Skip to main content
Every number on this page is computed directly from message history on live client accounts — not surveys, not self-reported estimates. Accounts are described by niche only, without names or screenshots, so that neither the clients nor the people who messaged them are exposed.

7,742 Calls Booked

In the last 90 days across 48 accounts

16.3 Calls/Month

Median result per account

421.7 Calls/Month

Top account result in the same period

​
Case Study 1 — Personal Brand Mentoring

A personal brand mentor whose inbox was silently losing nearly half of all leads. Most people wrote in and never heard back within a day — not because the owner was ignoring them, but because the volume had outgrown manual handling.
Key insight: Nearly half of all leads were getting no reply before Setor AI. The problem wasn’t intent — it was capacity. Once every inbound message received an answer in seconds and follow-up ran automatically, the booked-call rate reflected the actual level of interest coming into the account.

​
Case Study 2 — Online Fitness Program

A fitness program with a fast-looking inbox that disguised a long tail. The median reply time appeared healthy, but the 90th-percentile told a different story: one in ten leads was waiting over five hours.
Key insight: The median reply looked fine — only the tail revealed that 1 in 10 leads waited over 5 hours. Those are often the most serious prospects, the ones who ask a detailed question and then judge the business by how long it takes to answer. Closing that tail turned this account into the top performer across the entire client base.

​
Case Study 3 — Online Personal Trainer

A personal trainer handling a high-volume inbox manually. Nearly four in five replies took more than five minutes, and the median wait was close to two hours — which at 9pm becomes the next morning.
Key insight: From two hours to one minute, including 3am. The before state was not unusual for a solo operator — most people cannot watch their inbox around the clock. But leads don’t know that, and by morning they’ve moved on. Instant response at any hour changed the economics of every evening message.

​
Case Study 4 — Online Courses and Business Mentoring

A business mentoring account running online courses. The reply coverage was reasonable before Setor AI — but the volume of conversations the account could handle was capped by human bandwidth. That cap came off.
Key insight: Nearly 6× the conversations, and the share of leads left without any reply fell from 1 in 5 to 1 in 30. The volume increase alone wouldn’t have mattered if reply quality had dropped — it didn’t, because the AI handled every thread individually rather than working through a queue.

​
Methodology

Understanding how these numbers are calculated is as important as the numbers themselves.
The cutoff is the first trace of Setor on the account — the earliest contact on record, not the date the official Meta API was connected. Several accounts were using an earlier Setor AI integration months before the API connection, so a window measured from the API date would compare the tool against itself.The before window is the full year preceding that cutoff point. The after window is the last 90 days. Months separate the two windows, which means conversation volume is affected by organic account growth and cannot be compared directly. Time-based and percentage metrics are not affected by this, and only those are placed side by side.
One denominator is used for everything: conversations where the lead sent at least one text message. Story reactions, bare emoji, and empty threads are excluded — identically on both sides of the comparison.
A follow-up is a message from the business that was preceded by another business message the lead never answered, separated by a gap of at least 30 minutes. That gap excludes multi-part replies within a single turn.
Bookings from before Setor AI was installed do not exist in any system — call records only begin once an account is connected. Calls were established by reading conversation content with a language model, using the same measurement on both sides of the comparison.Before publishing, the classifier was validated against 500 conversations from the period where the outcome is also recorded in the system. Precision: 98.3%. Recall: 92.8%. Accuracy: 95.6%. It errs toward undercounting — it will miss a booking more often than invent one.
Response time is measured only on pairs where the lead sent a message and the business replied within 24 hours. A reply after 26 hours is not a reply — it is a follow-up, and it is counted as one.