Here is how I’d fix it. Before you worry about whether your fitness tracker is spying on you, ask the question I keep coming back to: what is that step count actually for now? For years it was a personal record — a number you showed your friends and your future self. This year it started becoming something else: an input to a price.
The signs are all over the mid-year earnings releases, if you read them the way you’d read a service manual. A major fitness platform told investors in August that it is working with insurers and other third parties to turn exercise data, activity risk assessment and AI capabilities into tailored products. A big gym chain signed a partnership with a health insurer in June around exercise-linked health benefits and member incentives. Behavior-based pricing in health coverage is no longer hypothetical; it is already on the market. Strip out the marketing language and the signal is blunt: the thing on your wrist is now part of the underwriting conversation.
Now, my first reaction was the one you’re probably having — fine, so I turn off the sharing toggle and go back to counting steps in peace. But here’s the thing: that is not quite the fix, and I need to correct myself before we go further. The problem was never really the collection. It is the asymmetry. The insurer or the app knows what your data is worth, how it will be combined with other data, and what it predicts. You mostly know that it exists. Shutting it off protects your future; it does nothing for the ten years of history already sitting in the ledger.
Let me back up and show my reasoning, because this caught me off guard at first. I started writing this as a privacy piece — consent forms, data brokers, the usual suspects. But the more I looked at how these products are actually structured, the less that framing held. A privacy breach is a leak. What we have here is closer to a swap: you hand over a stream of behavior, and you get a discount, a badge, a premium inside a premium. Nobody is stealing anything. The mechanism is the point.
Here’s the concrete moment that made it real for me. A friend renewed his health coverage last month and the questionnaire had a new section: how many times per week do you exercise, and can we verify it through your activity records for a lower rate? He laughed it off and said no. Then he called me the same night to ask whether he’d just locked himself out of a discount he was entitled to. That is the feeling you need to get used to — not betrayal, exactly, but the sense that a quiet negotiation is happening around you, and you weren’t sure you were at the table.
The mechanism deserves a closer look, because this is where the hands-on thinking starts. An insurer does not care whether you walk ten thousand steps for its own sake. It cares about what your movement predicts: illness risk, hospital visits, claims. The logic is actuarially sound, which is exactly why it is dangerous to misunderstand. If step counts genuinely predict lower claims, then pricing on them is rational, and the person who disagrees with the price is simply a worse risk in the insurer’s eyes. You’ll see this argument made with perfect confidence in investor calls. What is rarely said out loud is the second-order effect.
Here’s how I’d test the actuarial logic before accepting it, the way you’d test any new tool before trusting it with real work. Ask what the model would do with a runner who gets injured and stops logging for six months — a perfectly healthy person on any reasonable measure, but a blank line in the data. Ask what it would do with someone who walks for work all day versus someone who jogs for fun. The point is not that the models are rigged. The point is that the data is a partial map of the territory, and the pricing treats the map as if it were the land itself. You’ll see the gap the moment you ask a question the data was never built to answer.
The second-order effect is where I had to stop and think for a while. When behavior becomes a price signal, the incentive is no longer to be healthy — it is to be healthy in a way the algorithm can see and verify. A thirty-minute walk you forgot to log does not count. A dance class in a room with no tracking does not count. The data becomes the reality, and the reality outside the data slowly stops mattering. I’ve seen this pattern before, in a dozen different tools, and the feel of it is always the same: the measurement quietly becomes the goal. That is the point where I stop being enthusiastic and start being careful.
So what do you actually do about it, if you don’t want to live in a bunker? Here is how I’d approach it, in three passes, the way I’d tune a machine.
Pass one: audit your consents like you’d audit a work order. Open the settings on every app that touches your movement — the tracker, the gym chain’s app, the rewards program, the food logger, the one you forgot you installed. You’ll see a wall of toggles, and the pattern is usually the same: data shared with affiliates, data shared for ‘analytics’, data shared with ‘partners’. Read the partner list. If an insurer appears, or a health data aggregator, treat it as a pricing input with your name on it. You are not closing a leak; you are closing an input you didn’t price.
Pass two: understand what a discount is actually buying. A behavior-based discount is not a gift; it is a selection mechanism. The insurer is sorting customers into risk tiers, and your data is the sorting key. The honest question is not whether the discount is fair — it is whether you’re comfortable being sorted in real time, year after year, by a model you’ll never see. Some people will find that trade fine. I do not think less of them. But make it a conscious choice, not a default you drifted into because the toggle was green.
Pass three: keep a record of your own. The most practical thing you can do is maintain your own health data outside the apps — a simple spreadsheet, an annual checkup summary, a note on your resting heart rate over time. This is not paranoia. If behavior is going to be priced, the person with the fullest record negotiates from a different place than the person who only exists inside an app’s export file. You’ll see the advantage the first time you dispute a rate and can walk in with your own numbers.
Where is this heading? Here’s how I’d read the next three years, for what that’s worth. The behavior-pricing products will get more sophisticated, the health pictures will get fuller, and the regulatory questions will multiply — the agencies that watch consumer protection will eventually have to decide how much of a discount a step count can buy, and what a carrier may and may not ask. The direction of travel is already set by the partnerships announced this summer; the speed is the only unknown. If you build the habit of owning your data now, you will not need to make any of those decisions under time pressure later.
There is one thing I have not mentioned, because it only became clear to me while writing this. The people who will navigate this shift best are not the privacy maximalists and not the early adopters; they are the ones who treat their health data the way a careful owner treats a machine — knowing the manual, keeping the log, catching the small deviations before they become expensive problems. That is a maker’s temperament, and it turns out to be exactly the temperament the new data economy rewards. The irony is not lost on me: the people who never wanted their lives reduced to numbers are often the ones most prepared to meet an economy that runs on them.
Let me be honest about the limits of all this advice, because I don’t have this fully figured out either. I do not know how far behavior pricing will go, whether regulators will cap it, or whether the discounts will actually be worth anything in ten years. I was skeptical of the whole trend when I first saw the partnership announcements — my instinct said overhyped, a press release wearing a business model. Then I watched the rate questions start appearing in real renewals, and I revised that read. The skepticism was right about the hype and wrong about the direction.
There is one more angle I want to put on the bench, because it’s the one nobody is selling. If activity data becomes an underwriting input, then the people who benefit most are the ones who can control the quality of the data they present. That is a skill, and it is learnable the way any hands-on skill is learnable: you get the tools, you practice, you understand the material. The person who knows their own body’s numbers — the trends, the baselines, the red flags — is not at the mercy of a model built on everyone else’s averages. The tool does 80% of the work; knowing which one is the skill, and here the tool is your own record.
And the feel of it, when you get it right, is a good one. It is the same satisfaction as walking out of a workshop with a machine that runs better than when it came in. You are not outsmarting anyone. You are simply refusing to be the last person who understands your own data.
So here is my bottom line, offered in the spirit of a bench note rather than a sermon. The tracker on your wrist is now a pricing instrument with a step counter attached. You can’t un-invent that, and you probably shouldn’t want to — the same data that prices risk can also prove your health to a skeptic. But treat it the way you’d treat any new tool in the shop: read the manual, check the tolerances, and never let the measurement stand in for the thing itself. Your premium will follow your data, whether you read it or not. The only hands-on question is whether you’ll be the one doing the reading.
Hands-on beats theory every time, at least for the first hour — and the first hour of understanding your own data is the cheapest insurance you can buy.