A new dish can taste promising in a quiet kitchen and still create problems during service. The portion may vary between cooks, its final garnish may compete for space at the pass, or the description may set an expectation the plate does not meet. A launch trial brings those questions together before the dish becomes a standing promise to customers.
This guide is about deciding whether and how to add one dish. It assumes the recipe's food controls and required information are reviewed by the responsible people. It does not approve a recipe's safety, set a selling price or replace a tested production method. The feedback counts below are original illustrations, not a launch threshold or a forecast of demand.
Identify the exact dish being tested
Give the candidate a recipe identifier and revision. Attach the intended portion, presentation, ingredients, approved preparation method and relevant service conditions. Keep the cost card and proposed menu description linked to that same revision. A tasting of one portion cannot validate the cost or presentation of a larger portion sold under the same name.
Record the customer occasion and channel. A dish for a quick lunch needs a different service trial from a dish prepared for a planned banquet. If delivery is part of the intended offer, evaluate the actual approved packaging and journey separately. Do not treat a good result on a plate at the pass as evidence of delivery performance.
Set the questions before collecting opinions
Separate the product question from the operating question. Product feedback might ask whether the description matches the served dish, whether the portion fits the occasion and what the tester would change. Operating evidence might ask whether another trained cook can reproduce the portion and whether the finishing step fits alongside the existing menu.
Choose how responses will be recorded before the tasting. Keep wording and options consistent for the same question. If you ask about willingness to order, show the intended portion and proposed price and record those conditions. A complimentary sample judged without a price does not answer the same question.
Make feedback voluntary and allow criticism without pressure from the cook or owner. Collect only the information needed for the decision. Anonymous response references are usually enough for the shared trial record; individual contact details belong in the appropriate restricted system if follow-up has been agreed.
Worked example: ten forms from twelve testers
A fictional team invites 12 people to evaluate one candidate revision. Ten return a complete answer to a stated willingness-to-order question: 7 say yes, 2 are unsure and 1 says no. Two invited testers do not return that answer. The team keeps those two missing responses visible rather than assigning them a preference.
The observed yes share is 7 divided by 10, or 70% of completed responses. Response coverage is 10 divided by 12, about 83.3% of invited testers. Neither number means that 70% of the restaurant's future customers will buy the dish. The small invited group may differ from the intended audience, and a stated intention is not a completed purchase.
The comments matter alongside the counts. If several testers interpret the menu description differently from the plate they receive, the next action may concern wording or presentation. If the recipe changes after that feedback, the new version needs its own evidence; do not quietly combine old and new responses.
| One original tasting question | Count | Interpretation |
|---|---|---|
| Invited testers | 12 | Defined invitation group |
| Completed responses | 10 | 7 yes + 2 unsure + 1 no |
| Yes | 7 | 70% of completed responses |
| Unsure | 2 | A distinct response, not a missing answer |
| No | 1 | Recorded preference |
| No completed answer | 2 | Unknown preference; retain separately |
Test the work around the dish
Use the intended recipe revision and a trial scope approved for the kitchen. Observe the preparation, portioning, finishing and handoff with the people and equipment expected to perform them. Record where an extra helper, a pre-prepared component or an unusually quiet service changes the conditions. A workaround is useful evidence, but it should not disappear from the report.
Check the effect on existing dishes as well as the candidate. A new finishing step may share a person, surface or piece of equipment with another item. Record the overlapping work and the observed effect before deciding that more equipment is needed. If the trial uses a larger or smaller recipe batch, validate that yield separately; a proportional ingredient calculation alone does not establish the resulting quality or process.
Keep the cost card, purchasing specification, ingredient information and available storage consistent with the final candidate. These are separate decisions with named owners. A favourable tasting result cannot compensate for unresolved ingredient information, an untested process or an unavailable specified component.
Decide the scope of the first release
Record whether the candidate is held for revision, tested again, released within a stated scope or dropped. Identify the approved recipe version, menu wording, portion and channel, the evidence reviewed, any restrictions and the person authorising the decision. There is no automatic pass percentage in the worksheet.
If the first release is limited to a particular service or channel, make that limit visible to the people taking orders and producing the dish. Confirm that the correct recipe and customer-facing information are available before taking those orders. An approval to trial a candidate is not an approval to offer it everywhere.
Review real service without rewriting the trial
Choose the review date and observation scope when the release is agreed. Compare actual orders, comments, production variation and effects on the existing menu with the assumptions that supported the decision. Keep the tasting record intact so the team can distinguish what it expected from what happened.
Low sales can reflect availability, visibility, price, customer fit or execution; the count alone does not identify the cause. Investigate the relevant evidence before changing the recipe again. Give each change a version and a focused check so the next review is about a known offer.
What recipe-evaluation guidance supports
The Institute of Child Nutrition's recipe-standardisation guide describes evaluation after recipe verification, using feedback from the preparation team and intended customers and selecting questions that the evaluation needs to answer. That is general recipe-evaluation context here. Its school-programme requirements, suggested group sizes and sample rating forms are not applied to this restaurant trial. The feedback example and decision record are original.
Put it into practice
Your action checklist
Your progress saves in this browser. No sign-up needed.
0 of 8 complete
From reading to doing
New-dish trial and launch decision record
The Menu Launch Trial worksheet links the candidate revision to recipe, customer and service evidence. A feedback summary reconciles invited testers, yes, unsure, no and missing answers, showing response coverage and the yes share among completed responses. Manual sections record trial conditions, operating observations, changes, approval scope and the first-service review. It does not predict sales or approve a launch automatically.
Open the Menu Launch Trial tab in the launch workbook. Keep a copy for your own business.
Download the free workbook See what’s inside the toolkit →Sources & further reading
- USDA Recipe Standardization Guide for School Nutrition Programs (Institute of Child Nutrition with USDA FNS) ↗
2022 edition, retrieved 7 November 2026. Product Evaluation Phase used for general recipe-evaluation context only. School-programme requirements, suggested group sizes and source rating forms are not applied. All numerical examples and register design here are original.
Published by Keystonne Grow. Product links connect to the Keystonne ecosystem; check the linked product or service for its current scope.
