Reports

The web chat could not finish a booking. Three faults were in the way, and all three are fixed. On the phone, Ken's "no email, no booking" rule is live and proved on two test runs. Numbers are said right, and the test line had more than twenty faults found in-house. The phone is current with the code. Five decisions are Ken's.

Web chat only. 473 scripted chats across 57 customer situations, on all four early-adopter pages. Five faults a customer would have read on screen, in Josh's own words. All five fixed and re-checked live. One decision for Ken and Dil: updates hang up on whoever is chatting at that moment.

Josh already asks all that the ISN forms ask. We put Ted's, Glen's and Chad's ISN forms next to what Josh asks on a call. Not one house question is missing. The forms ask less than Ken's own list. The gaps are small: lockbox and gate codes, cross streets, a few contact details. Two calls are Ken's: how the add-on questions become standard, and whether to build the six boxes.

Two leaks closed on the phone. Josh forgot what a caller had spelled, and "1800 square feet" was being read as a house number. What 39 calls said, why the test runs were failing on their own, and a twelve-call proof run on Property Masters' line.

Six faults fixed on Chad's line, measured before and after. We ran the same nine test calls twice, an hour apart, changing only the code. Josh's score went from 70 to 81. The worst fault: he threw away an email the caller had just confirmed. Two faults are still open and we name them. Nothing reached a client.

Two sessions in one day. One built a machine that tests Josh every two hours and drafts the fixes overnight. The other worked down Chad's list and fixed six of his nine complaints. Eight sets of changes merged. Nothing reached a client.

Read this first today. Eighteen test calls instead of one. Nine faults found in Josh, six fixed in code. The number of callers at once the phone can actually take — and a straight answer on what to buy, and what not to.

Read this first today. Callers heard nothing from ten to three this morning. It is the phone company, not Josh — proved from both ends. What happened, how we know, what we are asking Telnyx, and what we should have in place for next time.

Start here. Plain English, five minutes. We ran the test: one client costs 54 cents a month, not the five dollars I first said. What the four helpers would be, what the checker found while it was in there, and the three things we need from Ken.

Can helpers run the services we sell? They own the middle of every one and neither end. All seven accelerants walked to the exact point where each stops — plus the full stack Dil asked about: client websites, WordPress, Google Business Profile, social, Monday.com, Telnyx, Analytics and Search Console.

The detail behind report 1. What was measured, what is still a guess, why pointing it at Chad was safe, and the two mechanisms Ken said he did not quite believe.

The first report, which Ken has read. Still worth keeping — but note its $5-per-client estimate has since been measured at 54 cents. Reports 1 and 3 replace that number.

The six calls Chad placed to GC's line, triaged: the seven Josh fixes shipped (the invented $500, "we don't work Fridays," and five more), why the stall is mostly the caller and far-field audio rather than a code bug, and the confirmed Telnyx fault behind the silent calls (case #635168).

The overview Ken asked for twice: four agents at any size, which accelerants need one, and a correction to Tuesday's report.

The measured bills, the quality control agent's three runs, and why the phone is 96% of it.

What moved today.

The draft homepage, live on its own address for Ken to send to members.

The eleven jobs on the phone agent, and what each one fixes.

Step by step, for the two people who need it.

Beth's costing work — what a minute on the phone actually costs us.