A gym never closes, so the overnight setback that most multi-site energy playbooks are built around does not exist here.
Take a standard energy program written for retail or quick-service and drop it onto a chain of fitness clubs, and the first three recommendations fail on contact. Set back overnight. There is no overnight. Match the schedule to store hours. The doors never lock. Reduce ventilation during low-occupancy periods. Occupancy is the one thing you cannot compromise in a room full of people breathing hard.
The load profile is genuinely different, and it needs its own approach. Here is what that looks like.
Three things stack.
Operating hours are the obvious one. A 24-hour club runs its HVAC, lighting, and ventilation continuously. A retail store with a twelve-hour day gets half the year off by comparison.
Ventilation is the less obvious one and often the larger cost. Occupant density in a fitness floor or a group exercise studio is high, and the metabolic rate is several times what a code table assumes for a retail sales floor. That drives outside air requirements up, and outside air is expensive to condition because you are heating or cooling ambient air rather than recirculating air you already paid for.
Then there is the swing. A club can go from near-empty at 2pm to packed at 6pm and back to a handful of people at 11pm. Equipment sized for the peak spends most of the day oversized for the actual load, which is where short cycling and poor part-load efficiency show up.
For scale, a 24/7 mid-sized facility commonly runs $5,000 to $8,000 per month in utilities, and larger operations average roughly $63,000 to $68,000 annually per site.
Five things, in rough order of how often they turn out to matter.
Runtime by unit. Not temperature, runtime. A rooftop unit running 22 hours a day in a shoulder-season month is telling you something that a comfortable indoor temperature will hide completely.
Actual space temperature against setpoint, per zone. The gap between what the thermostat is asking for and what the room is doing is the single most useful diagnostic signal in the building. A widening gap means the equipment is losing ground.
Override frequency and duration, by site and by time of day. Covered in its own section below, because at a gym this is the number that predicts everything else.
Outside air damper position and economizer function. An economizer stuck closed means you are mechanically cooling on a 55-degree morning. An economizer stuck open means you are conditioning the outdoors in August. Both are common, both are invisible without instrumentation, and in a high-ventilation building both are expensive.
Interval electrical data, ideally at 15-minute granularity. Demand charges are set by your highest short interval, not your total consumption, and a club's evening peak is exactly when that interval gets set.
Because the person overriding the thermostat is usually right.
At a retail store, an override is often somebody being careless. At a gym, an override is a staff member responding to a member who just told them the room is too hot during a spin class. That member has a month-to-month membership and a competitor two miles away. The staff member is doing their job.
This is the structural difference that gets missed. A comfort complaint at a retail store is a slow margin problem. A comfort complaint at a gym is a churn signal, and churn is the number the operator is actually measured on.
So a policy that locks the thermostat and calls the problem solved will either be circumvented within a month or will cost memberships. Neither is a win.
The approach that holds is narrower. Give local staff a bounded adjustment range they can use without asking, so the response to a legitimate complaint is fast. Log every override with its time and duration. Then read the log as data rather than as a compliance problem, because a site that overrides every Tuesday at 6pm does not have a discipline problem, it has an equipment or airflow problem that shows up under load, and the override is the symptom telling you where to look.
Fix the underlying cause and the overrides stop on their own. That is the only version of this that survives contact with a real club.
If you cannot set back on time, set back on load.
A 24-hour club is not equally occupied for 24 hours. Between roughly 11pm and 4am, a location that is technically open may hold a handful of members. The building does not need the same conditioning it needs at 6pm, and the difference is recoverable without ever going dark.
Practically, that means a modest overnight adjustment rather than a deep setback, timed to actual traffic rather than to the clock, and paced on recovery so the building is back at full condition before the morning arrival rather than fighting to catch up at 5:30am.
Recovery pacing matters more across a portfolio than at a single site. If forty locations all recover to occupied setpoint at the same moment, they can build one coincident demand peak that costs more than the overnight adjustment saved. This is the specific problem GlacierGrid's Smart Setpoints and Adaptive Recovery is designed for.
This vertical has less published benchmarking than restaurants or convenience retail, so set expectations accordingly.
Across GlacierGrid's customer base, which spans quick-service restaurants, convenience stores, and retail rather than fitness, the published figures are roughly 10% energy savings, about a one-month payback, and 15% fewer service calls. Those are cross-vertical numbers, a reasonable starting frame for a fitness portfolio and not a fitness measurement.
Long operating hours should move that estimate up, because more hours means more hours in which a correction applies, and high ventilation load makes economizer faults costlier when they occur. Less refrigeration load and less room for deep overnight setback move it down.
If a vendor quotes you a gym-specific savings percentage, ask how many fitness locations it was measured across. That question applies to us too.
Pick your three highest-utility-cost locations and pull thirty days of runtime, space temperature against setpoint, and override logs. Look for the sites where overrides cluster at a consistent time of day. That cluster is where your equipment is failing under load, and it is almost always cheaper to fix than it looks.
Start a free 90-day pilot and get that view across your locations without committing to anything.