Keith Camhi
Published in

Keith Camhi

photo: unsplash alex-knight-199368

Why Your Fitness-Tech Product Will Fail, Part 2: Automating the Wrong Things

Understanding the human-computer symbiosis for fitness-, wellness-, sports-, and consumer-health-tech

In Part 1 of this series, I described the background to these articles, and covered major categories of fitness-tech product concept flaws, including gimmicks, design by and for self-motivated super-fit people, and concepts that are easy to copy. Future articles will cover implementation flaws and go-to-market flaws.

In this article I cover a nuanced but critically important aspect of fitness-tech product strategy, namely what parts of the experience are best implemented directly to the user, and what parts are better handled indirectly through a coach.

Human-Computer Symbiosis

Human-computer symbiosis, the idea that technology can better augment or complement, rather than replace, certain types of human capabilities, has proven valuable in many fields. Erik Brynjolfsson and Andrew McAfee, in their outstanding book The Second Machine Age, discuss this at some length and cite freestyle chess tournaments as an example: presently the most effective chess “teams” combine a human (for strategy) and a computer (for crunching data on the implications of individual moves). A good computer plus a good human chess player can consistently beat either a grandmaster or a supercomputer alone (at least for now).

Helping exercisers get results and stay motivated to stick with their programs is arguably an even more complex and nuanced challenge and is also well suited to this joint approach: combining data collection and analytics with communications, empathy, and psychology-based support. It also becomes relatively clear, when you break down all the components of support that are valuable in assisting an individual’s fitness journey, from both the physiology and behavioral literature, which components are best handled by human or machine.

Yet product after product has come out for decades that gets the mix wrong.

Human-Computer Symbiosis Flaw #1: It Automates the Wrong Things

For fitness, technology is really good at certain things that human coaches are bad at, or at least inefficient for, like sensing and monitoring; collecting, storing and reporting data; and sifting through large volumes of data for trends, milestones, and early indicators of challenges.

Analyzing workout data is a good fit for tech.

Picture a human coach trying to go through a file cabinet of hand-written workout cards at the end of the week to see which exercisers are on or off track. It’s incredibly time consuming to get even a high level read. Compare that to the fraction of a second it would take a computer, with a database of workout result data and the proper algorithms, to do a much more thorough job reviewing a range of key status indicators for thousands of club members, players on a team, or training clients.

By contrast, technology is really bad at other things that human coaches happen to be really good at, like talking to other humans about their goals and aspirations, providing encouragement, discussing challenges, and changing programs to redirect or re-motivate someone when progress slows. While a computer might be able to collect and sift through the data with ease, when it tells me, “great job today,” it just doesn’t seem like its heart is in it.

Nevertheless, many fitness-tech products try to automate the human parts instead of the parts that tech can do well. It never works.

Example: Program Design

Program design is a good example. The initial consultation that kicks off a program for a new member at a fitness center is well-suited for human to human interaction. There are both objective and subjective aspects to it with inputs ranging from goals and schedule, to motivation, past experience, and physical constraints. Plus, it’s just good to make a human social connection at that point.

It’s one thing to use tech for this at a self-serve club targeting self-motivated, experienced exercisers, but it just doesn’t get the mix right at a well-staffed, retention-minded facility (other than to assist the coach). Nevertheless, I saw the first computer kiosks in such gyms as far back as 1995, designed to replace that first visit experience with a computer-created program for new exercisers, using a small fraction of the inputs that the fitness instructor in the office next to the kiosk would have used to create one better tailored to their needs and motivation profile. The product didn’t last long, but it did get funding and other attempts have followed.

Other Examples

Using technology to provide robotic messages of affirmation and encouragement, or to make major program updates (instead of just incremental tweaks), tends to fall short as well.

On the other hand, having coaches roam the gym floor trying to identify who might need support (other than for obvious poor form) is virtually impossible as well. The people who are losing motivation and are on their last visit tend not to self-identify. But the clues are in their workout data. Once identified, the coach (rather than computer) has the better shot of turning things around.

Ultimately, if you are designing a product for an audience or setting in which human coaches are part of the equation, take the time to consider if any given feature is better implemented through a coach rather than direct to the user. Most are obvious. Others may take trial and error to optimize over time.

Counterexample: Heart Rate Monitors for Group Classes

Heart rate monitoring for group fitness classes is a good example of getting human-computer symbiosis right. Participants wear wrist or chest straps that send their heart rates in real time to large monitors. A color coded box on the screen for each exerciser shows where they are in their own age-adjusted heart rate zone. This provides motivating feedback to each individual, and it also allows a single coach to oversee a couple dozen exercisers simultaneously.

Polar’s group heart rate dashboard (

In this case, the technology is doing things that it is uniquely good at: measuring exertion continuously in a way that a coach, even one-on-one, couldn’t do, and then compiling and presenting the data in real time in an easy to digest format.

The coach can then do things that he or she is uniquely good at: providing the human to human support to coax some, cheer others on, and let a few know they might be overdoing it.

This creates real value. However, as mentioned in Part 1 of this series, given competition, it’s not clear whether the value is more likely to be captured by the tech provider or by the business customer (such as Orange Theory and the like).

Counterexample: Great Play Interactive Gyms for Kids

At Great Play, one of the companies that I co-founded, we leveraged the concept of human-computer symbiosis to help coaches make it fun for kids to learn motor skills and sport skills.

Great Play coaches teach children the next step in a complex sport skill, such as throwing.

Human coaches are great at providing individualized instruction tailored to each child’s current ability, and then providing encouragement and pointers during practice and play periods, which are designed to give children the repetitions needed to embed a newly learned or improved skill into so-called “muscle memory.” However, it would be very inefficient for coaches to individually oversee every practice rep (either you need private lesson pricing for 1–1 attention, or a lot of down time for children waiting for the coach in a group lesson). It is also difficult for coaches to be the source of all the energy and creativity needed to keep classes fun all day every day, so practice gets boring, leading to dropout.

Getting practice reps at Great Play by breaking virtual bottles on the wall.

Our technology is great at the latter two tasks — supporting practice and helping the coaches keep it fun. Each patented interactive gym is outfitted with eight projectors, a sound system, lighting, and sensors, all under computer control. The lead coach moves through each pre-programmed group class with a simple remote control, first bringing up images to support instruction and then a series of games that turn practicing the skill into play. For instance, after working on a throwing skill, the kids cycle through stations and get reps by breaking virtual bottles on the wall, throwing into an interactive strike zones, etc. A younger class working on locomotor skills might play a big game of hide and seek with Buddy the animated mascot, chasing him as he appears on different screens running, jumping, or skipping, and then doing the same. The coach is the star of the show and provides instruction, encouragement, and pointers, and the technology adds energy and makes it fun for the kids to get the reps.

See a video of this in action.

The combination lets us create an effective and fun learning experience for kids, do it cost effectively, and drive 2–3x the conversion and retention rates vs. traditional low-tech children’s gyms.

Human-Computer Symbiosis Flaw #2: It Automates the Right Things but Leaves Out the Coaches

Closely related to automating the wrong things is automating the right things, but then leaving it at that and, for instance, presenting performance data exclusively to the end user without adding the final key pieces of functionality that would allow human coaches to assist participants in a leveraged way.

Example: Wearables

Many wearables seem to make this mistake. A typical wearable is great at collecting exercise data, organizing, and presenting it, but it just isn’t that credible when trying to provide words of motivational support or helping users know when and how to change their programs.

For the self-motivated, do-it-yourself exerciser with good fitness knowledge, that might be fine, but for the much larger audience that doesn’t fit that mold, it’s a significant missed opportunity.

Incorporating a workflow, for instance, with a dashboard for a coach to quickly comprehend the situation for each of many clients and provide them with individualized feedback and help, can make all the difference for a large potential market base. There are a number of approaches like this in the behavioral science that collectively can dramatically improve program adherence. Neither technology nor coaches alone can deliver on them, but in combination, they can.

I’d argue that this missing link is, in fact, a primary reason that makers of even very good wearable fitness tracking products are running into difficulty, with rapid churn in their customer base combined with a not-endless supply of new prospects leading to declining stock prices. After a period of explosive growth driven primarily by rapid customer acquisition for very cool new technology, the products are following the same predictable drop out curve that plagues self-serve fitness centers, as potential regulars are lost due to lack of proper behavioral support that a coach could have readily provided with the right set of extra tools. The difference between many wearables and self-serve clubs is that it’s an out-of-center experience. Their similar B2C-only mindset is causing them to miss a phenomenal B2B2C (or B2C2C with the middle C being “coach”) opportunity.

Counterexample: FitLinxx

FitLinxx Training Partner: device attached to strength equipment to coach users through exercise based on settings input by their coach, and collect exercise results for both.

At FitLinxx, we started incorporating coaches into the offering early on, and did extensive research and testing on how technology could both support exercisers directly and support the coach’s’ efforts to support the exercisers — both physically and psychologically. That focus (combined with incorporating both physiology and psychology into the product) allowed FitLinxx to beat better funded competitors, including various other startups as well as giants like Life Fitness, and go on to become the most widely adopted commercial fitness tech product to date, with millions of users in 500+ locations.

A high-level breakdown of how we divided the computer and coach tasks is summarized in this vintage product brochure. Advancements in technology have provided more sources of data (especially with wearables), but they don’t fundamentally change the division of labor.

Learn More



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store