How a shift trade settles

Coverage problems are market problems

When someone calls out at 6 a.m., the question is not "who do we know?" — it is "who is qualified, authorized, rested, and willing, and can we prove all four afterward?" Multi-location operators run that question constantly. Treating each answer as an informal message thread loses the proof, and eventually loses the fairness argument too: the same willing people get asked first, every time.

The shift marketplace treats an open shift as a small market with rules, and every trade as a state machine that either completes cleanly or returns the shift to open. Nothing settles quietly.

The states, plainly

• PROPOSED. A trade exists on paper: a specific shift, an offering party, a candidate or candidate pool. Nothing has been agreed; the shift still belongs to its original assignment.
• PENDING. The parties have signaled acceptance. The trade is real enough to check, and this is where the compliance evaluation runs — before anything finalizes, not after a problem surfaces.
• AWAITING_REVIEW. The trade passed automated checks and now sits with the designated reviewer — typically the location manager — because operator policy says a human confirms coverage changes. A review that times out does not approve the trade; it returns it.
• COMPLETED. The assignment changes hands, the roster updates, and the record — proposal, checks, approvals, outcome — is written as a durable entry.

Declines, expirations, and cancellations are exits back to open, not dead ends. A shift that fails to trade is visible as still-open coverage, with the history of attempted trades attached.

What the compliance check evaluates

The checks are policy, expressed as data and evaluated at the transition into PENDING — the last moment a bad trade is still cheap to stop:

• Qualification match. The candidate holds the role and certifications the shift requires.
• Hour and overtime limits. Accepting the trade keeps the candidate inside configured weekly and pay-period limits.
• Rest and spacing rules. The trade does not create back-to-back or insufficient-rest patterns that policy prohibits.
• Location authorization. The candidate is granted to work at that location, inside that brand's boundary.

A failed check is not an error; it is the marketplace working. The trade does not proceed, the reason is recorded, and the candidate pool advances.

Why a state machine and not a group chat

The informal alternative — a flurry of texts and a manager's memory — fails in three specific ways. It cannot prove who agreed to what. It cannot prove the compliance questions were asked. And it concentrates opportunity on whoever is fastest to reply to the manager personally, which is a fairness problem that compounds into a retention problem.

Explicit states fix all three by construction. Every transition is recorded with actor and timestamp; every check outcome is preserved, including the denials; every eligible candidate sees the same opening on the same terms. Six months later, a manager — or counsel, or a franchisor — can reconstruct any trade from the record rather than from recollection.

The same mechanics extend across locations inside a brand. Where operator policy allows it, an open shift at one location can reach qualified, authorized staff at a sister location — through the same states, the same checks, and the same grants, never around the brand's boundary. Coverage stops depending on who happens to know whom. The constraint model rosters are built from is documented at scheduling constraints & the roster engine.

What we will not claim

We will not quote fill-rate or time-to-cover figures, because the honest ones belong to your locations and your labor market. The claim is structural: trades that settle through explicit states, compliance checks that run before finalization, and a record that survives the shift itself.