have no setter and how many have a link and a call history that disagree. Nobody then mistakes the gap for a setter's weak week.
Shown. We count it, with care. A person turning up depends on the reminder, the closer and the lead as much as on the setter. What a setter does control is whether a bad booking gets onto the calendar. When a setter cancels a booking they judge to be no fit, that call is neither a show nor a no-show, and the rules in the show rate post leave it out of the rate. A setter who clears bad bookings should not be punished for it, so we ask for one line in the setter SOP: if you cancel it, say why.
Two ratios, with the counts beside them
The setter page shows two ratios per setter: dial to conversation, and conversation to booking. They separate two jobs. The first is mostly the list, the hours and the dialer: who answers. The second is the setter on the phone: what happens once someone does.
Every ratio is printed next to its counts, because a week is short. A setter with eight conversations who books two is at 25%. One more booking moves it to 38%. Nobody should be reviewed on that.
A worked example
This is a made-up example. It names no client, and it exists to show the arithmetic.
Two setters work for two weeks.
| Dials | Conversations | Booked | Conversation to booking | |
|---|---|---|---|---|
| Setter A, week 1 | 400 | 40 | 4 | 10% |
| Setter B, week 1 | 150 | 30 | 6 | 20% |
| Setter A, week 2 | 400 | 40 | 8 | 20% |
| Setter B, week 2 | 150 | 30 | 3 | 10% |
| Setter A, both weeks | 800 | 80 | 12 | 15% |
| Setter B, both weeks | 300 | 60 | 9 | 15% |
Read week 1 alone and B is twice as good. Read week 2 alone and A is. Over both weeks the two sit level at 15%. What stays different is the step before: B reaches a conversation on 20% of dials and A on 10%. That is a question about their lists and their hours before it is a question about skill.
A review that used one week would have coached the wrong person twice.
What not to judge a setter on
Closed deals and cash. They arrive weeks after the setter's work, they depend on the closer and the offer, and there are too few of them in a month to rank anyone by. A setter's page should end at a call that happened.
Dials alone. It is the easiest number to move and the least informative.
One week's ratio. Use four weeks, with the counts shown. That is our rule of thumb, not a law, and a team with few dials needs a longer window.
A show rate without the setter cancellations taken out. It charges setters for the good work of removing bad bookings.
What to do this week
- List every user id in your dialer's call log and match each to a person with a start and end date. Count the dials from ids you cannot name.
- Write down your conversation floor in seconds and put it under the number on the page.
- Decide how a booking is credited when the link has no setter name, and count the bookings that fall into that gap.
- Print the counts beside every ratio, and review setters on four weeks.
- Add one line to the setter SOP: if you cancel a booking, say why.
If you want to see which of these you can answer today, the Revenue Blind-Spot Score takes about a minute.
Step 1 is where we started. The busiest dialer on that first page was a real person with a real record, and the fix was a roster with history in it, so the name stays attached to the dials after the login is gone.