A wellness benefit is working when people keep using it — not when they sign up for it. The report that survives a finance review therefore carries five numbers and nothing else: how many people are eligible, how many activated, how many actually used it in the period, how often the active ones came back, and how all four moved since last quarter. Everything else is decoration, and anything more granular than that is a data-protection problem you do not need.
Sqwod Pod is a private, app-booked gym in Berlin with locations in Weißensee and Kollwitzkiez (Prenzlauer Berg) — individual fully-equipped private rooms of roughly 30–50 m², booked by the hour. We sit on the supply side of corporate wellness programmes, which means we see the usage data behind them. This guide is about measuring the thing honestly, whoever provides it.
What is the difference between sign-up and usage?
Roughly a factor of four, on the national numbers.
The BIBB/BAuA Faktenblatt 41, published by Germany's Federal Institute for Occupational Safety and Health on a representative survey of 20,012 employees, found that about 46% of employees were offered workplace health promotion measures by their employer — and that 59% of those offered took part. Multiply the two and roughly 27% of German employees actually participated. That last figure is our own arithmetic on the two published numbers, not a figure BAuA states directly, but the direction is unambiguous: the gap between "we offer it" and "they use it" is the whole story.
This matters because the two numbers get reported interchangeably, and only one of them is evidence. "We rolled out a fitness benefit to 180 people" describes a procurement decision. "Fifty-one people used it at least once last quarter, averaging 2.4 sessions each" describes an outcome. Finance can act on the second. It cannot act on the first, which is why the first is usually the one that gets presented.
Which numbers belong in the report?
Five, in this order. Each one answers a question the previous one raises.
| Metric | Definition | What it tells you |
|---|---|---|
| Eligible headcount | Everyone entitled to the benefit in the period | The denominator. Fix it once so quarters stay comparable. |
| Activation rate | Share of eligible people who signed up, registered or claimed access | Whether your communications reached people at all |
| Active-user rate | Share of eligible people who used it at least once in the period | The headline number. This is "utilisation". |
| Frequency per active user | Sessions ÷ active users | Whether it became a habit or stayed a novelty |
| Cost per active user | Total spend in the period ÷ active users | The number finance will calculate anyway. Bring it yourself. |
Two design rules make the set defensible. Set the period before you look at the data — a quarter is usually right, because a month is too noisy and a year hides the drop-off. And report the trend alongside every figure. A 28% active-user rate is meaningless on its own; a 28% rate that was 19% last quarter is a programme working, and one that was 41% is a programme dying.
How do you report usage without touching health data?
By never holding individual-level data in the first place. Aggregate, anonymous reporting is not a compromise here — it is both the legally clean option and the one finance actually wants, because nobody in that meeting needs to know who trained.
The legal position is unusually simple. Under Recital 26 GDPR, the data protection principles do not apply to anonymous information — data that cannot be related to an identified or identifiable person, or that has been anonymised so the person can no longer be identified. Genuinely anonymised aggregate usage falls outside the regulation altogether, including for statistical purposes.
The practical question is therefore how small a group can get before "aggregate" stops being anonymous. There is no fixed statutory answer, but supervisory authorities have published working numbers. The Baden-Württemberg data protection authority, in its Ratgeber Beschäftigtendatenschutz (4th edition, April 2020), writes that a minimum group size of three or five persons seems clearly too small and recommends evaluating only from seven responses upward — rising to twelve or more in medical research. The passage is quoted in full by the data protection practice DSN Group in its note on statistical evaluations of employee data, which also makes the sharper point: a statistic with no names in it is still not anonymous if the surrounding detail identifies someone.
Health insurers apply the same logic at a larger scale. Several will only issue an anonymised company health report where at least 50 employees are insured with them — Mobil Krankenkasse states this threshold openly as a data-protection precondition. If a provider offers you a departmental breakdown of a twelve-person team, that is not a better report. That is a report you should refuse.
Three rules follow, and they fit on one slide:
- Company-wide totals by default. Break down by site or department only where the group clears your minimum size.
- No health data, ever. Usage of a room is not a health record. Keep it that way by never collecting what was trained, how heavy, or how the person felt.
- Suppress small cells rather than publishing them. "Fewer than 7 — not reported" is a perfectly good line in a table, and a far better one than a number that names someone by implication.
Does the works council need to be involved?
If there is one, involve it early — and do it before you choose the provider, not after.
§ 87 Abs. 1 Nr. 6 BetrVG gives the works council a co-determination right over the introduction and use of technical devices suited to monitoring the behaviour or performance of employees. German case law reads that broadly: what matters is whether a system is objectively capable of such monitoring, not whether the employer intends to use it that way. A booking platform that can produce a per-person attendance list is objectively capable of it, whatever your reporting policy says.
That is a reason to design the reporting narrowly, not a reason to avoid the conversation. Walking into a works council meeting with "here are the five aggregate figures we will see, here is the minimum group size below which nothing is reported, and here is the confirmation that no individual-level data reaches HR" turns a contested topic into a short agenda item. Walking in after the contract is signed does the opposite.
This is general information on benefit administration, not legal or tax advice. Have your data-protection officer and, where applicable, your works council review the specific setup before rollout.
What does finance actually want to see?
Cost per active user, a trend, and an honest statement of what the number cannot prove.
The temptation at this point is to reach for a return-on-investment figure. There is a credible German one: the iga.Report 40, published jointly by BKK Dachverband, DGUV, vdek and IKK e. V., re-examined the evidence for 2012–2018 across 49 high-quality systematic reviews containing more than 900 individual studies and confirmed the positive cost-benefit picture, with the average return on investment of 2.7 established in the preceding iga.Report 28.
Use it as context, not as your result. That figure describes structured workplace health promotion in general, across many employers and many programme types — it is not a forecast for your line item, and it depends entirely on participation. Presenting an ROI that rests on 27% participation as though it rested on 100% is the fastest way to lose the room. Presenting your own cost per active user, next to that benchmark, with the caveat stated out loud, is the way to keep it.
For the spending side of the same conversation — the €50 monthly Sachbezug limit, the €600 annual allowance and how the models compare — see what workplace fitness actually costs a Berlin employer.
What do you do if the number is bad?
Report it, then read it as a design signal rather than a verdict on the idea. Low usage is nearly always structural, and the structure is usually one of three things.
- Distance. A benefit with a detour is a benefit used in January. In a city like Berlin, one location people can reach in fifteen minutes beats a directory of ten thousand they never open.
- Queueing. Anyone with a 45-minute gap between meetings will not stand in line for a squat rack at 6pm. Peak-hour congestion turns a benefit into a scheduling problem.
- Intimidation. A large share of the people you most want to reach will not walk onto a busy training floor at all. Programmes built for the already-fit look fine in a headcount and change nothing.
All three are measurable, which is the useful part. Split your active-user rate by site and by time of day and the cause usually announces itself. We wrote up the design side of this in how to build a workplace gym employees actually use, and the wider Berlin programme picture in our guide to corporate wellness in Berlin.
Frequently asked questions
How do you calculate benefit utilisation rate?
Divide the number of eligible employees who used the benefit at least once during the period by the total number of eligible employees, then multiply by 100. Keep the denominator and the period fixed across quarters so the figures stay comparable, and report activation (sign-ups) separately — the two measure different things.
What is the difference between enrolment and utilisation?
Enrolment counts people who signed up. Utilisation counts people who actually used the benefit in a given period. The gap between them is usually large: BIBB/BAuA found that 46% of German employees were offered workplace health promotion and 59% of those took part, which works out at roughly 27% of all employees participating.
Can an employer see who used a wellness benefit?
It should not, and in most setups it does not need to. Aggregate anonymous reporting keeps the data outside the scope of the GDPR under Recital 26, and where a works council exists, § 87 Abs. 1 Nr. 6 BetrVG gives it co-determination over systems objectively capable of monitoring individual behaviour or performance. Design the reporting to be aggregate-only and agree the minimum group size in advance.
How small can a reporting group be before it stops being anonymous?
There is no fixed legal threshold. The Baden-Württemberg data protection authority considers a minimum group size of three or five clearly too small and recommends evaluating only from seven responses upward, with twelve or more in medical research. Several health insurers require at least 50 insured employees before issuing a company health report at all.
What is a realistic utilisation rate for a fitness benefit?
There is no reliable published benchmark for fitness benefits specifically, and the figures circulating on provider websites are not traceable to primary research — so treat your own trend as the benchmark instead. Measure one quarter, change one structural variable such as location or booking friction, and measure the next.
Want a usage report you can take to finance?
Sqwod Pod runs private, fully-equipped training rooms in Weißensee (Jacobsohnstraße 27, 13086) and Kollwitzkiez (Kollwitzstraße 20, 10405), booked by the hour through the app. Because the booking happens per hour rather than per contract, the usage data is a direct record of what your team did — reported as aggregate, anonymised figures, with no individual health data anywhere in the chain.
See the Company Wellness programme →
Ask us for a sample usage report →
Last updated: 21 September 2026.
