Protocol v1.0, 12 October 2026
The pilots Kalisme offers to organisations test seven hypotheses, written before the first measurement, each with the threshold that would prove us wrong. Their results will be published, whether they prove us right or wrong, aggregated across all the pilots and without ever naming an organisation.
| Hypothesis | Measure | Refutation threshold | When it is measured |
|---|---|---|---|
| H1 · Activation. A voluntary offer, announced by the organisation, is activated by a significant share of the people it is offered to. | Share of seats activated | Refuted if fewer than 40% of seats are activated at 30 days. | 30 days |
| H2 · Coming back within the week. Someone who opens Kalisme comes back to it within seven days. | Questionnaire (self-reported) | Refuted if, in the 30-day questionnaire, fewer than half of the respondents who say they opened it say they came back within the week. | 30 days |
| H3 · Repeated use. Use lasts beyond the first discovery, without being daily. | Questionnaire (self-reported) | Refuted if, in the 90-day questionnaire, fewer than half of the respondents who say they used it report at least two uses over the past month. | 90 days |
| H6 · Adoption. The fact that the tool is offered by the employer does not put people off trying it. | Questionnaire (self-reported) | Refuted if, in the launch questionnaire, those who say they are “less inclined” outnumber those who say they are “more inclined”. | Launch |
| Hypothesis | Measure | Refutation threshold | When it is measured |
|---|---|---|---|
| H4 · Perceived value. The scheme is understood to be private, and judged useful by those who have tried it. | Questionnaire (self-reported): average trust and relevance | Refuted if, in the 90-day questionnaire, average trust is below 4 out of 5, or average relevance below 3.5 out of 5. | 90 days |
| Hypothesis | Measure | Refutation threshold | When it is measured |
|---|---|---|---|
| H5 · Willingness to pay. An organisation that has seen the pilot continues as a paying client. | Paid continuations | Refuted if fewer than 2 of the first 5 pilots continue as paying clients. | Review of the first 5 pilots |
| H7 · Who pays, and why. The budget is held by human resources or occupational health, on the grounds of care for people, not of a performance indicator. | Reason given for continuing | Refuted if the organisations that continue justify it first by a performance or absenteeism indicator, or by a budget with no owner. | Decision to continue |
Publication rule
No results are published before at least three pilots have ended. Each organisation may refuse to have its data included in the aggregate, without having to give a reason.
What is measured: the share of seats activated at 30 days, what people report in a voluntary, anonymous survey in three waves (at launch, at 30 days and at 90 days), and whether the organisation wishes to continue. Coming back to the app, and how often, is measured by what people report, never by tracking their use.
The organisation sees which seats are activated, never what is done with them. From the survey, it receives only an aggregated summary, indicator by indicator, never below 10 responses.
Two short scales, one of well-being and the other of perceived stress, are added to the launch and 90-day questionnaires only once their usage rights are confirmed. They carry no hypothesis about effects, and the organisation receives nothing from them: their average appears only in the publication covering all the pilots, without naming any organisation.
Before launch, each organisation writes three to five criteria with Kalisme, read once only, at the review, against the 90-day report. If they are met, the organisation is committed to continue as a paying client, on at least the pilot’s seats and never fewer than 20. A criterion that cannot be read counts as not met, and stopping after the 90-day wave has opened counts as the review: the criteria are read all the same. The proposed thresholds:
None of the following can be a criterion: an activation rate on its own, a volume or frequency of use, a change in the well-being or stress scales, a human resources indicator (absenteeism, sick leave, turnover), or any measure concerning a team, a site or a person.
Hypotheses published on 8 October 2026. No results have been published yet.
aee6d2c57cba7fc7c14c0cdc481ab065d5cb9774b7060fd9ce586d4879f5e07dc9924001cb46d6a8e94aad6aed9fa33417ac39ddfff04effb657bb5e5ad65f8eTo check a downloaded file, recompute its fingerprint with shasum -a 256 Kalisme_Protocole_pilote_v1.0_EN.pdf and compare it with this one.
The French version is authoritative.