A lot of tools will now let a chatbot answer questions about your business. The trouble with that is a chatbot will happily hand you a figure that looks right and came from nowhere. Loupe does not work that way. Software does the arithmetic, from rows in the files you sent us, and the wording around each figure is a written template — no model touches your numbers today. The rule below is the boundary that holds if one ever does.
$41,961 stale quote value/mo × 75% close rate = $31,300/mo
Every term in that line is something counted in your own records. Change a record and the figure moves with it.
“57open quotes have been sitting past 30 days. The line above is what that pile is worth in a month, and here is where each part of it came from.”
Same figure, plainer words. Today that sentence is a template with the computed numbers dropped into it. If a model ever writes it, this is still the ceiling.
Nudge the 75%. Round the $31,300 up. Fill a gap in your export with something that looks about right.
And when your records genuinely cannot support a figure, the finding does not appear at all.
Plenty of owners now have an AI receptionist or booking agent picking up the phone. When you import your jobs, you mark the ones it booked. You set that label — we never guess it from a name or a number, because guessing it would be inventing something about your business.
AI-booked work goes through the same checks as the rest: what got quoted, what got invoiced, what actually got paid. If the agent is booking work that never turns into money, that shows up as a finding, with the records behind it.
Every figure above is from the sample report — 1,976 records read, 5 open findings, all demo records.
“You’ll always know who’s looking at your numbers. I review every dashboard personally — and we walk through it together, every month.”
Both free. The check takes about 30 seconds and asks for nothing; the workspace runs the same detectors on one export of your own. Or start a conversation →