Building a working biometric wearable is challenging, particularly if you’ve never done it before. Adding biometrics to a wearable is not just another version in your product line. It’s a completely different animal. There is a long list of interconnecting factors you need to consider when building a biometric wearable that works. And without an experienced partner, even if you properly account for all those factors, you’re likely to run into unanticipated challenges.

Valencell has been involved in more biometric wearable product development cycles than any other company in the world. We’ve seen many companies with little to no expertise or experience in biometrics try to do it themselves. Those companies very often reach out to Valencell for help when they can’t get their product to work with the levels of accuracy and performance they expect (or were sold by their OHRM vendor).

We thought we’d share some lessons learned from more than 10 years of building biometric wearables. Below you will find a few indications that a biometric wearable project needs help, but this is by no means comprehensive list – only the most common signs we see that signal negative outcomes.

1. Reference designs don’t work

Key indicator:

Your vendor should be able to show you a wearable reference design that works at the level of accuracy and performance you expect.

Why it matters:

Even if your design is very different from the reference design, your vendor should at least be able to show you they can make it work in some design. And this can’t be done by putting your finger on a dev board. You should be able to wear the technology and it should deliver the user experience you want to bring to your customers. If your vendor can’t show you this right now, it’s an indication they are in uncharted territory for your project and you are their guinea pig. It’s not a good place to be.

2. Testing problems

Key indicator:

You or your vendor are only testing your prototypes on a few people around the office (or any other very small population of test subjects).

Everyone is different

Why it matters:

The old saying “everyone is different” holds true here because biometric wearables perform very differently on different people. Every new version of your prototype should be tested on at least 20 different people under varying workloads and these people need to be all shapes, sizes, skin tones, genders, and fitness levels. You also need to test for your use case – intense running, lifestyle, or whatever you expect your customers to be doing with the wearable. We’ve had companies come to us just to test their production-ready devices because they know we have the experience, expertise and facilities to test correctly. It’s a part of biometric wearable product development that goes woefully underserved in too many cases. Never underestimate testing.

3. The Whack-a-Mole problemwhack-a-mole

Key indicator:

As the product development cycle continues on, you find that as one problem gets solved then one or more other problems will pop up, like the arcade game Whack-a-Mole.

Why it matters:

Biometric wearables work best when the biometric sensor system is designed as a holistic solution – hardware, software, and testing together. If they are not designed to work together, the system that has been cobbled together does not respond well to changes in other parts of the design. This is like trying to build a car with a Ford engine, Chevy transmission, and Toyota chassis. It might look like a car, but it won’t run well (or at all). The whack-a-mole problem is a sign your vendor doesn’t have a good handle on the system-level solution required for a successful product and you will spend precious time and resources whacking moles instead of getting your product in the market.

4. The “just another couple weeks” trap

Key indicator:

You are experiencing project time-delays and your team and/or vendor is telling you “just give us another 2 weeks and we’ll have it solved”.

Why it matters:

It’s never that simple. Multiple time delays without very clear indication that the team understands the problem means that it’s not likely to get fixed without help from an experienced expert. Doing the same thing over and over, expecting a different result is the definition of insanity.

5. Manufacturing a biometric wearable for the first time

Key indicator:

The contract manufacturer chosen to build the product has no experience building biometric wearables.

Why it matters:

Building biometric wearables at scale is very different than building your standard consumer electronics product, or even a simply activity tracker. Your vendor should have a network of experienced contract manufacturers to choose from who have successfully manufactured biometric wearables at scale. It should be noted that experience building activity tracking devices alone (devices that use accelerometers to estimate steps and calories) is not sufficient – they need to have proven experience with validated wearable biometric optical sensing technology.

6. “We’ll just fix the algorithm”

algorithm

Key indicator:

You find your vendor pushing multiple algorithm updates with no meaningful improvement in performance of your prototype.

Why it matters:

If all you have is a hammer, everything looks like a nail. We see many companies working with silicon vendors that think they can overcome inferior optomechanics and hardware problems with algorithm updates. It’s never that simple. A sure sign that you’re caught up in an algorithm trap is if you find that each new algorithm update exposes new problems (the Wack-a-mole described above). No matter how good an algorithm is, it can’t fix bad sensor data.  And no matter how good your sensor is – bad algorithms, poorly designed optics, poorly designed mechanicals, and inconsistent manufacturing will result in a failed user experience for your customers.  Your vendor should have broad experience in all of these fields, including how they interact with one another.

7. The 5-month threshold

Key indicator:

You are five months from go-live with your product and you still don’t have a prototype with target accuracy and performance.

Why it matters:

We’ve seen this time and time again when people think they will be able to get the product working in a short timeframe. It doesn’t happen and the project gets delayed or cancelled. Don’t let yourself be deluded. Get help now.

8. The disappearing features phenomenon

disappearing
Key indicator:

You find that the features outlined in the original vendor’s spec sheet are getting diminished over the product development cycle, because the device is not performing as expected.

Why it matters:

These feature-sacrifices will limit your product’s differentiation in the market and significantly decrease your chances of success with the product. Some examples we’ve seen are companies that originally wanted continuous HR monitoring, but found their prototype would only support resting HR when the user was not moving. Or companies that wanted to do heart rate variability or stress analysis, but couldn’t get to the level of precision required to support that use case, so it got dropped from the final product. Before getting locked into a vendor’s solution, be sure to check out products using their technology existing in the marketplace (if any). See how those product’s features are being marketed to consumers.  Is your vendor promising continuous heart rate while running, but all of their customers’ product market “one touch” style of heart rate at rest?  How are those products ratings for accuracy?

9. The disappearing roadmap phenomenon

Key indicator:

Similar to #8 above, the feature roadmap you were sold doesn’t materialize once you get into the product development cycle.

Why it matters:

Biometrics are becoming standard features for wearables today and basic heart rate features alone will not win new customers in the future. If you can’t provide a roadmap for biometrics that deliver continually interesting insights to your customers, it severely limits your ability to add new features and upgrades in future products.

10. Your designs are crossing other company’s intellectual property

Key indicator:

In the early stages of the process, every design your team comes up with clearly crosses well-established patents in this space.

Why it matters:

Despite being a relatively new space, there is an extensive patent landscape in biometric wearables and high-profile companies are litigating in this space. This includes areas such as optomechanical designs, data processing, data assessments and visualization of wearable data. Make sure your team is well-versed in the IP landscape and your designs are free from any IP complications that could cause problems after the product is in the market.

Many biometric sensor vendors claim they have solutions and expertise in this market, but it’s usually just their ambient light sensor and with “heart rate monitor” on the datasheet.  It takes more than that to get a successful wearable user experience into the marketplace.  Only one company in the world has a proven track record of making it work — Valencell.