SHFT.run - Initial Thoughts

My photoshop skillz are second only to my running style

As a rule, triathletes are data and gadget freaks - and I am no exception. In the pool, you can usually spot a triathlete by the selection of "toys" they bring down; on the bike cyclists despair at our covering our beautiful machines in power meters and storage boxes; and I feel the run revolution is just getting going. Garmin's Run Dynamics (now a proper ANT+ profile) and Stryd's power measurement are well known - I had the opportunity to try out SHFT.run's two-pod system and report my experiences from my first few runs here.

Background

I have had a Garmin HRM-Run heart rate belt for a few years, but find that whilst it's interesting to know my Ground Contact Time and Vertical Oscillation etc, it's mostly a load of extra data to ignore as I don't know how to interpret it and work on it. I've not used Stryd - but I get the impression they're mostly focussed on pushing the physiology, using power numbers in the same way we use them on the bike - for pacing during races and intervals.

In common with most former rowers, I had (have?) a terrible running style. People who've seen rowers run will confirm the "slap slap" noise of feet coming down hard - and I was bad even for a rower. When I took up running, I couldn't run for more than a few minutes without terrible knee pain, shin splints, you name it - and it was only by treating running as a technical exercise that I graduated to enjoying running as I do today. So when I saw the SHFT IQ Kickstarter campaign, promising technical coaching for those of us who don't have a proper coach to watch us, I signed up. SHFT's idea is to turn those esoteric running metrics into actionable intelligence - moving the athlete towards a more efficient, and hopefully injury-free, running style.

After I joined the Kickstarter, the guys at SHFT.run kindly offered to let me have a look at their earlier product - a two-pod "Pro" solution. The technology may be different to the upcoming "IQ", but I looked forward to an opportunity to explore their philosophy - and get myself some run coaching!

 

The Hardware

The SHFT.run devices arrive well-packed in a cylinder. There are two pods (they're interchangeable), a USB charger, and a chest strap. The pods are triangular, with a sturdy clip and a single notification LED, which flashes blue when waiting for a connection and pulses green when all is well. Indoors this works well, but outside it can be difficult to work out what's going on, especially when it's on your foot.

Connection is via Bluetooth - unfortunately there's no ANT+ here - which means that you have to start taking your phone along on runs (Ray Maker has made a lovely chart of the crossover between serious runners and those who run with their phones, opining that they're fairly different groups). I tried with my Android watch-phone (Zeblaze Blitz, a full Android phone in a watch) and the app appeared to work but connection with the pods was almost impossible to sustain. My proper phone (Android Elephone P9000) found the pods rapidly for pairing, and once connected (the big "Start Running" button actually means "Connect Pods", you get another button and a countdown when you actually want to start running) I was off.

 

The Software

Obviously I can only talk about the Android version, though I believe the iOS version is almost identical. In setup, the app asks you for your details - though frustratingly while it offers an option to enter imperial units, it seems not to convert them - so initially I was 6cm tall and weighed 190kg. After fixing this, I was presented with three simple options - to Run, to view my History, or to watch Training videos. Obviously I went for the Run. With my bluetooth headphones on, I pressed 'Start Running', waited for it to connect to the pods, pressed the actual start button, and was counted down to the arrival of my new running coach.

The coach is the real jewel of SHFT.run. A slightly robotic, but very clear, voice guides you through a 5-minute warmup followed by a 2-minute "screening run", and then three 6-minute technical sessions working on the metric that the software has decided will bring the most improvement to your running. This is the piece that the other manufacturers are missing - I can get great data from my Garmin HRM-Run, but frankly I'm bewildered as to how best to work with it. SHFT.run's virtual coach told me that my Ground Contact Time needed working on, and despite the fact that I really had no idea what to do about it, it guided me clearly into the zone it had chosen and kept me there with repeated live feedback. One of the best rowing coaches I ever had the pleasure of learning from (Harry Mahon) was famous for his style - he'd come along side you and each stroke give you a simple "Yip" if you'd done it right that stroke, or a "Nope" if you'd done it wrong. No amount of careful explanation can make up for that simple, instantaneous feedback - and SHFT.run can definitely deliver.

However, despite the promise - and delivery, there has been an element of frustration too. When there's an issue with the Bluetooth connection the coach simply announces that connection to a pod has been lost, and falls silent. No more actionable intelligence, and worse - the app doesn't seem to make any effort to reconnect (sometimes it announces that it's trying, but after 40 minutes of silence I have to assume it's not going to). If one pod is lost, half the metrics go. If the second goes, the app seems to give up entirely - it abandons the GPS, all the metrics sit on zero, and when you press 'stop' your entire run so far is lost into the ether. The pods unfortunately seem to have quite a limited Bluetooth range - the website specifically warns against having your phone behind you; I found that even the slightly-behind-the-midline pocket on my running shorts was too much and I would get the "pod disconnected" error within a minute or so. If it could automatically reconnect it would be less frustrating, but each time it meant wasting a couple of minutes with my phone out, turning Bluetooth off and back on and restarting the run on the app to get the pods reconnected before I could get back on my way. I also quickly decided that my Bluetooth headphones would have to go, to minimise interference given the penalties for a dropped connection. The best connection appeared to be with no headphones, Bluetooth off on my Garmin, and the phone on my waist at the front (I also tried it slung across my chest, but even that seemed to cause issues after a few minutes).

First Run - Coaching

Please note incorrect foot pod position!

 I have to confess to being a bit of an idiot here. On my first run I had the foot pod in the wrong place (I put it central like a traditional footpod, rather than on the right side of the shoe), and I think I may have had the chest pod upside-down (at least, that's the only way I can explain the data being as strange as they are). However, I can comment that the system seems to cope quite well with the foot pod being wrongly positioned - at least, my subsequent data look quite similar. The chest pod, though, really needs to be the right way up.

For the start of my very first run with SHFT.run, I had Bluetooth headphones, my phone was in front of me in a small waist-belt, and I went for the two-strap option with the SHFT.run chest strap positioned slightly above my normal heart-rate belt. I had scheduled a 45-minute recovery run, so some technique work seemed to fit in perfectly. My new coach asked me to warm up for 5 minutes, so I set off at an easy jog. Words from the coach seem to come at a nice pace - not so often you want her to be quiet, but often enough that you feel you're being coached properly. After 5 minutes she instructed me to pick up the pace slightly for a 2-minute screening run. I wasn't sure quite how fast to go, so went for an easy Zone 2 jog. She identified that my stride rate should increase, and that became the focus for my technical runs. During each of these, I was nicely guided into a 'zone' - when I experimented by going too high I was told to back off slightly, and overall I had that really nice feeling I remember from being coached - that someone has a definite idea of what they want me to do and is giving me the feedback I need to get there.

Unfortunately, Bluetooth issues struck during the second technical run and I had to start the process again. This time during screening I had a slightly higher stride rate, and was pleased that it didn't blindly decide that it was still below 180 (I have long legs!) and try to coach me higher. This time it was my Ground Contact Time I needed to work on. I was quite pleased as this seems to be an important metric that I get from my Garmin HRM-Run, but one I have no idea how to fix. And this is why I can say the coach (she needs a name!) definitely works - I was free to experiment, all the time getting feedback that directed me into the right zone. And I got there - and the graphs afterward clearly show that during the technical segments she was guiding me right into her target band. It became clear to me that SHFT.run isn't another generator of inscrutable numbers, but is a genuine running coach capable of moulding the way you run.

Second Run - Intervals

I decided to try SHFT.run again, except during an interval session - a classic 12x 1-minute on, 1-minute off that I do on a nearby cricket pitch with a circumferential path that's almost exactly 400m. With the coach switched off, I hoped that the changes in pace might show something interesting in the variety of metrics that SHFT.run offers. For some reason, the app gave up its GPS connection part-way through the run, but it managed over 50 minutes without Bluetooth issues. I'll go through each metric in the order displayed in the SHFT.run portal:

• Watt - I'm not sure why this isn't called "Power". I'm not convinced by the values here. 180W when jogging along seems plausible in comparison with what I get cycling. But 240-270W when going very hard (and nearly twice as fast) during the intervals doesn't make sense - from cycling I'd expect something nearer 450-500W, and even from simple mechanics I should have to use at least twice the power to go twice as fast.
• Running Efficiency - This looks like a really interesting metric. The fact that mine is near 20% is a worry, as they state it's the proportion of energy going to moving me forward. Is the other 80% something I'm doing wrong? Or are they accounting for energy loss as heat? I'm sure it's measuring something important as I could see it dropping on each interval, presumably as I became tired and a less efficient runner. They're definitely on to something useful here!
• Pace - Unfortunately this wasn't much use due to the GPS dropouts. I do wonder why the SHFT.run pods don't calculate pace themselves, given that a standard footpod can usually do the job more accurately than a GPS (see fellrnr for more details!)
• Brake Effect - This is again potentially very interesting, but it seems that the position of the chest pod can affect this far more than running style. Initially I had the pod outside my clothes, and was getting Brake Effect readings in the 30W range. When I put it next to my skin the values dropped to 5-15W. Presumably slippery triathlon clothing allows the sensor to move a little more and messing up the readings?
• Body Bounce - I feel like I have a bit of a handle on this from the HRM-Run. I know that I bounce a bit when I'm jogging, but it drops a lot when I run hard. SHFT.run doesn't agree and says that my bounce is about 6cm regardless of what I'm doing (the HRM-Run figures are about 11cm and 8cm). Even when I tried my bounciest gait on the way home, SHFT.run insisted I was still about 6cm - I think it's wrong here.
• Ground Contact Time - The one my lovely coach was working me on. The absolute values are about 20% higher than the Garmin ones, but they definitely vary in the same way and who's to say which figures are correct? I can imagine it depends a lot on whether we're talking complete loss of contact, or just that the force drops below a certain value etc. I'm happy to trust SHFT.run here, and just be aware that I can't compare with old Garmin values.
• G-landing - I have no idea how to interpret these. About 10G during intervals (though one spiked up to 18G which has me worried!). I wonder if she'll have me working on this at some point, having it too high sounds like an injury risk.
• Landing angle - I was surprised to find I'm considered a "midfoot runner". I've been working on it, but I keep seeing photos of me at the end of triathlons with what looks like an imminent heel-strike. As predicted, during the intervals I move slightly closer to forefoot running. I really liked this (though the 3D visualisation of foot movement is a bit gimmicky!) - looks like a good way to watch yourself for overstriding.
• Step length - Unfortunately this just seems to be calculated from speed and steps per minute so with the GPS misbehaving it was useless. Again, I'd hope this could come direct from the foot pod.
• Steps per minute - Cadence is pretty basic; I'd be surprised if this is inaccurate - and it certainly seems to match my HRM-run at a variety of cadences.
• Time in Air - I'd have assumed that this would be calculated from Ground Contact Time and cadence, so that the total time would represent one step period - but it seems to be something else. I certainly have segments where both GCT and TiA go up without a change in cadence. It appears that there's another time period not being displayed - presumably where the foot is just leaving the ground.
• Toe Off Angle - Not something I'd encountered before; apparently it's the angle of your foot to the ground just before leaving it. It seemed to go up when I was accelerating at the start of an interval, which makes sense. I'd assume this might be something useful to work on, so you can keep the power going during a stride.
• Landing Position - This doesn't give any information on how foot landing changed during the run, only a 9-sector "heat map" of the foot. I was awarded "Neutral Midfoot" despite the picture suggesting I ran on the outside quite a bit - though I'm not sure how bad that is (surely it would be difficult to land on the arch?). I was just pleasantly surprised not to be told I spent a lot of time heel-striking.
• Body Angle - This is mentioned on the website but does not appear in the metrics - which is a shame as I read the "Chi Running" book shortly after starting running, in which this is a big thing. Presumably, though, it's quite difficult to tell whether a runner is leaning from the ankles (good) or waist (bad) with only a chest pod.

Third Run - Treadmill

I've been running a lot on the treadmill this year owing to small children (latest just hit 10 weeks!), and moving to an area where the main outdoor running is steep bogs (or joining the "pavement lane" as Bradford's drivers seem to consider it).

The good news was that with the treadmill, I had no Bluetooth disconnections (despite using Bluetooth headphones, although they were connected to my PC rather than phone). The app seemed to work well, though it was difficult to know which screen to leave it on - the main run screen only allows a single metric to be displayed, or to rotate through a few one at a time. There's also a "Data" screen which displays all metrics, but they're not laid out with easy reading in mind. However, I appreciate the app is intended to be mostly used with audio.

Unfortunately, as mentioned above, the app doesn't properly support indoor running and still tries to take pace from the GPS - I hope that it might be possible to add "footpod"-like features to the pod firmware and take a more accurate pace direct from there.

I was doing a 2-hour glycogen depletion run so I didn't have the coach turned on, but as far as metrics go there were a few points of interest: G-landing was much lower, presumably from the treadmill's cushioning. Something to watch when I'm expecting my legs to cope with the world outdoors! Running Efficiency showed the same gradual decline, presumably as my legs became more tired (and out of glycogen!). Time in Air and Ground Contact Time appeared much closer to the mirror-image I'd expected (though the cadence was near constant). Brake Effect was higher, but the chest pod was in a different place so I'm not sure how to interpret that.

Overall

I'm impressed. This seems to be the first product that doesn't just present you with a morass of numbers about your running and expect you to deal with them. I've not had a coach since my rowing days, and my experience with SHFT.run is the closest I've had to that - I was being guided to change my running technique with immediate, usable feedback. Whether the technique I was being guided towards was a good one is not something I can comment on - but at least I can say it doesn't seem to be anything as simple as a blind "forefoot strike, 180 cadence" that you sometimes see on internet fora.

Whilst the coach is brilliant, I feel the software around it could do with a little work. The ability to work without GPS - using the foot pod as a footpod - seems a logical improvement, and would suit those who like to do their recovery runs on treadmills. And the process around Bluetooth connections could do with some work too. I have to assume my phone doesn't have as strong a Bluetooth antenna as the ones SHFT.run have tested with (though I've not had issues with other Bluetooth devices), but equally the app should be able to reconnect after a signal loss rather than just giving up entirely (and dropping GPS, and losing all results up to that point). On the plus side, the SHFT.run team have been responsive in looking into my connection issues, and even seem interested in getting it to work on my watch!

SHFT IQ

And back to how I became interested in SHFT.run. During one moment when the chest pod but not the foot pod lost connection, I noticed that most of the more 'iffy' metrics seem to be coming from the chest pod (eg Watts, Brake Effect, Body Bounce) whereas the foot pod ones seem to make much more sense. SHFT IQ is a new single-pod device which has just finished on Kickstarter, and is now available for pre-order on Indiegogo. I gather the internals will be different (they're using Intel Curie), but this could be a very interesting device at a nice price point. If it adds any kind of ability to work without a phone (I'm assuming the coaching would be difficult) then it could be an excellent footpod, with the added benefit of technical coaching. I look forward to seeing it!