The promise on the timetable
We've gone from one gym to nine in a few years, and for a long time I told people it was a growth story. It wasn't. It was a story about a promise we were quietly failing to keep, and I didn't see it until it nearly cost us everything we'd built.
With one gym, my co-founder and I kept that promise with our phones and our memory. With nine gyms, the phones and the memory fell over, and we started breaking promises we didn't even know we were breaking.
People don't join Forge for the dumbbells; you can get dumbbells anywhere. They join for a specific instructor teaching a specific class at half six on a Tuesday, and they build their whole week around it.
When goodwill was the whole system
An instructor would get double-booked across two sites because two managers had grabbed her without talking to each other. A class would have no cover when someone was ill, so forty people turned up to a locked studio and a Post-it note. A new site would open with a timetable that was half guesswork.
What kept me up at night wasn't the logistics. It was what came after. In our world, a cancelled class isn't an inconvenience, it's the first domino in someone cancelling their membership. Somebody plans their Tuesday around a class, the class vanishes on them twice, and by the third time they've stopped planning their Tuesday around us at all.
For ages I treated the timetable as an operations chore, the boring bit you delegate and forget. It isn't the boring bit. It's the product and it's the retention engine, and we were running the single most important thing in the company on goodwill and a WhatsApp group. Saying that out loud in a board meeting was genuinely embarrassing.
A launch that went wrong on purpose
We brought in Northpeak Crew to run one timetable across every site, with instructor certifications built in, so it physically can't schedule a spin class with someone who isn't signed off to teach spin. When someone calls in sick, instead of a manager panicking down the phone, Crew shows them instantly who else is qualified and free, often at a site twenty minutes up the road. Cover went from a forty-minute crisis to a two-minute job. Pulse sits on top of that and gives me fill rates, cover and cancellations across all nine sites on one screen, so I can spot a studio starting to drift weeks before it ever shows up in the membership numbers. Pulse sits on top of that and gives me fill rates, cover and cancellations across all nine sites on one screen, so I can spot a studio starting to drift weeks before it ever shows up in the membership numbers.
It didn't all go smoothly. We opened our Liverpool studio in the middle of the rollout and got cocky, assuming the system would sort the opening timetable on its own. We hadn't put the local instructor data in properly, and launch week was a mess: mislabelled classes, gaps, the chaotic opening we'd been trying to leave behind, except this time I couldn't blame the old way of doing things.
We owned it, fixed the data, and learned the hard way that the tool amplifies whatever you feed it; garbage in, faster garbage out.
What changed
The next studio we opened, in Sheffield, went live with the timetable full and fully staffed from day one. It had the strongest opening month of any site we've ever launched. The Liverpool mess directly bought us the Sheffield success; we'd never have got the data discipline right without that scare.
Cancelled classes are now a handful a month. I won't publish the before-figure, but I will say it was embarrassing enough that I stopped putting it in board decks. Fill rate is up, because we're building timetables around what members want and who's strong at teaching it, not around whoever happened to be free. The number I really watch is thirty-day churn after a let-down: how many people cancel within a month of a booked class falling through. We only started measuring it because of all this, so I can't give you a clean before-figure. What I can tell you is it's now low enough to have stopped being the thing I'm frightened of.
The night Sheffield opened, my co-founder and I stood at the back of a packed class, neither of us on our phones fixing something.
We didn't start a company to administer spreadsheets; we started it to build rooms full of people who walk out feeling brilliant.
The one thing you can't wing
My advice to anyone scaling a multi-site business: find the one operational thing your customers actually feel, and make it bulletproof before you scale, not after. For us it was the class timetable. For you it might be something else entirely, but you'll know it, because it's the thing that, when it breaks, customers take personally.
Don't kid yourself that heroics and memory will scale the way they did with one site. They won't, and the day they finally fail, your growth quietly goes into reverse, with nobody noticing until it shows up in the numbers months later. Find that thing now, while it's still fixable on your own terms.