Intelligence proposes. Chairfirst helps the institution commit — and learn.
Multi-site institutions make thousands of consequential operating decisions every month and have no governed way to make them, record them, or learn from them. Data and AI recommendations are inputs, not decisions. A proposal becomes consequential only when someone with authority commits the institution — and only then does the outcome mean anything.
Chairfirst, Inc. · Phoenix, Arizona · Founded by John E. Hancock
The bottleneck
Head office sets the targets. The sites make the result.
The centre declares the numbers — revenue, margin, labour, capacity, service. Performance is created somewhere else: one site, one shift, one commitment at a time. A general manager decides tomorrow’s production. An operations manager decides which customer gets short-shipped, and what the expedite costs. Each works from partial information, against a real authority limit, on a decision that will not wait.
Central systems can see the variation between sites. They rarely answer the question that would let anyone act on it: which part of that gap is decisions, and which part is just structural?
Scarce, perishable capacity
A shift, a kitchen line, a delivery window. Unused, it is gone — there is no inventory of yesterday’s labour.
Competing commitments
Every yes to one customer, product or site is a no somewhere else. The trade-off is the decision.
Divided authority
The person closest to the facts can commit only so far. Past that line, someone else has to.
The decision card
What an operator recognises on sight.
It arrives inside the shift, not in a weekly report. The constraint applies before the choice, and declining is a legitimate answer. The record is made before the outcome, which is what makes the review afterward contemporaneous rather than reconstructed.
High consequence · Store 14 · Labour
Add 22 labour-hours to Fri–Sat dinner
Why
Forecast demand exceeds scheduled capacity. Projected service-time degradation and sales loss if unaddressed.
Constraints
Labour-percentage guardrail, service standard, and the general manager’s own authority limit.
Who may decide
The GM at this amount. Above it, the decision routes to the area director.
Uncertainty
Stated as a range — and the system declines to recommend where evidence does not support one.
Evaluation
Estimated versus realised, under a method declared before the shift.
Representative rendering, not a screen capture.
What happens around the card
The card is what you see. This is what runs.
The product is the governed decision lifecycle. The card is how a person takes part in it. Nine stages run in order, most of them before anyone is asked to choose.
Something changes in the data that might warrant a decision — a demand shift, a shortfall, a constraint tightening.
Most things that could be surfaced should not be. The system chooses the few decisions worth a manager’s attention.
The decision arrives with its reason: what changed, what it implies, and what the recommendation rests on.
Options that would breach an operating floor are removed before the choice is offered, not flagged afterward.
The decision routes to whoever is allowed to commit at that amount, and no further.
A person with authority commits the institution. The record is made here, before the outcome is known.
The commitment flows back to the systems that carry it out, with its evidence attached.
Estimated versus realised, under the method declared before the commitment.
What is learned from resolved decisions changes how later ones are selected, explained and constrained.
Seven stages run at the live deployment. Measurement is being instrumented now; the improvement loop is designed and not yet running.
How it fits
It composes with the systems you already run. It does not replace them.
Chairfirst reads from the systems that already hold the operator’s data — point of sale, workforce and scheduling, inventory, enterprise resource planning — and writes back the commitment and its evidence. It is not a system of record and does not try to become one.
That matters to a buyer for a plain reason: no migration, no replacement project, no team asked to abandon the tools it already runs. The decision layer sits above what exists and leaves it in place.
Reads from
Writes back
The commitment, who authorised it, the constraint it respected, and the evidence attached — returned to the systems that carry the work out.
The alternatives
Every alternative already exists inside the operator’s business.
Each does something real, and each stops somewhere specific. The point is not that they fail — it is where they stop.
| The approach | What it does well | Where it stops |
|---|---|---|
| Dashboards and reporting | Show the variation between sites clearly. | Decide nothing, and record neither who committed nor why. |
| An exceptional area director | Genuinely makes the right calls. | Does not scale, and leaves when they do. |
| Spreadsheets and the weekly meeting | Bring people together around the numbers. | Reconstruct the reasoning after the outcome is known — when it is least reliable. |
| Forecasting and optimisation tools | Produce a sharp recommendation. | The authority, the constraint and the record are still missing. |
| Doing nothing | Costs nothing to adopt. | The decision still gets made — just without a constraint, a record, or a way to learn from it. |
What a manager experiences
Designed for the person who has to decide.
When a manager changes a recommendation, the system asks one question — “What did you know that the recommendation did not?” — with quick options a manager would actually pick.
The one question
- A local event or unusual demand
- A staffing or equipment constraint
- A customer commitment
- An inventory discrepancy
- Weather or traffic
- Incomplete or stale data
- A brief note, in their own words
The design rules
- The decision is instrumented, not the person.
- A modification is treated as information gained, not as an error.
- Modification rate is never used on its own as a performance score.
- When a manager’s local knowledge changes how later decisions are made, they are shown that it did.
The record exists so the institution can learn — not so anyone can be second-guessed.
Where it runs today
One live deployment.
A multi-unit restaurant operator, running labour and production decision families inside the manager’s existing daily ritual. A working platform: signal, explanation, constraint enforcement, authority resolution, and the commitment record. Measurement is in progress. No measured operating-line result yet.
Live never implies measured; measured never implies independently attested.
How it grows inside an institution
An institution that governs one decision usually has several.
Labour is rarely the only recurring commitment that matters. Production, purchasing, inventory, capacity, maintenance and service commitments have the same shape: a scarce resource, competing claims, a constraint, someone authorised to decide, and an outcome visible within days.
Because a second decision family reuses the connection, the authority model and the evidence surface already in place, it is substantially faster to stand up than the first. That is a property of the design, not a promise about a number.
Why a finance leader engages
What a finance leader is actually buying.
Improve the decision-addressable portion of performance
Not the whole gap between sites — the part decisions can actually move.
Depend less on an exceptional manager at every site
The alternative is hiring one, which works, does not scale, and leaves when they do.
Hold a contemporaneous record
Made before the outcome, under a declared method. Useful in a board meeting, and in a sale.
None of the three is a promise about a number. The promise is process integrity: the outcome is measured, not warranted.
Who it is for
Multi-site institutions where the local manager holds real authority.
The buyer is an institution where a local manager holds real authority to commit a scarce resource, frequently, with a consequence that can be measured within a short window. The decision could be labour, inventory, pricing, capacity, maintenance — anything that commits dollars or service quality and whose outcome is visible within days.
The same decision structure appears wherever that condition holds.
The founder
John E. Hancock
Four decades inside operationally complex institutions, building software where the consequential decisions still came down to whoever was in the room. Chairfirst is what that career produced: software for the decision itself, not for the data around it.
In addition to Chairfirst, Hancock manages the Bacaro technology estate and its related advisory activities.
[email protected]How this starts
What engaging looks like.
A paid initial deployment
A small number of representative sites, one supported decision family.
The expansion framework agreed first
Eligible sites, acceptance criteria, decision authority and rollout mechanics established before deployment begins. Each expansion remains a separate decision the customer makes.
An acceptance review on a date
On the calendar before the deployment starts, with a named person on the customer’s side.
Rollout in waves
Sized to what can actually be delivered well.
A paid initial deployment tests whether the product is valuable; a free one tests whether the attention is welcome.
Ownership
Your records remain yours.
The operator’s data and its decision records belong to the operator, and can be exported. Chairfirst retains its platform, its methods and the improvements it makes to them, and it improves the system through protected patterns rather than by exposing any operator’s information to anyone else.
Your records remain yours. The system improves through protected patterns — not by exposing your operating information to anyone else.
Have a conversation
Tell us about one decision that matters.
A member of the Chairfirst team will follow up, usually within one business day.