Keepme.ai ads

I have been around long enough to remember the first time the business software industry made this mistake. In the mid-2000s, operators bought CRM, marketing automation, customer service and billing from four different vendors. The combined effect was that the same customer existed as four separate records in four separate systems. The market eventually consolidated around platforms because operators were tired of paying the integration tax. I am watching the AI agent space for fitness operators line up to make the same mistake. This week's newsletter covers why, including the five questions any operator should ask a vendor to test whether they are being shown a real platform or several agents in a trench coat. Steffie Bryant Hilary McGuckin MCIM Michelle Thavasi Jack Murray Chyna Blain Sherilyn Milch Tom Bohlin Jess Gronholm Krisha Tiwari Jasmin Mortensen Mantas Dvareckas

Retention has been a problem in fitness for decades. I have been working on it from the technology side for longer than most. In 2018, Keepme deployed a retention platform at over a thousand sites worldwide using random forest ensemble methodology, trained customer by customer on each operator's own data. Not a generic industry model. One built for each business, capturing its specific geography, pricing, and demographics. Accuracy ranged between 91% and 98%. The prediction worked. It still works. Eight years on, I have not seen another solution in fitness that comes close. But prediction was only half the problem. The other half was running continuous, adaptive engagement across an at-risk population at scale. Automation rules could not do it. Conversations require something that can adapt. That arrived with the agent platform. Ember on Antares now brings the prediction layer we built in 2018 together with an engagement layer that can finally act on it. Continuous, conversational, behaviourally informed engagement across every at-risk member, 24/7. The full piece is in this week's newsletter on why this is the first time both halves of the retention problem have been in the same system. Steffie Bryant Hilary McGuckin MCIM Jess Gronholm Jack Murray Jasmin Mortensen Mantas Dvareckas Charly Stenson Chyna Blain Sherilyn Milch Tom Bohlin Michelle Thavasi Krisha Tiwari

Two new agents shipped on Antares this month. Clarion and Atlas. The constellation is filling out nicely, and we're rapidly running out of celestial nouns — the next one might just be called Steve. (Do you have a name suggestion for the next one? 🤪 ) Clarion takes the phones. Inbound, round the clock. Opening hours, pricing, is there a pool, can I book a tour. And anything that needs a person, it routes to the right one. Atlas handles member services across text and voice. Class times, account queries, the bank holiday opening hours question that anyone who's worked a front desk knows by heart. Five agents now: Nova for sales, Beacon for AEO, Ember for cancellations, Clarion for voice, Atlas for member services. One platform, one contract, all five included. Last week I wrote about one of our multisite customers paying $0.26 an hour for an agent handling every inbound conversation across more than 60 sites. That number lands because every new site inherits what was already built into the first. People scale linearly. Agents scale logarithmically. The gap between those two curves is what compounds across a multisite estate. The same logic runs through the platform itself. When we ship the next one, it turns on alongside the others. No new procurement, no second integration, no fresh negotiation. The cost of adding capability keeps falling. But the economics is how this conversation opens. It's not where it lands. What I keep coming back to is what happens to your team when the transactional load comes off them. Front-of-house spends less time answering the same fifteen questions for the thousandth time, and more time sitting with the member who is thinking about leaving. The team stops firefighting and starts coaching. The operating model finally gives people room to do the work our industry has always needed them to do — the work that compounds into retention, into referral, into reputation. The team doesn't get replaced. The team gets given back to the member. Keepme.ai Mantas DvareckasIan MullaneSherilyn MilchJack MurrayJasmin MortensenChyna BlainHilary McGuckin MCIMKrisha TiwariWilli Trocio

$0.26 an hour. That is what one of our multisite customers pays for an AI agent that handles every inbound conversation across every channel a member or prospect uses to reach them. This is not a chatbot answering questions. It is an agent that books the tour, confirms the appointment, runs the trial follow-up, chases the no-show and attempts the cancellation save. It does the work, takes the action and owns the outcome, across thousands of concurrent conversations spanning more than 60 sites. A front-of-house team member at a multisite operator costs $22 to $28 an hour once you load in payroll tax, benefits, paid leave, sick cover, training and the management hours that go into hiring and replacing them. An AI agent on our platform costs around $4 an hour at a single site, runs every hour of every day in more than 200 languages, and holds the same standard at 2pm on a Tuesday as it does at 2am on a Sunday. What changes the conversation is what happens as you add sites. For a front-of-house team, every site is a near-linear cost, and serving members at site 50 costs about the same as serving members at site 1. On our platform, every new site inherits the intelligence already built into the first one, and the per-hour cost falls with every site you add. A 10-site operator runs materially below $4. By 60-plus sites, the cost lands in cents rather than dollars, which is where the figure at the top of this post comes from. People scale linearly. Agents scale logarithmically. The gap between those two curves is what compounds across a multisite estate, and it is the reason the labour conversation in our industry is about to change shape. The economics are how this conversation starts, but they are not the headline. The headline is what your team gets to do once the transactional load is off them. This is the part of the conversation I find genuinely exciting. Our industry has long asked front-of-house teams to be salespeople, schedulers, retention specialists, customer service operators and brand ambassadors on the same shift, for the same wage. We built operating models that asked too much of the people closest to the member and then wondered why service quality was inconsistent. Agents take the transactional load, and what that frees up is the work the industry has always needed its people to do and never had the operating margin to let them do properly. Sitting with a member who is thinking about leaving and actually understanding why. Coaching the new joiner through their first eight weeks. Building the floor culture that turns a gym into a club. The work that compounds into retention, into referral, into the reputation a brand carries in its market. Agents do not replace the team. They give the team back to the member. Keepme.ai Ian Mullane Chyna Blain Tom Bohlin Jack Murray Jasmin Mortensen Jess GronholmJoel O'SullivanKrisha Tiwari Hilary McGuckin MCIM Mantas Dvareckas

Last week I shared the finding that the global fitness industry is scoring an average of 21 out of 100 on AI search readiness across 901 operators in 27 countries. The response prompted more questions than I expected — about why the scores are so low, and what is actually going wrong on these websites. Before I get to the fix, those questions deserve a proper answer. This week I have gone into the detail of what we found: the specific failures, the patterns that repeat across operators of every size, and the one technical mistake that is costing operators visibility they do not know they are losing. Hilary McGuckin MCIM Mantas Dvareckas Steffie Bryant Jack Murray Jasmine Gan Chyna Blain Charly Stenson Sherilyn Milch Jess Gronholm Tom Bohlin Michelle Thavasi Krisha Tiwari

In March, at HFA, Jon Brady stood up at the CEO drinks and told the room that AI agents were already affecting his business. That comment sent me down a rabbit hole I have only just climbed out of. Three months and 901 audits later, the picture is uncomfortable. The fitness industry is, on average, scoring 21 out of 100 on AI search readiness. Not one operator in the world is fully ready. And some of the largest groups have rules in place that actively forbid ChatGPT and Claude from reading their websites. This part one of a two on Answer Engine Optimisation. Steffie Bryant Sherilyn Milch Ant Martland Chyna Blain Hilary McGuckin MCIM Mantas Dvareckas Jasmin Mortensen Tom Bohlin Jess Gronholm Michelle Thavasi Krisha Tiwari Jack Murray