In a recent development, spend enough time comparing power meters, and you’ll quickly learn that they don’t always agree. Put a power meter on your crank and another on your pedals, and the numbers may not match exactly. That can be frustrating when you’re trying to figure out which one is right, but for most riders, a small, consistent difference isn’t necessarily a problem.

According to recent reports, if you use a power meter for training, what matters most is that the numbers you’re seeing today are comparable to the ones you saw last week or last month. Whether your threshold is actually 250 watts or 255 watts matters less than knowing that an interval at that power represents the same effort from one ride to the next.

According to recent reports, but what if your power meter isn’t the only thing affecting those numbers? What if the bike computer recording your ride is changing the data before you ever upload it to Strava or TrainingPeaks?

Sources indicate that I recently noticed something odd while testing a cycling computer. At first, I was doing this to compare GPS track accuracy, but since I was testing, I also paired all the devices to my heart rate strap and Quarq power meter. When I compared the ride files afterward, what stood out was that their power numbers didn’t match. Initially, the differences were small enough. But when I looked more closely, I noticed it immediately. Every time I stopped pedaling, one computer still recorded power.

Before I get into what the heck sticky watts are, I should say I first learned about the issue in 2025 when I saw a video from Shane Miller, who goes by GPLama on YouTube. The video is called “Power Meter Data Recording: The Wild West?”, and it’s worth watching. Shane was the first person I saw digging into this particular recording issue with cycling computers.

So what the heck is it? Sticky watts happen when a cycling computer keeps recording a power value for a few seconds after you’ve stopped pedaling. Instead of dropping to zero, the last power number gets “stuck” in your ride file until the computer finally catches up.

It has been revealed that let’s say you’re riding along at 250 watts and stop pedaling as you approach a corner. Your computer might continue recording 250 watts for a few seconds while you’re coasting. On a graph, that can look like a flat line extending beyond the point where the power should have dropped.

Industry observers note that A few seconds might not sound like much, but think about how often you stop pedaling on a typical ride. Every corner, descent, stop sign, or moment spent coasting in the group is another opportunity for those extra watts to show up in your file. And if you stop pedaling after a hard effort, the computer could be holding a much higher number than 250 watts, which can skew your data.

It has been revealed that it’s also worth distinguishing this from the setting on some computers that lets you exclude zeroes when calculating average power. That setting changes how an average is calculated. Sticky watts affect the power values recorded in the ride file itself, so those extra values can still be there when you upload the ride for analysis.

In a recent development, on one ride, I compared the DURA 2 with a Garmin Edge 850 and an Edge 1060. The two Garmins recorded almost identical power numbers, but the DURA 2 consistently came out higher. When I looked more closely at the files, I found repeated instances where the COROS continued recording the last power value after both Garmins had dropped to zero.

The difference was especially noticeable during a series of short, hard efforts. After one effort, the DURA 2 recorded 565 watts for four seconds while both Garmins recorded zero. After another, it held 656 watts for three seconds. Across a roughly three-minute section containing several of these efforts, the Garmins averaged about 260 watts, while the COROS recorded nearly 307 watts.

In a recent development, those short efforts make the problem easy to see, but the difference showed up across the ride, too. In the roughly 98 minutes of matched power data I compared, the DURA 2 averaged about 186 watts, while both Garmins averaged about 179 watts.

Recent findings suggest that on a second ride, I compared the DURA 2 with a Hammerhead Karoo, Wahoo ELEMNT ACE, and Garmin Edge 850. Again, the COROS recorded higher power through a section with several hard efforts followed by coasting. The Garmin recorded lower numbers than the Karoo and Wahoo in that section, which reminds us that differences can go both ways when computers record the same power meter.

I was honestly surprised to see this on the DURA 2. The original DURA had similar issues early on, but its power recording improved with subsequent updates, and it performed well in GPLama’s 2025 video presenting the matter. I don’t want to turn this into a story about picking on COROS. The firm has a good history of updating and improving its products, and I expect it can address this issue on the DURA 2 as well. For now, these files show why it’s worth paying attention to what your cycling computer records, even when you trust your power meter.

If you mostly ride for fun and don’t pay much attention to your power data afterward, a few extra watts in your ride file probably won’t change your life. But if you use that data to track your training, compare rides, and work with a coach, those differences can matter.

Recent findings suggest that the big problem is that sticky watts don’t necessarily add the same amount to every ride. A steady climb with little coasting might show no difference. But a ride with lots of corners, stop signs, descents, or short, hard efforts followed by coasting could produce a much larger one. That makes it harder to compare rides or judge how much work you actually did.

In a recent development, the easiest way to check is to record the same ride on two cycling computers paired to the same power meter. You don’t necessarily need to buy a second computer. If a friend has one you can borrow, pair both to your power meter and record a ride with a few clear transitions from pedaling to coasting.

Sources indicate that alternatively, you can look closely at the power files your head unit generates. Do a couple of hard efforts, like sprints, where you stop pedaling at the end, then zoom in on those sections. Look for a flat section where one computer continues recording the same power value for several seconds after the other has dropped to zero.

And yes, I know that I’m definitely in some deep weeds over here, and it can be tough to argue that a cycling computer recording a few extra watts is a huge problem worth tackling. For many riders, it might not matter at all. But the current crop of cycling computers is also the most expensive we’ve ever seen. These devices are marketed and sold as precision training tools, and accurate data recording is central to that. These devices should do that, and riders shouldn’t have to question it.

Sources indicate that test Editor Dan Chabanov got his start in cycling as a New York City bike messenger but swiftly found his way into road and cyclocross racing, competing in professional cyclocross races from 2009 to 2019 and winning a Master’s National Championship title in 2018. Prior to joining Bicycling in 2021, Dan worked as part of the race organization for the Red Hook Crit, as a coach with EnduranceWERX, as well as a freelance writer and photographer.

→ Read original article