I Spent a Career Trying to Give You Back Your Data. This Week, Garmin Took Some of Mine.

On Friday I wrote a field note counselling patience. Garmin had just bought TrainingPeaks, and my shoulder had just handed me a rule for the month ahead. I’d had the injection last week, negotiated a few days of careful rehab, and mapped out a month of attentive training and climbing, nothing loaded before it’s ready. Don’t overreach. Push a joint too fast and you undo the patient work in a single session. So I applied the same logic to the news. Back up my files, look at the alternatives quietly, don’t cancel everything in one furious weekend. Go slow.

It’s Monday now. I spent the weekend turning it over, and I’m no longer convinced by my own advice. The patience I prescribed was starting to look like the comfortable option dressed up as the wise one, and I hold a principle, Adventure Over Comfort, that exists precisely to catch me doing exactly that. This is the kind of moment it’s for.

Because underneath the fitness story is an older one. I’ve spent the better part of two decades on the wrong side of a losing argument: that the most intimate data you generate, where you go, how your body performs, what your coach asks of you, should stay yours. Lent out when it’s genuinely needed, pulled back the moment it isn’t. On Friday, phone in hand and still sweating from that run, I watched the argument lose again. This time it reached my own athletic life.

I put real time and real money into that platform. Months trying to understand something genuinely complex, failing at parts of it, paying for the Uphill Athlete programs, not because I wanted another mainstream fitness app but because I was looking for a tribe. A small, serious gang who cared about proper training and not about boxes, badges, and upsells. That’s what drew me in.

The formal version: Garmin acquired TrainingPeaks, TrainHeroic, and TrainingPeaks Virtual (formerly indieVelo) from Peaksware Holdings.

I love Garmin. I’ve said so for years. But reading the news, the feeling wasn’t neutral. It was betrayal, the quiet kind I get every time something small and specific gets folded into something big and generic.

The pattern

This isn’t Garmin’s first rodeo, and that’s what tells you where it goes.

Officially, TrainingPeaks stays a separate business unit and remains brand-agnostic; Wahoo, Coros, and Suunto users can still sync. That’s the reassuring version. But the same reassuring version came with Tacx, the Dutch indoor-trainer market leader Garmin bought in 2019, undisclosed terms, “business as usual” messaging, fully absorbed within a couple of years. And again with Firstbeat in 2020, whose third-party licensing narrowed gradually until it stopped licensing to new customers altogether. Tacx, Firstbeat, now TrainingPeaks and TrainHeroic. Absorb, integrate, then narrow what used to be open, over time. No reason to assume this one plays out differently.

I understand why Garmin needs it. I’ve sat on the product side of a table like that. Apple isn’t a fitness accessory company anymore, the Ultra is a real endurance watch eating share; Suunto and Coros ship better hardware every cycle; Chinese and Korean manufacturers close the gap faster than anyone expected. Hardware margins alone don’t hold a company up, so the move is services. Connect+ is the revenue path, and TrainingPeaks’ coach-annotated compliance data, not just what an athlete’s body did but what a coach asked it to do and how well they complied, is exactly the asset that makes a premium tier defensible.

Strategically, it’s obvious. Understanding it doesn’t make it feel right.

What’s actually bothering me

It isn’t the open API, or paying more for Connect+. It’s the shape of what gets built when you combine these data sets.

Garmin already holds years of raw GPS and heart-rate data from people like me, thousands of activities across skimo, trail running, climbing, and cycling. Raw data is messy on its own. TrainingPeaks adds the missing layer: coach-annotated, longitudinal, structured, tracked over years. That combination, physiology plus intent plus compliance, is one of the most revealing personal datasets a person can generate. And once data like this exists in one place, anyone’s, the next question is always what trains on it. A model built on this kind of data, deployed with the wrong incentives or a badly specified objective, is no longer a hypothetical risk. That’s a worry about the architecture we’re all defaulting to, not a prediction about any one company.

We’ve normalised this everywhere at once, health, training, behaviour, consolidating into fewer hands, public and private, with less friction each time. If you want the public-sector version of the same story, I wrote recently about the UK’s NHS Federated Data Platform, run by a Palantir-led consortium, where NHS England’s own privacy assessment didn’t match who could actually reach patient data. Different institution, same shape.

I have been here before

None of this is new to me, and that’s exactly why it stings.

Years ago at Vodafone I worked on location services, when the industry was first waking up to what a phone in a pocket reveals. Location data is the most intimate telemetry there is. It doesn’t record what you say; it records where you sleep, who you visit, which clinic you walked into, how long you stayed. We understood even then that this was not “just metadata.” It was a map of a life. The question that consumed me was never whether we could use it, we plainly could, but whether any architecture let a person share the useful part without surrendering the whole map.

That question took me to digi.me, and to the most powerful and difficult product idea I’ve helped build toward: a personal data wallet with consent as the operating principle rather than the afterthought. Not the consent theatre we all click through, the twelve-page policy, the single irreversible “I agree.” Something atomic and continuous instead. Data that lives with you and leaves your wallet only when a specific use is relevant, for a specific purpose, for as long as that purpose lasts, and not one query longer. Consent you grant narrowly and withdraw cleanly, again and again, as your relationship with a service changes. The user as the point of integration, not the product being integrated.

It was cool, and it was hard, and it was hard for the reason it mattered: the entire commercial gravity of the industry pulls the other way. It is far easier to build a business on a pool of data you hold than on a stream a user lends you. Every acquisition this week is that gravity winning again. Garmin doesn’t want a relationship where my .FIT files leave a wallet I control only when a coach genuinely needs them. It wants the pool. Health, wellness, and the coach-athlete relationship itself, the three categories we agonised over at digi.me precisely because they are the most sensitive a person generates, all flowing into one reservoir. Held, not lent.

I still think that vision was right. Atomic, continuous, revocable consent, data leaving your wallet only when relevant, isn’t a niche privacy feature. It’s the only architecture that lets you keep your athletic life, or your medical life, as your own while still getting the benefit of sharing it. What I’ve watched for over a decade is the market choosing the opposite at every fork, and calling it convenience.

The harder test

So will I actually follow through this time, or is this just another paragraph I write and don’t act on?

I like to think I hold to principles like Adventure Over Comfort, the same conviction that had me pull back from Meta’s products entirely, WhatsApp included. That wasn’t a casual call; I set out the evidence in an earlier piece. Walking away wasn’t free, and it isn’t finished. I did it because the evidence didn’t leave me an honest alternative. The question this week is whether I extend the same discipline to Garmin, or whether love for the hardware quietly becomes the exception that proves I didn’t mean it.

What I’m doing about it

I can’t fix the consolidation. I can control my own copy of the truth, and keep building toward the version where the user holds the wallet.

1. Back up my .FIT files. Full local export, now, not “eventually.”
2. Diversify the toolkit. Intervals.icu, GoldenCheetah, Final Surge, real analytics, no lock-in.
3. Learn to parse raw data myself. Basic Python with fitparse, so my dashboards don’t depend on who buys whom next.
4. Keep building toward consent-first infrastructure. This is exactly the moment that brings the personal-data-wallet idea back into focus, the same one I chased at digi.me and Sir Tim Berners-Lee still champions with his data pods. It felt niche a few years ago. It doesn’t anymore.

I still love Garmin. I still think the strategy makes sense on a spreadsheet. But something about watching this happen, again, tells me the healthiest move is to make sure my data doesn’t need anyone’s permission to exist.

What about you: is your training history backed up anywhere that isn’t a corporate cloud you don’t control? If not, this is the week to fix that, not the week after the next acquisition.

Sources:

• Garmin acquires Tacx (DC Rainmaker analysis)
• No Surprises: The Caldicott Principles and the Gap NHS England Admitted on 15 June (fabrizio.deliberali.com)
• When Fiction Becomes Field Manual: Why We Need to Break Up with Big Tech (fabrizio.deliberali.com)

---

On my use of AI

The thinking here is mine. The reading, the connections, the argument about data consolidation, the line I draw from Vodafone location services through digi.me to this week’s acquisition, and the position on consent architecture that should be atomic and continuous rather than one irreversible click, all came out of my own reflection, my own reading history, and the governance work I do with clients.

English is my second language. I use AI the way I would use a good editor: to structure an argument I have already formed, to tighten sentences, to catch the errors that a non-native writer makes and often cannot see, and to check facts, dates, and attributions against sources before anything is published. Where a claim in this piece is checkable, it has been checked, and the sources are listed above so you can check them yourself.

What I do not do is ask a model to have the idea. If a sentence here makes an argument, I made it. If it draws a line between a mobile network’s location logs in the 2000s and a fitness acquisition in 2026, that line is mine.

I take full responsibility for everything published under my name, including any error that survived the process. If you find one, tell me and I will correct it in the open.