Is Your Workspace Program Working? What 6,138 Company-Months Say Good Looks Like
Workspace programs are budgeted as though adoption ramps. Croissant tracked 115 company programs from their first month onward and found the opposite shape. Engagement holds steady for two years. What decays is whether the program runs at all, and by month eight the median program books nothing in a typical month.
Every workspace program is budgeted on the same assumption: that it starts slow, people discover it, and usage climbs. Rollout plans are written around that curve. So are the reviews, which is why the first serious look at the numbers usually happens at the twelve month mark, when the renewal is already on the table.
We tracked 115 company programs from their first month with at least five seats, across 6,138 company-months, and the curve is not there. Something else happens instead, and it happens early enough that a program can be read long before anyone schedules the review.
What does a healthy workspace program look like?
Start with the distribution, because most companies have no idea what normal is. Across 1,960 months in which a program recorded any bookings at all:
| Measure, in a month with any activity | 25th | Median | 75th | 90th |
|---|---|---|---|---|
| Share of seats that book | 12.5% | 22.2% | 40% | 60% |
| Company booking days in the month | 2 | 5 | 11 | 18 |
| Seats on the account | 6 | 9 | 16 | 25 |
If roughly a fifth of your seats book in a month, you are at the median. Two fifths puts you at the 75th percentile. That is the whole benchmark, and it is lower than almost every program owner expects, which matters because programs get judged against an imagined 80% and quietly written off at 25%.
Then there is the number that reframes all of it. Across all 6,138 company-months, including the quiet ones, 68.1% recorded no bookings at all. The median company-month is not a low-participation month. It is an empty one.
Programs do not fade, they switch off
Line the months up by program age, so month six means month six for every company regardless of when it started, and two measures move in completely different directions.
| Program month | Share of programs with any bookings | Median participation, all months | Median participation, when active | Booking days when active |
|---|---|---|---|---|
| Month 0 | 82.6% | 40% | 41.7% | 4 |
| Month 3 | 66.4% | 19.1% | 33.3% | 4 |
| Month 6 | 50.5% | 5.3% | 30% | 4 |
| Month 8 | 47.5% | 0% | 30.8% | 5 |
| Month 12 | 46.9% | 0% | 20% | 4.5 |
| Month 23 | 31.8% | 0% | 20% | 6 |
Read the last two columns against the first. When a program is running, it engages between a fifth and a third of its seats and produces four to six booking days a month, and that holds almost flat for two years. The intensity of a working program is durable.
What decays is whether the program runs at all. The share of programs recording any bookings falls from 82.6% to 31.8%. By program month eight, the median program books nothing in a typical month, and it never recovers across the remaining fifteen months we observed.
Programs do not wind down through declining engagement. They keep the same engagement and lose months. Which means the thing to monitor is not how much people are using workspace, but whether anyone is using it at all this month.
This is the program-level counterpart to something we found at the employee level: workspace demand is low frequency and highly concentrated even among people who actively want it. A program built to expect steady broad usage is measuring itself against a pattern that does not occur.
Onboarding is not adoption
The obvious explanation for a quiet program is that people never got access. The data closes that door firmly. Median activation, the share of seats onboarded onto the account, is 100% through program month eleven and never falls below 93.8% across the full two years. Access was granted, and it was kept.
It is worse than that for anyone using onboarding as a health metric. Among companies observable through their first year, those with activation at or above 90% in the first 90 days reached the healthy outcome 28.2% of the time. Those below 90% reached it 32.5% of the time. The lift is 0.9, which points mildly the wrong way.
Getting everyone onboarded is not evidence of anything. It is the most common thing a stalled program has already done.
Concentration arrives immediately and never changes
A reasonable hope is that a program starts as a small group's habit and broadens. It does not.
| Program phase | Companies measured | Median share of bookers producing half of all bookings |
|---|---|---|
| Months 0 to 5 | 55 | 25% |
| Months 6 to 11 | 30 | 25% |
| Months 12 to 23 | 26 | 23.3% |
Whatever concentration a program has, it has from the beginning. This sits alongside the employee-level finding that a median 15.4% of active employees account for half of all bookings, measured on a different unit and arriving at the same shape. Concentration is not a maturity problem to be fixed. It is what workspace demand looks like.
What predicts a program that lasts: the 90 day read
Of 111 companies observable through program month eleven, 29.7% reached a median of 20% participation or better across months six to eleven. Two things visible in the first 90 days separate them, and both earn their place on the negative case rather than the positive one.
| Signal in the first 90 days | Reached when present | Reached when absent | Lift | Share of successes caught |
|---|---|---|---|---|
| 5 or more company booking days | 37.2% | 12.1% | 3.1x | 87.9% |
| 2 or more distinct bookers | 32.6% | 12.5% | 2.6x | 93.9% |
Look at the third column rather than the second. A program that cannot produce five booking days or two separate people inside three months lands at roughly 12% against a 29.7% base rate. That is the useful part: the absence of early activity is a far more reliable predictor than its presence. It is the same shape as the market-level rule for deciding when a city has earned a fixed commitment, applied to the program instead of the market.
The bars are deliberately low. Five booking days in ninety, or two people, is close to the least a real program could produce. That is what makes the negative case strong, and it is also why we do not report a stricter version: demanding five or more early bookers actually weakens the signal, from 2.6x down to 1.3x on a smaller group. More early activity is not monotonically better, and we are not going to claim it is.
Where the threshold sits, and why we are showing you
The definition of a healthy program moves the base rate substantially, so here is the whole grid rather than the one number that flatters the argument:
| Definition, over program months 6 to 11 | Companies reaching it | Base rate |
|---|---|---|
| Active in 3 or more of the 6 months | 49 of 111 | 44.1% |
| Active in 4 or more of the 6 months | 44 of 111 | 39.6% |
| Median participation 15% or better | 40 of 111 | 36% |
| Median participation 20% or better | 33 of 111 | 29.7% |
| Median participation 25% or better | 27 of 111 | 24.3% |
Every figure above uses the 20% line. Move it and the base rate moves with it, between roughly a quarter and a half of programs. The direction of the finding does not change, but anyone quoting a single number should know which one they are quoting.
What buyers say about measuring a benefit
Workplace and finance leaders describe the same pattern in their own terms. These are illustrative quotes from recorded buyer conversations, not a measured sample, and they are included because they explain the behaviour rather than prove it:
On what finance actually watches: "Variance is one of the best indicators to look at for finance."
And on a comparable benefit, a learning budget, from a finance leader who does not police individual spend at all: "I don't even really track it to the individual. If they went over, they would have to go way over before I really looked into it. We're 50 to 60% under projection for that typically."
That last figure is worth sitting with. It is a different benefit, measured by the buyer rather than by us, and it lands in the same place: budgeted in full, consumed at a fraction, and nobody watching closely enough to notice which.
Four things to measure, and when
- Whether anyone booked this month. Not participation, just presence. Given that 68.1% of company-months are empty and the median program goes dark by month eight, this is the single measure that separates a working program from a dormant one, and almost nobody computes it.
- Participation among active months. Compare against 22.2% for the median and 40% for the 75th percentile. If you are inside that band your engagement is normal, and the question is continuity rather than depth.
- Booking days and distinct bookers at day 90. Five days and two people. Below either, you are on a path that reaches a healthy outcome about 12% of the time, and you have nine months of runway to change it.
- Concentration, once, early. Around 25% of bookers producing half the bookings is normal at every stage. It will not improve with time, so do not build a plan that assumes it will.
The instrument that makes all four of these cheap is treating workspace as a variable people-cost rather than a fixed property line. Under usage-based pricing an empty month costs nothing and shows up in the data. Under a per-seat arrangement it costs full price and shows up nowhere.
Methodology and limits
This analysis covers company-attributed usage on Croissant: 6,138 company-months at 115 company accounts holding at least five seats, drawn from 71,395 company-attributed bookings, from each company's first qualifying month on record through 1 September 2026. A booking is one coworking day visit, guest visit, meeting room booking, or private office booking. Cancelled bookings and sessions longer than 24 hours are excluded.
Program month 0 is a company's first month with at least five seats, not its first booking. Participation needs a seat denominator, so the clock starts when that denominator exists. Participation itself is bookers in a month divided by that same month's seat count, taken from weekly account snapshots rolled up to the month by their maximum. Measuring it per company-month rather than against a final snapshot matters: comparing a stale seat count against a multi-month booker count produces participation above 100%, which is meaningless.
Every maturity figure carries its eligible denominator. A company that started four months ago cannot have a month-twelve reading, so the cohort shrinks with program age, from 115 companies at month 0 to 97 at month 23. The shrinkage is reported rather than hidden, and the outcome analysis is restricted to the 111 companies observable through month eleven.
Concentration is reported by phase rather than by month because a single company-month rarely carries five bookers, below which a share-of-population figure is noise. Companies falling under the five-seat floor leave the qualifying set and count as not reaching the outcome, which is the intended reading, but it does conflate a program that shrank with one that ended.
Finally, this describes companies using Croissant, not companies in general. These are organisations that already decided to buy flexible workspace centrally, which is a selected population. Employees who also have a company office available may be using it on the days they do not appear here, so these figures describe the health of a flexible workspace program rather than total workplace attendance.
Measure Your Program Before Month Eight
Croissant reports participation, frequency, and concentration per team and per market on one account, across 800+ workspaces in 61 countries.
- ✓ Participation by team and market, computed for you every month
- ✓ Pay for booked usage, so an idle month costs nothing
- ✓ A 90 day read on whether a program is taking hold
