What it actually takes to launch your own health band
By Shailendra Kumar Ram · · 12 min read
Somebody at a factory in Shenzhen will sell you a finished fitness band tomorrow for about ten dollars, with your logo on it and a free app. That is why so many people think launching a wearable brand is a sourcing problem. It is not. The hardware is the cheapest and most solved part of the whole exercise, and the three things that will actually decide whether you have a business are certification, the claims you make, and the app.
This is written for the person financing that decision rather than the person soldering it. The engineering is covered elsewhere on this site; what follows is the order of operations, where the money goes, and the two or three places where people who skipped a step lose a year.
The market you are walking into
Two numbers set the shape of it. The recovery and readiness category is genuinely large and growing: Whoop reported a $1.1 billion revenue run rate and 2.5 million members before raising at a $10.1 billion valuation, and Oura passed $1 billion in revenue in 2025 having doubled year on year.
The second number is the one nobody quotes at you. IDC forecasts the wristband form factor to shrink 6.8% in 2026, to about 40.2 million units, squeezed by smartwatches from below and smart rings from above, after growing 14.7% the year before. So the appetite for health tracking is rising while the appetite for a band specifically is falling. That does not make a band a bad choice, but it does mean you are not going to win on the form factor. You will win, if you win, on what the software does with the data.
The live example is the Luna Band, which went on sale worldwide this week at $149 with no subscription, from an India-based team, positioned squarely against Whoop's monthly fee. Google's Fitbit Air arrived in May at $99 with an AI coach. That is your competitive set: one competitor removed the subscription, another removed most of the price, and both of them ship a screenless device where the phone is the entire interface.
Where the band itself comes from
There are three routes, and the difference between them is how much of the product is yours rather than how good it is.
| Route | What you get | What it costs you |
|---|---|---|
| White label | An existing band, your logo, the factory's app with your name on it | Cheapest and fastest. You are selling the same device as everyone else who called the same factory |
| ODM with customisation | Their reference design, your industrial design, sensor choices and firmware changes | The usual middle ground. Real differentiation is possible, mostly in software |
| Own design | Your schematic, your enclosure, your firmware, a contract manufacturer building it | Slowest and most expensive, and the only route where the hardware is defensible |
Shenzhen dominates this: industry directories list dozens of established wearable ODMs, and minimum order quantities for a white-label band commonly start around 500 units, at roughly $8 to $14 for a basic band depending on display and sensor set. Those are indicative figures rather than a quote, and they move with your specification, but they are the right order of magnitude for planning.
The important thing about that price is what it tells you. At ten dollars a unit, the hardware cannot be your moat. Your competitor can buy the identical device. The only durable differences are the sensor quality you pay up for, the firmware you control, and the software experience on the phone.
- Ask which chipset and which optical front end, not just "does it have heart rate". Sensor quality is where the price differences hide, and it is the one hardware decision you cannot fix later in software.
- Ask whether you get the firmware source or only binaries. If you cannot change firmware, you cannot fix a protocol bug, and every future feature is a purchase order.
- Ask for the GATT table in writing before you order. If the factory cannot produce one, your app team is going to reverse-engineer the protocol, and that is a real line item.
- Ask who owns the tooling if you paid for a custom enclosure. This is a normal question and a bad answer is a warning.
What a band like this can actually measure
Almost every consumer band is built on the same two sensors, and understanding what they physically do is what stops you writing marketing copy you cannot support. An optical sensor shines light into the skin and measures what comes back, which is photoplethysmography or PPG. An accelerometer measures movement. Everything on the box is derived from those two, plus a thermistor if you paid for one.
| What you can offer | Where it comes from | How solid it is |
|---|---|---|
| Heart rate | PPG | Strong. Agrees closely with ECG at rest |
| Heart rate variability | PPG, beat-to-beat intervals | Good at rest, noisy during movement |
| Blood oxygen, SpO2 | PPG at two wavelengths | Workable on a wrist, much better on a finger or ear |
| Respiration rate | PPG, derived | Reasonable at rest |
| Steps, activity, calories | Accelerometer | Steps are solid. Calories are an estimate wearing a confident font |
| Sleep stages | PPG plus accelerometer | Inferred, not measured. Useful for trends, not for a diagnosis |
| Skin temperature | Thermistor, if fitted | Relative trend rather than core body temperature |
| Stress or readiness score | Your own algorithm over HRV | Entirely your invention, and therefore entirely your responsibility |
Blood pressure and blood glucose are not on that list, and no amount of supplier enthusiasm changes that. Both can be estimated from PPG in a research setting, but the practical error is large: an error of a single millisecond in pulse transit time translates to roughly 1 mmHg of blood pressure error. A wrist band with an optical sensor is not a cuff and it is not a glucose sensor. If a factory datasheet claims otherwise, that claim is now your legal problem rather than theirs.
Your marketing copy is a regulatory document
This is the section people skip and then regret. In the United States, what determines whether your band is a regulated medical device is not the sensor inside it. It is what you say it does.
The FDA published updated guidance on general wellness devices in January 2026, and the shape of it is helpful for anyone launching a consumer band. Products intended to maintain or encourage a general state of health, without claims about diagnosis, cure, mitigation, treatment or prevention of disease, sit in general wellness and outside active oversight. You are allowed to prompt someone to see a doctor when a reading looks unusual, provided you avoid disease-specific language, do not label a result as pathological, and do not set clinical thresholds.
| Safe territory | Regulated territory |
|---|---|
| "Track your resting heart rate over time" | "Detects atrial fibrillation" |
| "Understand how your sleep varies" | "Screens for sleep apnoea" |
| "See how your body responds to training" | "Diagnoses overtraining syndrome" |
| "Some users find this helps them manage stress" | "Treats anxiety" |
| "Consider speaking to a doctor if this looks unusual" | "Your reading indicates hypertension" |
Three practical consequences follow, and each one has cost a real company real money:
- Your landing page, your app copy, your ads and your founder's conference talk are all evidence of intended use. A careful product with an enthusiastic marketing page is a regulated device with bad paperwork.
- Blood pressure is the tripwire. The FDA has previously warned a manufacturer over blood pressure insight functionality on the basis that such measurements are inherently tied to diagnosing hypertension. Treat any BP feature as a decision to become a medical device company.
- Deciding this late is what costs money. Moving from wellness to regulated changes your documentation, testing evidence, release process and timeline, and it is not a label you bolt on in month six.
None of this is legal advice, and the equivalent questions in the EU under the MDR, and in India under CDSCO, have their own answers. The point is only that this decision belongs in week one, alongside the sourcing conversation, rather than after the packaging is printed.
The certifications nobody puts in the first budget
You cannot legally sell a Bluetooth device without qualifying it, and this surprises almost every first-time hardware founder. Bluetooth is a trademark licensed by the Bluetooth SIG, not a free-to-use radio standard. Adopter membership itself is free, which is where the confusion starts, but qualifying the product is not.
There is a legitimate way to make most of that smaller. Build on a pre-certified Bluetooth module and you inherit its FCC ID and much of its test evidence, which removes the expensive radio testing and reduces the SIG work to a simpler declaration against an already-qualified design. Almost every ODM band is already built this way. Ask your supplier for the module's certification identifiers before you order, and confirm in writing which certifications transfer to you and which you must hold in your own name as the brand placing the product on the market.
Then there is the market-by-market layer, which is separate from all of the above:
- United States: FCC certification, because a Bluetooth device is an intentional radiator under Part 15.
- European Union: CE marking under the Radio Equipment Directive, plus GDPR obligations the moment you process health data, which is a special category under the regulation.
- India: WPC ETA approval from the Department of Telecommunications for any device using unlicensed bands. Without it the device is not legally importable or sellable, and the process typically runs four to six weeks with correct documentation.
The app the factory gives you is not the app you need
Every ODM offers a white-label app. It has your logo, your colours and your name in the store listing, and it costs nothing or nearly nothing. For a first hundred units it is genuinely a reasonable way to prove there is demand. As a business it is a rented storefront on someone else's land, and the rent is your product.
- The differentiation is skin deep. Your competitor who called the same factory has the same app with different colours. Everything you told investors about your experience layer lives in a codebase you do not control.
- You cannot ship a feature. Every change is a request to a factory whose priorities are not yours, on a timeline that is not yours.
- Your customers' health data is on infrastructure you did not choose. You are the data controller in the eyes of GDPR whatever the factory's contract says, and you have to be able to answer where the data lives, who can reach it, and how a user deletes it.
- Your store listing is your legal exposure. Apple requires a privacy policy for apps using health data and restricts sharing it with third parties, and none of that obligation transfers to the factory.
- You cannot leave. When you outgrow the app, your users' history is in a database you have no export from, and migrating a customer base off it is the most expensive project in this whole article.
There is also a hard deadline sitting in the middle of this. Google's Fit APIs shut down at the end of 2026, and Health Connect is the replacement on Android. Any white-label app still built on the old APIs is on a countdown, and whether it gets migrated is the factory's decision rather than yours.
What your own app actually has to do
This is where a wearable stops being a sourcing exercise. A screenless band is the extreme case: with no display, the phone is not a companion to the product, it is the product. Everything the customer experiences, they experience through your app.
- Collect the readings taken while the phone was away. A band records continuously and stores what it could not send. Fetching that history is a separate mechanism from live notifications, and getting it wrong loses data silently. This is the single most common defect in wearable apps, and it is covered in detail in the readings your health device app never collected.
- Stay connected in a pocket. Background operation on iOS and Android is the largest single multiplier on the cost of a Bluetooth app, and an app that only syncs while open is a different product from one that just works.
- Update firmware over the air. You will ship a firmware bug. Without OTA, your fix requires a recall. See BLE OTA updates that survive a dropped connection.
- Write to HealthKit and Health Connect. Neither platform ingests your device automatically. Your app writes every sample, and it has to avoid duplicating history on a resync.
- Own the account and the data. Export, deletion and portability are legal obligations in most of your markets and a competitive asset in all of them.
- Turn readings into the score you are actually selling. The algorithm is your product. It is also the only part of this list a factory will never build for you.
How that translates into hours and money is worked through in what a Bluetooth build actually costs, and the questions worth settling with your supplier before development starts are in the iOS BLE hardware integration checklist.
The order to do it in
Most of the expensive mistakes in this article are sequencing mistakes rather than spending mistakes. This is the order that avoids them:
- Decide your claims first. Wellness or medical. It sets your budget, your timeline and your marketing before you have spent anything.
- Then choose the sensor set the claims require, rather than the sensor set the factory is promoting this quarter.
- Then pick the supplier, and get the GATT table, the module certification identifiers and the firmware terms in writing as conditions of the order.
- Then start the app, in parallel with sampling rather than after it. The app is the long pole and it is the thing you are actually selling.
- Then certify, using the module's existing evidence wherever it legitimately transfers.
- Then order volume. Not before. Five hundred units of a device your app cannot properly sync is the most common way this ends.
At ten dollars a unit, nobody is buying your hardware. They are buying what your software says about their body, and how much they believe it.
Figures here are drawn from the Bluetooth SIG's published fee schedule, the FDA's general wellness and device software guidance, IDC's wearable shipment tracking and published test-house pricing, and they are planning numbers rather than quotes. Regulatory positioning is a question for counsel in each market you sell into. If you are working through a wearable launch and want an honest read on the app scope before you commit to an order, send over the GATT table and the claims you intend to make and we can work through what the software actually has to do.