Meet Tsuno at Sirha Lyon 2027 · Stand S.57Learn more
EN FR
Try the demo
Menu

AI and tools

Gadget or tool? The 60-day test

Before you sign, run 60 days: voluntary use at day 14 and 60, repeated questions, manager time, a routine without the tool and the team's anonymous view.

Published on

On the counter of a closed restaurant, a desk calendar with days ticked off next to a tablet that is switched off.

To tell whether a tool is a gadget before you sign, test it for 60 days on one or two real routines and look at five things, written down in advance: voluntary use at day 14 and day 60, a fall in repeated questions, the time the manager spends on it, what happens to a routine when the tool is taken away, and the team's anonymous view on how useful it is and whether they feel watched. A gadget shines in the demo and then disappears within three weeks, whereas a tool proves itself in month two.

Thursday, 4 p.m., on the counter

Joris opened his restaurant six years ago. Two weeks ago he sat through a flawless demo: the tool answered everything, quickly, in a pleasant voice. He almost signed that evening.

On the counter he has put a desk calendar. Mélissa, his sous-chef, ticks every day someone used the trial tool without being reminded. The first ten days are all ticked. Then the blanks arrive, one, two, three in a row. The tablet sits next to it, switched off.

"It's not that it works badly," says Mélissa. "We just don't need it for closing. We know closing." Joris looks at the calendar. He won't sign tonight. He'll call the vendor back to test the tool on opening, where the new starters ask all their questions.

(Scene and characters are fictional.)

Why so many tools end up in a drawer

The demo shows the best day

A demo proves the tool can work, not that it will be useful in your kitchen on a wet Tuesday with a temp who turns up late. To test the answers themselves, the method is in how to test a voice assistant in a kitchen. The question here is different: will the team still be using it once the novelty has worn off?

Curiosity looks like adoption

In the first few days, everyone has a go. That's when people sign, and it's the only period when usage means nothing.

Forced use hides uselessness

A mandatory tool gets used a lot, but you will never know whether it helps, because nobody has a choice. Only voluntary use, without reminders, tells you anything.

The manager's time stays invisible

The tool saves the line cook time, but the manager spends evenings fixing procedures, chasing people and checking. If you don't count that time, you mistake a shifted workload for a gain.

Teams go quiet when they feel watched

A tool that logs questions or actions makes people uneasy. An uneasy team doesn't say what's wrong; it uses the tool just enough not to get called out. Honest feedback only comes if you guarantee it's anonymous and that a question asked never becomes a file on someone.

The test, in five criteria

Before day one, pick one or two routines where the problem really exists, and write down with the team the five criteria below and their benchmarks. Tell everyone what will be recorded: in France the Labour Code requires it (Article L1222-4), and elsewhere it is simply the honest way to run a trial.

  1. Voluntary use at day 14 and day 60. Every day, a tick if the tool was used on the routine without anyone asking. At day 14, at the pre-shift briefing: "Who used it this week without thinking about it, and for what?" At day 60, the same question. You count days and routines, not people.
  2. Repeated questions going down. In the week before the test, the person in charge makes a stroke for every procedure question asked twice or more, and notes the procedure. They repeat the same week around day 60. If the number doesn't fall, the tool isn't helping people learn, or the procedures are incomplete.
  3. The manager's time. Two numbers, over a typical week: interruptions for a procedure question, and time spent keeping the tool going (procedures to fix, reminders, technical problems). Net time is what counts.
  4. The day without. Around day 50, take the tool away for a day, with notice to the team, and keep the paper version. If the routine runs less smoothly and the team asks for it back, the tool has become a tool. If it runs better without, it's a gadget, or a tool placed on the wrong routine.
  5. The team's anonymous view. Four questions on paper, unsigned, in a box, at day 14 and day 60. "Does it help you? With what? Do you feel monitored? What gets in the way?" In a team of fewer than five people, paper anonymity doesn't hold: do it in a meeting, or ask someone from outside the team to collect it.

Set the rule before you start

The decision is made on day one, not day sixty. Write down what must have moved, in which direction, and what you'll do otherwise. The same principle applies to money: see how to measure restaurant software ROI in 90 days.

The decision grid

The benchmarks below are starting points to discuss with your team, not standards. They come from no study: adapt them to your size and your routines, but write them down before day one.

CriterionWhat you recordSuggested benchmark at day 14Suggested benchmark at day 60Gadget signal
Voluntary useDays ticked without a reminder, per routineThe tool is used on most days on the chosen routineAt least as much as at day 14Use halved or worse between day 14 and day 60
Repeated questionsStrokes per procedure, typical weekNo conclusion yetDown on at least one routineNo fall, on any routine
Manager's timeInterruptions and time spent keeping the tool goingMay go up: this is the start-up phaseNo higher than before the testHigher than before, and rising
Day withoutHow the routine runs without the tool, around day 50Not applicableThe team asks for it, the routine is less smoothThe routine goes better without
Anonymous viewFour questions, usefulness and monitoringMost people name a specific useThe same, with named usesVague usefulness, or a growing feeling of being monitored

One line is a deal-breaker on its own: if the feeling of being monitored grows between day 14 and day 60 and a clear answer hasn't brought it down, stop, whatever the rest says.

Common mistakes

  • Testing on the routine the team already knows. As Joris found, a tool is useless where there's no problem.
  • Counting questions by person. The test becomes surveillance, the team goes quiet and the numbers stop meaning anything. Count by procedure and by station.
  • Forgetting the manager's time. A tool the team likes but that costs the manager two evenings a week won't last.
  • Starting the trial without consulting. In France, companies with at least 50 employees consult the works council (CSE) before putting such tools in place, and a Nanterre court ruled in February 2025 that a pilot phase already counts. Details are in should you consult employee representatives before using a voice assistant?. Outside France, check the equivalent rules where you operate.
  • Deciding nothing at day 60. A half-adopted tool stays half-adopted. Keep, adjust or stop: you need one of those three answers, with a date on it.

What to measure

  • The ratio between voluntary use at day 60 and at day 14, routine by routine.
  • The number of repeated questions per procedure, in the week before the test and around day 60.
  • The manager's net time: interruptions avoided minus time spent keeping the tool going.

What we propose for a pilot

We would rather be judged on this test than on a demo. Here is what we propose to businesses that want to try Tsuno San, as a starting point for discussion. Terms are agreed with you case by case, and nothing below is a contractual offer.

  • Write down together, before day one, the criteria and benchmarks in the grid above, on your own routines.
  • Work with a volunteer team that has been briefed: nobody has to wear an earpiece, and anyone who prefers the screen keeps the same access.
  • Hold a check-in at two weeks and another at two months, with the team, on what helps, what gets in the way and what has been dropped.
  • Keep the freedom to conclude that it isn't for you.

Before committing a team, you can also start smaller: the trial built from three non-sensitive operational documents, in an isolated environment that is deleted on the agreed date. The process is on the setup page. Our rules for kitchen trials are in why we test in the kitchen before we sell.

Where Tsuno San fits in

Tsuno San is the intelligent voice assistant for frontline teams, connected to the Tsuno environment: procedures, tasks, checklists, routine checks, equipment and information about the site. It acts within the person's own rights, launches nothing without their confirmation and assesses nobody. San Voice, the voice interaction, is available as a demo; San Audio, the hands-free experience, and San 360 are under exploration and testing.

What a good tool should do for this test, and what we are working on: count use by routine and questions by procedure without ever linking them to a name. We don't describe it as done. The sheet, the calendar and the box of slips are enough to run the test, with San or any other tool. The Tsuno San page presents the assistant. The online demo lets you question Tsuno about a fictional group of three restaurants, and San Voice is shown in a guided demo.

Further reading

To measure whether the team is really learning with the tool, see independent, not dependent: measuring that the team is learning, and on resistance, why teams reject new tools. For the people side of a rollout, read the tool, but how? Adopting AI without losing the craft. And for the underlying question, talking to your restaurant: gimmick or new interface?. The pillar page Tsuno San, the work companion for frontline teams describes the tool under test, and why we show our failed trials explains how we measure ourselves.

Key takeaways

You spot a gadget in month two, not during the demo: voluntary use that collapses, questions that don't fall, a manager spending evenings on it. Write the criteria and benchmarks before day one, with the team, and decide at day 60 without telling yourself stories.

Frequently asked questions

How can you tell whether a new tool will still be used in two months?

Test it on one or two routines for 60 days before committing, and look at five things: voluntary use at day 14 and day 60, a fall in repeated questions, the manager's time, what happens when you take the tool away for a day, and the team's anonymous feedback.

Why 60 days rather than a one-week trial?

Because the first week measures curiosity. Around week two you see who keeps going without a reminder; around month two you see whether it has become a habit or ended up in the drawer.

What thresholds should you set to decide whether to keep a tool?

There is no universal threshold. You can start from simple benchmarks, discussed with the team before you begin: use at least stable between day 14 and day 60, repeated questions down on at least one routine, manager time not going up. What matters is writing them down before you start.

How do you collect the team's views without anyone feeling watched?

With four questions on paper, unsigned, dropped in a box at day 14 and day 60. In a small team, do it in a meeting or through a neutral person. Ask whether the tool is useful and whether anyone feels monitored, and answer what comes up in writing.

Do you need to consult employee representatives for a simple trial?

It depends on the country. In France, companies with at least 50 employees must inform and consult the works council (CSE) on new technologies (Labour Code, Article L2312-8), and a court in Nanterre ruled in February 2025 that a pilot phase of AI tools already counted as implementation. Elsewhere, check local rules before the trial.

What should you ask the vendor before a 60-day test?

That success criteria are written together before the start, that the trial runs on your own routines rather than a demo, that check-ins are planned at two weeks and two months, and that you can say no without penalty if the test doesn't convince you.