Personal Activity Monitors

After my Fitbit review was published I was sent some varied and interesting feedback. I can split most of it into two camps of people saying:

  1. You are wrong.
  2. That’s what I thought all along / You are right.

That’s pretty typical of anything that I write. What made this go around a bit different is why people felt that I was wrong. Specifically why people think I am wrong compared to why I think I am spot on.

A significant amount of the “you are wrong” emails I received had anecdotes about how devices like the Fitbit have helped them. “I lost X pounds in X weeks.” “I am way more active.” “I actively make different decisions to my benefit because of this.”

I don’t want to discredit these people as there certainly is a psychological effect to wearing such a device. That effect though isn’t enough to convince me that this is a good device. I am sure it can be a good device, but there are to many conditions to make that true.

Given these emails I can see now that I wasn’t clear enough about why I discount the legitimacy of the Fitbit. Instead of appending an update to the post that a great many people won’t see I wanted to expand on my thoughts about these types of “personal activity monitors”.

More specifically I want to elaborate on two things:

  1. What these devices are currently good for.
  2. How these devices would be useful to everyone that strapped one on.

These two things are actually very different. These devices are like accounting statements. Where a general ledger, income statement, and balance sheet are useful tools — but only useful if you fully understand all of the data, where it comes from, and how to use it. As a pure source of data accumulation and reporting I find that these tools excel.

Where they fall flat though is in their ability to help a user come to a decision about what should be done next when it comes to their health.

Current Use

I think the best case usage scenario for these types of devices (as they currently exist) is for people who have a clear goal in mind and know how to reach it (and what their goals should be). That could be a few different types of people.

  • It could be someone who simply knows that they want to walk/run X miles per day. Thus they want a way to track such data. For that the Fitbit would be an excellent choice.
  • It could be someone who just wants to increase their activity by a factor of X over the next few months. Again, a great option. That’s a goal that is easily quantifiable.

Really these devices seem to be more tailored to people that are currently “in the know”. These are the knowledgeable set that want to learn, analyze, and conclude on their own. For this subset I have no doubt that the Fitbit will work astoundingly well.

(Don’t forget about the psychological effect that I mentioned above.)

What happens though when you are someone that just says: “I want to be more healthy.” That’s not something that can be quantified in steps taken per day, miles walked, stairs climbed, hours slept — yet this is exactly the type of goal the target market for these devices are seeking help with.

Moving Forward / What I Want

What people really need (and what I think Fitbit and UP fail to deliver) is not personal statistics about activity levels, but a personal assistant that can translate those statistics into digestible and understandable information.

It’s the difference between data, useable data, and actionable data. Right now these solutions only succeed in the first two and fail in the latter. Yet the actionable data is the most important data — without action we just have a plastic piece of junk weighing us down for nothing (both physically weighing is down and psychologically).

We need an intelligent means of translating this data into actionable information. A way where we as the lazy beings that we are don’t have to lift a finger to learn the what we should be doing.

What This Looks Like

I completely understand that most won’t “see” what I mean in the above. Let me try to better explain my vision by providing an example of what I think would be neat — regardless if things exist to be able to achieve such results.

First and foremost the device would tell me in a matter of fact way whether I gained, lost, or maintained my weight that day. The value in that alone is helpful — don’t skirt this because you want to spare feelings. Just state it. Time to get real.

Certainly the Fitbit can do this with the calorie burn tracker and the food consumption deal, but it fails to do this because the food consumption piece is too cumbersome for an average user to ever use.

If I drink a Pepsi and want to enter that, here is what the current screen looks like:

That is a UX nightmare. That makes me want to avoid this entry screen at all costs.

I entered what I ate everyday for a week and I regret wasting that time. It was an effort in futility and frustration.

There are three current shortcomings of the system:

  1. Too many duplicates.
  2. Too much user knowledge required about portions and calories.
  3. Too few assumptions.

Why not just simplify the screen and instead respond to me wanting to enter “Pepsi” as an item by saying the following:

How big?

  • Small Cup
  • Medium Cup
  • Large Cup
  • XL Cup
  • Can
  • Standard Bottle

Pepsi’s calories are a knowable quantity. You can estimate the rest of the sizes and be close enough — we don’t need everything broken down by: Taco Time Medium Cup, 8fl oz, 2 cups, Large KFC Cup, Mexican sub-full-sized-Cancun-special bottle. That’s just confusing for the user and will encourage them to not use this functionality.

Adding what you ate should be fun in the best case scenario and fluid in the worst case.

Strip out the shit and take some guesses. A bad guess is infinitely more useable than no input. We must embrace the inaccuracy of such a system.

Further, what if I eat a home cooked bowl of chili?

How do I input that? I need to guess the serving size and calories? I don’t even know where to start with such a guess. The Fitbit app lists a slew of “chili” options, which only confuses me more. None of these were my wife’s chili recipe — nor would I expect that they would be. 1

Just let me enter “chili w/ cheese, 1 bowl”. From there you guess at the rest — I can assure you that your guess will be far more accurate than mine.

That’s step one and it is a necessary one.

Step two is eliminating a stupid little sensor that I must carry with me while adding in more useful sensors.

Imagine this: you place a small film sensor under the lining of your shoes (these sensors would ideally be cheap enough that you get a few so as to keep them in more than one pair of shoes).

More sensing activity is gained from your phone/ watch. (In other words: things you already have on your person.)

There is an additional sleep sensor and it’s all built in to your pillow or a pad underneath your sheet.

All of this syncs with your phone constantly — with no user input and no need to every worry about carrying additional items.

That’s how you make a useful system.

The pad in your shoes detects two things:

  1. Weight
  2. Weight distribution

Using the weight sensor everything knows how much you actually weigh and if you are carrying additional weight with you (in other words is has a baseline for your weight). This will give a more accurate sense of calorie burn when you are carrying a computer bag and other items and a recommended activity level.

These weight sensors could also be a better way of counting a step, as a step would theoretically reduce the weight on that foot. Coupled with the GPS sensor talked about later this could give very precise measures of distance walked with a low false-positive rate. Likewise it would stand to reason that when running the pressure hitting each foot would be different in such a way that he software could determine you are running and adjust calorie burn accordingly.

The second shoe sensor would be able to help with posture problems that likely plague many people. This is done much in the same way that the Wii Fit module detects these things. For instance I stand all day and it would be really nice to know if I have good posture, or if I should work on something. Perhaps I stand on the balls of my feet too much — a weight sensor would know this.

The phone/watch sensors would track:

  1. Pedometer
  2. Altitude
  3. Distance
  4. GPS 2

What we want to know is just how active we are and what we should change. I want to add in the GPS portion so that we can disregard data gained while traveling on major highways (it is not likely we are walking down them). It would also be coupled with the movement tracking to see if a person is actually moving about, as opposed to shaking a leg. This would give a better idea of actual calories burned.

Now that we have most of what we need to track down, I want to look at how we use that data. As I see it a good device needs to do a few things:

  1. Encourage me to be more healthy.
  2. Show me how and where to be more healthy.
  3. Explain what I am doing right.
  4. Explain what I am doing wrong.

This is where the Fitbit falls on its face.

Please do not confuse my saying the data needs to be useable as my saying that the data should be infographic in nature — that’s not what I want. Instead the data must be understandable without prerequisite knowledge and instantly understandable to every user. Such a device shouldn’t be scaring off users because they can’t understand what the device is telling them.

Telling me that I burned 2400, of 2560 calories consumed is nice — but what should I do with that? Is that good? Is that bad?

What would be better is saying something more like: You maintained your weight today, loosing weight at today’s activity level would have been as simple as not drinking that Coke at lunch.

This may sound offensive to some, over stepping to others, but it is hard to deny the helpfulness of such a response. More importantly that information is incredibly actionable. It encourages the user with the knowledge that they are on the right track and gives them something tangible that they can act on: drinking one less Coke.

That seems pretty simple to implement. Further such a device should be willing to recommend things when a users asks. If I get home from work and want to know if I was active enough during the day and the software says I wasn’t — why not then say:

  • “You still have some energy to burn, may I suggest a 20 minute walk around your block?”
  • “Maybe walk the dog to walk off today’s stresses.”

That is helpful.

Perhaps the user can even ask what they should eat:

  • “Given your consumption today snacking on a Clif Bar would be great about now.”
  • “You had a light day today, perhaps just a salad for dinner?”

Those in the proper mindset would thrive with such an intimate and personal tool — likely you are in this mindset if you are willing to buy such a system. This is something that is locked away in a computer that you control instead of in the, perhaps, judgmental eyes of a trainer or nutritionist.

Most importantly the information is clear and actionable. A chart that says you have walked 2,467 steps out of your 5,000 step goal doesn’t tell you much. But telling a user that they should go walk around for 30 minutes is actionable — and something that is far less daunting because we know what 30 minutes of walking is like.

Is a user more likely to try and hit a goal of 5,000 steps if they see that are only halfway there, or are they more likely to accept that they need to walk for 30 minutes? How long does it take to walk 2,500 steps? I don’t know and I doubt most people know.

The Goal

As you can see I think that a device that is truly useable is one that helps the user make decisions driven from data by recommending actions. Not one that helps provide the user with data.

Data is nice and fun to look at, but many users can’t quantify and translate that data into actionable information. A consumer device needs to be designed to be used and understood by any user.

  1. I am pretty sure she makes it up every time. Still delicious.
  2. Yeah, battery life. This is make believesy.
Originally posted for members on: December 7, 2011
Follow along on RSS, App.net, or Twitter.
~I would appreciate it if you considered becoming a member.~