What Does Good Desk Utilization Look Like in 2026?
There is no single good number, but there are useful reference points. CBRE reports average building utilization of 53% and average peak utilization of 80% in its 2026 benchmarking, and JLL reports global office utilization of 56% in its 2026 report. Use those figures as context, measure your own building the same way every period, and set your target from your own busiest days.
Two reports on the same floor can disagree and both be right. A booking report counts intent. A sensor report counts presence. A badge report counts people who came through the door. This article separates published benchmarks from worked examples, defines the terms so you can read any report against them, and shows how to calculate and target desk use for your own building.
What do the published benchmarks say?
The table below collects the latest published figures from CBRE and JLL, with each study’s population and date. The worked examples later in the article use round numbers to show the math. They are not benchmarks.
| Source | Metric | Figure | Population and method |
|---|---|---|---|
| CBRE Published Jan 20, 2026 | Average building utilization | 53% (38% in 2024, 35% in 2023) | CBRE client data, 303 million sq ft, average portfolio 5 million sq ft; summarizes five years of benchmarking and sentiment surveys. CBRE’s public report pages do not define how utilization is calculated. |
| CBRE Published Jan 20, 2026 | Average peak utilization | 80%, against a target of above 65% | Same population as above |
| CBRE Published Jan 20, 2026 | Occupancy rate | 111% (101% in 2024) | Same population. CBRE defines it as more people allocated to buildings than there are physical seats, so it is an allocation ratio, not a count of people present. |
| CBRE Published Jan 20, 2026 | Utilization targets | 87% set explicit building utilization targets; 79% name a target above 65%; nearly half aim for 76% to 85% | Same population; share of survey respondents |
| CBRE Published Jan 20, 2026 | Target sharing ratio | 48% target 1.01 to 1.49 people per seat (21% in 2024); 33% target above 1.5 to 1 (62% in 2024); no respondent targets 1 to 1 | Same population; share of survey respondents |
| CBRE Published Jan 20, 2026 | Day of highest attendance | Tuesday 73%, Wednesday 23%, Thursday 3%, Friday 1%, Monday 0% | Same population; share of respondents naming each day as their highest attendance day |
| JLL Published May 2026 | Global office utilization | 56% (54% in 2025, 49% in 2024; pre-pandemic level 61%); gap between actual and target utilization of 18 points (25 points in 2024 and 2025) | Survey of 84 organizations, 175 region-specific responses, more than 716 million sq ft, 100+ questions. JLL states the report is not intended as a space utilization calculator. |
| JLL Published May 2026 | How utilization is tracked | Badge swipe data used by 92%; averaging daily peaks is the most popular calculation method (69%) | Same population as above |
| JLL Published May 2026 | Actual average seat sharing ratio (people per seat) | Latin America 1.5, Asia Pacific 1.3, North America 1.2, EMEA 1.2 | Same population as above, by region |
Three points keep these numbers honest.
- The sources do not match, and they should not. CBRE shows 53% and JLL shows 56%. They draw on different clients, different years of data and different definitions. Neither is a desk-level figure. Both describe buildings or portfolios.
- The peak sits far above the average. CBRE’s average peak utilization is 27 points higher than its average. CBRE notes that peak levels above target may mean offices are overcrowded at busy times.
- None of these is a target for your building. They describe what other organizations report. JLL says its own report is not a utilization calculator.
What do utilization, occupancy and booking rate mean?
Terms shift between reports and between vendors. These are the definitions used in this article. Check any report you read against them before you compare numbers.
| Term | Definition used here | Watch for |
|---|---|---|
| Desk utilization | Occupied desk-hours divided by available desk-hours, times 100. It measures presence across the whole period. | It needs presence data, such as desk sensors. Bookings and check-ins cannot supply it |
| Peak desk occupancy | The most desks occupied at the same time, divided by available desks, times 100. Say which period the peak covers: a day, a week or a month. | It is a snapshot of the busiest moment, not an average |
| Average of daily peaks | The mean of each day’s peak desk occupancy across a period. | JLL reports this is the most common calculation method. It is not the same as desk utilization |
| Booking rate | Booked desk-hours divided by available desk-hours, times 100. It records intent. | A booked desk may never have been used |
| Check-in rate | Bookings with a check-in divided by all bookings, times 100. It records an arrival signal at the start of a booking. | A check-in does not show how long someone stayed |
| Desk-sharing ratio | People assigned to a pool of desks divided by the number of desks. CBRE and JLL both express it as people per seat. | A ratio of 1.5 means three people for every two desks, not that 1.5 people are present per desk |
| Occupancy rate (CBRE) | People allocated to buildings divided by physical seats. | CBRE’s 111% is an allocation ratio, not a count of people present |
| Utilization in published reports | The share of available space or time used over a period, as each report defines it. | Reports that give a percentage without a unit or a period |
A booking is intent. A check-in is a signal. Neither is occupancy. Only a measurement of presence tells you a desk was used, and each measurement has limits of its own.
How do you calculate desk utilization?
Start with the data source, because it sets the question you can answer.
| Data source | What it measures | Level | Limit |
|---|---|---|---|
| Badge access | Someone entered the building | Building | Shows how many people came in, not which seats they used (JLL makes the same point). Misses anyone who enters without a swipe. |
| Desk or zone sensors | Presence at a desk or in an area | Desk or zone | Accuracy depends on the sensor type and placement. Zone counts describe an area, not a particular desk. |
| Wi-Fi | Devices connected to access points | Floor or zone | Counts devices, not people, and locates a device to an access point’s coverage, not to a desk. |
| Booking data | Who reserved what, and when | Desk | Shows intent, not whether the person came. |
| Check-in data | A person confirmed arrival at the start of a booking | Desk | Shows an event at the start of the booking, not how long the person stayed. |
JLL’s 2026 report says that knowing whether seats are shared requires space-level measurement, such as reservation data, sensors or video analytics, beyond building-level badge counts. With that in mind, the method is five steps:
- Fix the denominators. Count the desks in scope and decide the hours you measure, for example 8 a.m. to 6 p.m.
- Match the signal to the question. Use badge data for building attendance, sensors for presence, bookings for intent and check-ins for arrival.
- Calculate by day, not by week. Compute every metric for each day first, then roll up.
- Report average and peak by day of the week. State how the peak is defined.
- Put the signals side by side. Compare booking rate, check-in rate, desk utilization and peak desk occupancy for the same desks and the same hours.
Illustrative example: one floor on one TuesdayA floor has 100 desks, measured from 8 a.m. to 6 p.m. Ninety desks are booked for that whole period, so 90% of available desk-hours are booked: a booking rate of 90%. Seventy-two of those 90 bookings are checked in, a check-in rate of 80%. Sensors show that the most desks occupied at the same time was 60, at 11:00, a peak desk occupancy of 60%.
These three figures describe different signals: intent across the day, arrivals at the start of a booking, and presence at the busiest moment. They do not subtract. Ninety minus 60 does not mean 30 bookings went unused, because people come and go during the day and not every occupied desk has to be a booked one. To find unused bookings, compare each booking with presence at that desk during its booked hours. Report all three figures side by side rather than quoting the most flattering one.
Illustrative example: the same floor across a weekPeak desk occupancy by day is 30 desks on Monday, 60 on Tuesday, 50 on Wednesday, 40 on Thursday and 20 on Friday, out of 100. The average of those daily peaks is 200 divided by 5, or 40%. That figure is an average of peaks, not desk utilization, and it describes no single day. It hides Tuesday, when 60 desks were in use at once. Tuesday is the day that decides how many desks the floor needs.
What goes wrong when you measure desk use?
- Averaging away the peaks. A weekly average can look comfortable while one day is full. Report the peak next to the average every time.
- Ignoring the day of the week. CBRE’s respondents name Tuesday as the highest attendance day far more often than any other. A sample heavy on Mondays and Fridays pulls the average down. A Tuesday-only sample pushes it up. Measure several full weeks and report by weekday.
- Counting bookings as use. A booking nobody uses is a ghost booking, and it makes a desk look busy in the report. Use bookings to see intent, then test them against presence.
- Mixing levels. A building badge count, a zone sensor count and a desk sensor count answer different questions. Do not put them in one column.
- Changing the denominator. Dividing by desks, seats or assigned people gives three different answers. CBRE’s 111% occupancy rate exists because people allocated can exceed seats.
- Reading a short window. A holiday week, an all-hands event or one stretch of bad weather can distort a month. Check the period before you trust the figure.
- Treating a low number as an instruction to cut space. In his recap of Legal Workplace26 London, Simon Privett uses deliberately simple numbers to show how booking data and building entry data can tell different stories about the same office.
How do you set a target for your own building?
A universal good number does not exist. CBRE reports that 79% of its respondents target utilization above 65%, and nearly half aim for 76% to 85%. JLL’s gap between actual and target utilization narrowed in 2026 partly because organizations lowered their targets by 5 points. Targets move, and they reflect other organizations’ choices. Set yours from your own evidence:
- Measure first. Collect desk utilization and peak desk occupancy for the desks in scope across several typical weeks, by weekday.
- Find peak demand. Record the highest number of desks in use at the same time on your busiest day, by floor or neighborhood.
- Add a buffer on purpose. Agree how far above observed peak demand to plan, to cover normal variation, accessibility needs and people who need a desk near their team. Write down who decided it.
- Convert to a ratio. Peak demand plus the buffer tells you how many desks a pool of people needs, and from that the sharing ratio you can support.
- Compare with the benchmarks last. Use CBRE and JLL as a sanity check. If your ratio sits far from the 1.2 to 1.5 range in the table, ask why. There may be a good reason.
- Review on a schedule. Revisit the target when attendance policy, headcount or space type changes.
Illustrative example: from peak demand to a desk count150 people are assigned to a pool of 100 desks, a sharing ratio of 1.5. Peak demand on the busiest Tuesday is 60 desks. The team adds a buffer of 15% above that peak: 60 × 1.15 = 69 desks. That differs from leaving 15% of total capacity spare at peak, which would take 60 ÷ 0.85, or 71 desks rounded up. Say which one you mean. Either way, it is a calculation, not a recommendation. Before changing anything, check the other days, each neighborhood, accessibility needs and the mix of space types.
What can utilization numbers not see?
Booking, check-in and occupancy data describe how space is used. They say little about how well it works for the people using it. Gensler’s 2026 Global Workplace Survey, which polled 16,459 full-time office workers in 16 countries between July 22 and September 12, 2025, found that two thirds report “hacking” their workspace to compensate for gaps in how it performs, and one in four have resorted to DIY fixes for ergonomics, temperature or visual privacy. None of that shows up in a utilization report, which is why employee experience data belongs next to the numbers.
Where Desk Sign fits
Desk Sign helps at the point of decision, not the point of measurement. Connected to your desk booking platform, it shows each desk’s booking status at the desk, such as Available, Reserved or In Use, and where the platform and configuration support it, people can check in there. The booking platform remains the system of record.
Desk Sign does not continuously measure physical occupancy, and a check-in at the desk is still a check-in, not proof of presence. Desk utilization and peak desk occupancy still need presence data. What Desk Sign improves is the moment someone arrives on a floor and needs to know which desk they can use.
Frequently asked questions
Sources
- CBRE, 2026 Global Workplace & Occupancy Insights, Part 1, The Hybrid Reality: Why the Office Is More Important Than Ever (PDF), January 20, 2026
- JLL, Global Occupancy Planning Benchmark Report 2026: Measuring what matters, May 2026
- Gensler Research Institute, Global Workplace Survey 2026, fielded July 22 to September 12, 2025