Sample report
The shadow-mode report you get during deployment
During deployment, every automation runs in shadow mode: it watches your stores, decides what it would have done, and logs it, but sends nothing and changes nothing. After 14 days you get a report like this one, and you decide what goes live.
Harbor Lane Foods: shadow-mode report, days 1–14
214items checked across all stores
92things it would have flagged
89%of flags confirmed real by your managers
0changes made to any of your systems
Automation by automation
Each flag was compared against what your team actually did. “Confirmed” means a manager agreed it was a real issue; a false alarm is a flag that turned out to be fine.
| Automation | Flags | Confirmed | False alarms | What it would have done | Recommended next |
|---|---|---|---|---|---|
| Opener check | 9 | 8 | 1 | Nudged the opener at 4 stores; told the area manager twice | Autonomous |
| Missed clock-outs | 23 | 22 | 1 | Texted the manager each night with the suggested out-time | Ask first |
| Labor vs. target | 31 | 24 | 7 | Suggested trimming a shift at 10:30 on 31 store-days | Ask first |
| Truck order check | 3 | 3 | 0 | Flagged a missing item and two quantity typos before cutoff | Ask first |
| Deposit reconciliation | 4 | 3 | 1 | Listed 3 short deposits and 1 bank timing gap for the owner | Report only |
| Payroll cross-check | 22 | 22 | 0 | Listed hour differences between time clock and payroll before approval | Report only |
Messages it would have sent
- 48 texts to managers, sent as 17 digests instead of one text per issue
- 3 held for quiet hours (11:30 p.m.–5 a.m.) and delivered at 5
- 0 to hourly crew. Labor automations only ever text managers and owners
- 0 to guests
The false alarms, explained
- Opener check (1): a store opened late on purpose for a local event. Fix: add the event to store hours.
- Missed clock-out (1): a manager punched out at a second register. Fix: read both registers.
- Labor (7): all on two catering days. Fix: let managers mark a day as catering.
- Deposit (1): a holiday delayed the bank posting. It matched the next day.
From the log
Every decision is written down: what it saw, what it would have done, and what actually happened.
- Opener check, Store 4. No clock-in 2 minutes before open. Would have texted the opener, then the area manager at 7:10. Actual: store opened 7:14; manager called the opener by phone.
- Missed clock-out, Store 2. A shift still open 1 hour after close; the alarm armed at 9:22 p.m. Would have texted the manager “Close at 9:20? Reply Y”. Actual: found on day 8 during payroll prep.
- Truck order check, Store 7. Order had 40 cases of cups instead of 4. Would have asked the manager to confirm before the 3 p.m. cutoff. Actual: caught at delivery; returned.
- Labor vs. target, Store 1. Labor running 6 points over target with slow sales. Would have suggested sending one person home at 1 p.m. Actual: no change; day closed over target.
- Payroll cross-check. 11 hour differences across 5 stores, listed with the time clock and payroll values side by side. Actual: managers fixed 9 before approval; 2 were found after payday and corrected on the next check.
Proposed go-live
You choose each one. Nothing changes until you do, and every automation keeps its off switch.
- Opener check → autonomous texts
- Missed clock-outs → ask first (manager replies Y)
- Truck order check → ask first
- Labor vs. target → 14 more days in shadow, with catering days added
- Deposit reconciliation → weekly report
- Payroll cross-check → report before every approval
What shadow mode does not do
- Write to any system. Access stays read-only until you approve go-live.
- Approve payroll or move money. Those stay with a person after go-live too.
- Text anyone. Every message above is logged, not sent.
- Promise savings. The report shows what it caught, and you judge what that is worth.
Book a walkthrough See pricing and deployment How shadow mode works