Deployment study · pilot
Twelve stores, one set of books
What connecting a twelve-location restaurant group’s point of sale actually surfaced — including the 257 people the till knew about that the roster did not.
Everything in the “What we found” section is real, measured data from a live deployment. It is a diagnostic study, not an outcomes study.
The Results section is deliberately empty and marked [AWAITING CUSTOMER SIGN-OFF]. We do not publish outcome numbers until a named customer has confirmed them in writing. Until then this document is used in sales conversations as a diagnostic — which is genuinely persuasive, because the buyer recognises their own business in it.
Before
A twelve-location fast-casual group across four states, 349 people on the roster, running roughly $858,000 of net sales a month. Operationally competent, growing, and entirely typical in how they saw their numbers:
- Prime cost arrived with the accountant, about three weeks after each month closed
- Labour was reported per store from the scheduling system’s own roster
- Store-to-store comparison happened in a spreadsheet maintained by one person
- Nobody could state last week’s prime cost for any single location without asking
The question we open every deployment with: which of your locations made money last week? Not last month. Last week.
Implementation
| Step | Work | Owner |
|---|---|---|
| 1 · Locations | Twelve sites configured with addresses, business hours and geofence | Nexora + ops |
| 2 · Roles | Departments (Operations, Management, Sales, Front Desk) and roles (Cook, Cashier, Concierge, Shift Lead, Manager) | Ops |
| 3 · POS connection | Sales, orders and timeclock data connected | Nexora |
| 4 · Employee mapping | The gating task. Matching the POS employee list against the roster | Area managers |
| 5 · Expenses | Historical bank statements uploaded and classified against a chart of accounts | Finance |
| 6 · Access | 35 people granted scoped access: 7 super admins, 5 admins, 23 store managers | Owner |
Nothing was replaced. The group kept its point of sale, its payroll provider and its accountant.
What we found
1 · A third of worked hours carried no cost
Their labour was reported at 24.4% of sales. Costed across every hour actually worked, the real figure was closer to 35%. Every labour decision made in the preceding period had been made on the first number.
2 · A 52-point prime cost spread inside one brand
| Store | Orders | Net sales | Prime cost |
|---|---|---|---|
| Parkview | 5,617 | $94,078 | 23.60% |
| Westbrook | 1,584 | $32,622 | 24.51% |
| Brookside | 2,260 | $45,019 | 26.18% |
| Eastway | 6,406 | $115,567 | 29.51% |
| Hillcrest | 3,848 | $79,863 | 47.25% |
| Palmview | 3,819 | $78,532 | 53.86% |
| Lakeshore | 2,845 | $56,991 | 55.96% |
| Cedarfield | 3,669 | $76,512 | 61.29% |
| Northgate | 4,627 | $86,513 | 63.94% |
| Stonebridge | 3,540 | $62,379 | 71.66% |
| Riverbend | 3,580 | $79,324 | 72.22% |
| Millbrook | 2,596 | $50,586 | 75.84% |
| Group | 44,353 | $857,985 | 50.2% |
Read this table with the coverage figure in mind. Several of the lowest prime cost stores are locations whose POS employees were not yet mapped, so their labour was undercounted. That is exactly why the ranking and the coverage percentage have to be read together — and precisely the failure mode this deployment existed to expose.
3 · Overtime nobody had totalled
4 · Purchasing that had never been seen in one place
Results
[AWAITING CUSTOMER SIGN-OFF — do not fill from estimates.]
To be completed once the group has confirmed, in writing, the figures for a full period after remediation. The four metrics to capture:
- Labour cost coverage — before 67.9%, after [ ]
- Corrected group labour percentage — reported 24.4%, actual [ ]
- Overtime as a share of scheduled hours — before 11.0%, after [ ]
- Prime cost spread, best to worst — before 52.2 points, after [ ]
Plus one direct quotation from a named person with a job title, approved in writing.
A diagnostic study a buyer recognises themselves in outperforms an outcomes study they suspect. Our reader reads P&Ls for a living and has been sold to by eleven vendors this year — the moment one number looks rounded up, every other number on the page becomes suspect, including the true ones. The findings above are the strongest assets we have precisely because they are checkable.
Reusable template
For the next study. Same structure, same discipline.
| Section | Contains | Rule |
|---|---|---|
| Before | Size, markets, how they saw their numbers, what they could not answer | No adjectives about their competence |
| Implementation | The six steps, who owned each, what was not replaced | Name the gating task honestly |
| What we found | Coverage, spread, overtime, purchasing — with scope on every figure | Measured only. Never modelled |
| Results | Four before/after metrics plus one attributed quotation | Blocked until signed off in writing |
Run the same diagnostic on your group
Twenty minutes and your own POS. The coverage number alone usually changes how an operator reads every labour report they already have.
Book a 20-minute demo