The menu voice-readiness checklist.

Callers use spoken vocabulary that rarely matches your menu names. 'Pep' for pepperoni, 'small' for twelve-inch, 'the usual' — the last is unsolvable, the first two are trivially solvable with aliases. Adding aliases is the single highest-ROI menu prep task.

A voice agent that offers an 86'd item breaks caller trust immediately. If nobody on your team owns the sold-out workflow, add that owner before you launch — not after.

Sit down with the person who built the POS menu and open a shared doc. Sort your item list by last-30-days phone-order count. For the top 20 items, add every spoken variant a regular would use — abbreviations ("pep" for pepperoni), size shorthand ("small" for a 12-inch), plural-vs-singular forms, and any neighborhood vocabulary your regulars actually say on the phone. If your shop takes Spanish or Vietnamese calls, add those variants in the same pass rather than deferring them to a "phase two" that never happens. Fifteen minutes of this review typically covers 80% of the alias coverage your agent will ever need; the remaining 20% surfaces during the first two weeks of live traffic and gets added ad hoc.

Once the alias sheet is done, walk the "Operational reality" group in the checklist above. If the sold-out workflow doesn't have a named owner, stop and assign one before ticking any launch-adjacent box — that single decision predicts more of rollout success than the rest of the checklist combined. When you're ready for the broader rollout plan the alias review feeds into, move to the implementation checklist.

The checklist assumes a POS that exposes items, sizes, and modifier groups as structured objects. Menus maintained only as PDFs or printed sheets need an intermediate import step first — Fire It can consume a spreadsheet, but the checklist's "structure" items presume you're editing structured data, not scanning a flyer. Alias coverage also assumes your callers speak in fairly consistent regional vocabulary; extreme dialect variance benefits from a longer weekly tune-up rhythm than the default cadence implies.

Progress is stored only in your browser's localStorage under a stable key; nothing is transmitted to Fire It, so completing the checklist on a shared kitchen tablet means every operator on that device sees the same state. Use "Reset" to clear it. For the revenue math that justifies the rollout the checklist prepares for, use the missed-call calculator alongside the ROI calculator.

Not every checklist item blocks a launch. Some are ideal-state hygiene; others are non-negotiable. The three that block launch are: every item has at least one price, every required modifier group has a max, and someone owns the sold-out workflow. Without those, the agent can't function correctly. Everything else — alias coverage, holiday hours, upsell configuration — improves the experience but isn't blocking. Prioritize the blockers in your first hour of setup and revisit the rest weekly.

Alias coverage is never done. Every week produces a new mishearing where a caller used a word the agent didn't recognize. Fire It surfaces these as an 'unmatched utterance' list in the dashboard, and adding the utterance as an alias takes about ten seconds. Treat this as a weekly ten-minute review, not a one-time setup task. Shops that stop reviewing after month one drift; shops that keep reviewing keep improving accuracy for as long as the menu stays live.

Menu launches fail more often on sold-out discipline than on any other single factor. A voice agent that offers an 86'd item breaks caller trust immediately, and blame lands on the AI rather than the workflow. Fire It's admin exposes the toggle in one tap, but the operational habit — someone marks the item 86'd within thirty seconds of the last one going out — is a shift-lead job. Assign the job explicitly, put the toggle on the make-line tablet, and revisit whether the habit stuck two weeks in.