Every treadmill is a starting line.

It’s 6:40 on a Tuesday evening at a Planet Fitness off a suburban highway, and the treadmill row is nearly full. Twelve machines, twelve people, twelve small TV screens. A woman on machine 4 is on the ninth day of her membership. She’s walking at 3.2 miles an hour and watching a cooking show.

The man two machines down is running at 7.5. The teenager at the end of the row is on a summer pass and alternates between jogging and checking his phone. None of them know the others exist, and nothing about the next 30 minutes will be different from the last time they came, or the time before that.

Plenty of new members start on a row like this one, and it’s a lonely place to spend your first month at a gym. I think it could be a race.

What Planet Fitness is already saying

Planet Fitness is unusually candid about its problem. Monthly attrition averaged 3.8 percent in the first quarter of 2026 and 3.5 percent in the second, and management described the first quarter’s as “higher-than-expected attrition.” In May the company withdrew its three-year outlook.

The company has also said where it thinks the answer is. “We know that the first 100 days of membership influence long-term retention… early engagement, both digitally and in club, contributes to higher lifetime value,” chief executive Colleen Keating told analysts in February. Planet Fitness is building a first-100-day program with its franchisees, testing an AI model to predict who’s about to cancel, and relaunched its app on September 9 with a home screen that changes when a member walks into the club.

The newest members are the hardest to keep. Gen Z is the company’s fastest-growing group of joiners, and on the first-quarter call management noted that “younger consumers historically churn more than older cohorts.” In 2025, 3.7 million teenagers used the free High School Summer Pass, and about 8 percent became paying members.

What none of that gives a new member is a reason to come back on a Tuesday night.

What I’d build

I should say up front that this part already exists. My studio, Volt, makes games that are played by moving, and one of them, Volt Derby, is a combat racing game controlled by a treadmill. Your belt speed moves a cartoon runner around a track through a jungle, a city at night or a glacier. You change lanes to grab item boxes, fire a bolt at the runner ahead, and surge past the pack. It runs in a phone’s web browser, with no app and no account.

In a Planet Fitness, it would work like this.

1. 01 · ScanA sticker on the consoleA QR code per machine opens the race in the phone’s browser and knows which treadmill it’s on.
2. 02 · ReadThe belt, not the memberA small clip-on sensor reads belt speed over Bluetooth. Without one, the member taps plus and minus to match the console.
3. 03 · RaceThree to 12 minutesRivals, the ghost of your best run and, in time, the other members on the row.
4. 04 · ReturnA season, not a sessionResults, streaks and a ten-race starter season land in the member’s profile.

The phone sits on the console’s ledge, where most people’s phones already are. Every control is a tap: lane left, lane right, use the item. The game only reads the treadmill. It never changes speed or incline, and getting hit by a rival slows the cartoon runner, never the belt. Nothing starts until the belt is moving, and the first screen reminds people to clip on the safety key.

Why a walker can beat a runner

A leaderboard ranked by raw speed would be the worst possible thing to put in a Judgement Free Zone. The woman walking at 3.2 miles an hour would lose to the man at 7.5 every single time, and she’d be right to stop coming.

So nobody races on speed. Each player sets a comfortable pace, and the game measures their belt against it. The walker at 3.2 and the runner at 7.5, each holding their own pace, move at the same speed on screen. Going faster than your own pace charges a surge. A beginner who pushes from 3.0 to 3.4 miles an hour is working harder, relative to herself, than a regular cruising at his usual speed, and the race reflects that.

This isn’t a new idea in gyms. Myzone, whose heart-rate belts are common in boutique studios and YMCAs, scores effort as a share of each person’s own maximum heart rate for the same reason. It fits the members Planet Fitness says it wants: first-time gym users, people following its own 12-3-30 walking workout, and a growing older base.

Why now

Treadmill racing has gone mainstream this year, on other companies’ machines.

Every one of these is tied to its own hardware except the last. That matters for a franchise system that can’t change its treadmill software on its own schedule.

What it’s worth

Planet Fitness had 21.5 million members at the end of June, and its annual report puts average monthly dues at $19.51. That makes each tenth of a percentage point of monthly attrition worth a lot.

My arithmetic on reported figures, not a Planet Fitness number: 21.5 million × 0.1% × $19.51.

A game won’t move the whole membership, and it shouldn’t have to. The first place to look is new members in their first 100 days, which is where the company itself says the answer lies.

The objections

The treadmills can’t talk to a phone. Probably true, at least for now. Everything public about Planet Fitness’s fleet points to Matrix treadmills with a custom console. Matrix documents open Bluetooth speed data on its touchscreen consoles, but its simpler LED consoles list Bluetooth only for heart-rate straps. That’s why the pilot uses a clip-on belt sensor, which reads speed from any treadmill and costs far less than the machine it sits on. A native connection belongs in a later conversation with the equipment maker, timed to the next replacement cycle.

Phones on a moving belt are a hazard. The game is built for it. There’s nothing to steer, no gestures, and nothing asks anyone to sprint or stop suddenly. Its large numbers are readable at arm’s length, and it pauses when the belt stops. The industry norm is to leave speed control alone, and this never touches it.

Leaderboards feel intimidating. They do when they rank people by speed. These rank effort against each person’s own pace, are opt-in, and hide names by default.

It’s one more thing for club staff. The sensors ship configured, the stickers go on once, and the pilot measures staff time as a pass-or-fail condition.

How I’d find out if I’m wrong

I’d test it in stages, each with a number that can end it.

• First, two weeks without asking anyone for anything. Check what a Planet Fitness console actually broadcasts, and test the sensor against a treadmill display at home.
• Then two clubs, four to six weeks, 20 to 40 treadmills. It passes if belt speed reads within 3 percent of the console, at least 95 percent of members pair on the first try, there are no incidents, and staff spend less than an hour a week on it.
• Then 8 to 10 clubs against 8 to 10 matched control clubs, for 16 weeks. January through April, when the most people join. The one number that matters is 100-day attrition among new members who were invited to race, compared with the controls. If it doesn’t fall by at least 15 percent in relative terms, we stop.

The control clubs matter because people who opt into a game are already more engaged, so a before-and-after comparison would flatter the result. Detecting a drop from 20 percent to 15 percent takes roughly 900 new joiners in each group, which a handful of clubs can supply in a winter.

If it works, the next step is a region, then a Summer Pass league whose final lands right before the fall push to convert teenagers into paying members. If it doesn’t, the clubs keep a few dozen belt sensors and a clear answer, and it cost a season to find out.

The woman on machine 4 is still walking at 3.2. On day nine, she could be in third place.