The loops
Three questions.
The hard numbers, finished.
XPRNS takes the export and names where you are losing — the patterns that are awkward to bring up. You keep what the file cannot see: what happened on the floor. Together, that is the whole product.
The questions
- 01One location · the day
Was today actually good?
why it hitsThe numbers against that location's own usual — not a gut feel, not month-end. The brief names the shape. You already know whether it was a no-show streak, a walk-in flood, or a staff call-out.
you receive- Today against this location's own normal
- Takings, mix, new vs returning
- Who just became overdue
- Tomorrow's book: projected volume, regulars, first-timers, what to push
- 02The book
Where are we leaking?
why it hitsThe losses that do not show up in a good total: regulars slipping, empty chairs, promotions that paid the wrong people. Straight to the leak — before anyone has to say it in a meeting.
you receive- Who stopped coming — ranked by each person's own rhythm
- Gaps to fill — ranked call list and what filling the book is worth
- Did the promotion pay — what it bought vs what was already going to happen
- Overlap when you need it — what shares a customer base with what
- 03Owners · every location
Which door needs me?
Multi-locationwhy it hitsThe multi-site scan managers avoid: which doors are soft against their own usual, which look fine on money but thin on regulars. Only exceptions speak. Silence when the portfolio is fine.
you receive- How many sites are within their own usual range
- Only the exceptions named — quieter, busier, or money fine / regulars thin
- One concrete next step per flagged site
- A quiet brief when nothing needs you — that is the hours saved
Before you pay, the backtest grades our forecasts against your own past — so you see the error rate on your file, not a pitch deck.
Your first briefs come from your own file.
What lands
We take the awkward part
Slipping regulars. Soft sites. Discounts that bought nothing. The brief says it so nobody has to soften it in the room.
You keep the floor
A spike or a quiet day always has a reason the export will never hold. The loops flag the shape. You already lived the reason.
Each place against its own usual
A small shop is not punished for being small. Fair compare — then only the exceptions get named.
It says when it cannot see
A missing column prints “not available.” Never a confident zero. Guesses in the hours audit are labelled as guesses.
Tonight's brief
Sample · invent names
ALL LOCATIONS — Four-site salon2026-07-28———————————————————————————————————————— 2 sites within usual. 2 need a look.————————————————————————————————————————! Lake Street — money looks fine,! regulars don't 8 services against a usual 9 0 returning, 8 first-time! Riverside — quieter than its own! Tuesdays 2 services against a usual 9————————————————————————————————————————DO THIS - Win-back list at Lake Street, not a - discount blast. - Ask Riverside what changed before it - becomes a pattern.———————————————————————————————————————— 71 minutes of desk work this replaces.
That is one loop's output. See all seven beside the raw exports that produced them.
It speaks your industry
Restaurant, bar, cafe
guests · covers · menu groups · servers
Shop and retail
customers · baskets · departments · associates
Gym and studio
members · classes · instructors · attendance
Salon, clinic, appointments
clients · bookings · providers · rebooking
Same system underneath — only the words change. Not on the list? You get the generic wording until yours is written, which takes about a day. A dispensary, a studio, a shop desk — the file still names who, when, what, and how much.
What is actually built
Reads any export, no setup
143 tests, three export shapes including multi-location
Seven loops you can see run
every demo prints the command that made it
Day, leak, and all-locations packages
running today
Hours-back audit
your numbers, not our averages
Briefs arriving on their own, by email
the loops are ready; the sending is not wired
Ever run on a real business's data
not once — stated wherever it matters
The last row is the one most tools would leave off. No real business has sent us a file yet — every package is proven against synthetic data built to be awkward, and being first is what the founding rate buys.
Being first is the discount.