How accurate is Strava estimated power? Comparing Strava watts with 'real' watts from a power meter

When uploading a bike ride to Strava without using a power meter, you get an 'estimated power' number for your effort. But how accurate is that figure? And if it is any good, surely the days of expensive power meters limited? It's time to find out how accurate Strava's estimation is once and for all... 

2025 strava watts test Jamie riding road.cc kit (credit: road.cc)

> 1x vs 2x: Are single chainring set-ups the future of road cycling?

If you're a regular road.cc reader then you’ll know that a little while back I switched my Specialized Allez Sprint from Shimano Ultegra Di2 to 1x Sram Red AXS. I've mostly been very pleased with that change, other than one quite significant thing... I had to give up my power meter.

What that means is that I have several years' worth of data with a power meter, and then a year or so of relying on Strava estimated power numbers. Now, however, I'm back on the power meter band wagon and want to know if those estimated numbers are useful or not. 

2025 strava watts test Sram 1x power meter chainring (credit: road.cc)

> Best power meters for cycling

The new Quarq power meter is a good example of just how far power meters have come in recent years. They are now amazingly reliable, accurate, bug-free and for the most part, a lot cheaper than they used to be. Even so, this pro level setup would still set you back about £560 more than the non-power meter option.

It goes without saying that most amateur cyclists don't 'need' a power meter, and it's still very much a luxury. Let's do some tests and analyse the data to see if most of us really could be better off saving our money...  

Example 1

2025 strava watts test Jamie riding road.cc kit 3 (credit: road.cc)

As part of the video above, you'll see that I rode from the office to a cafe at a medium pace. Upon arrival, my trusty power meter (that has been tested against various turbo trainers and power pedals to ensure its accuracy) revealed that I had averaged a grand total of 180 watts.

To find out what Strava would estimate for exactly the same ride, I quickly turned off Bluetooth so it didn't automatically upload, stripped the power data from the ride using FIT file tools and then manually uploaded it to Strava.

Strava doesn't hang about, and revealed that it thought that it should have taken 153 watts for the same ride.

2025 strava watts test Strava screen capture (credit: road.cc)

> 12 reasons why you should buy, and use, a power meter

Clearly that's quite the difference, 15% in fact. I have a theory about why that might be, but first: how has Strava arrived at that number?

Well, Strava says: "When data from a power meter is unavailable, Strava can estimate power using information about an athlete's weight, speed, and elevation change. You'll need to do three things before Strava can estimate your power on a ride:

• Add a bike to your account by following these instructions and ensure that you include the bike's weight. If you don't create and select a bike for your activity, a default road bike weight of 9kg, or 20lbs, will be used.
• Enter your personal weight in your profile settings. It's helpful to enter your weight including your cycling clothes, shoes, and water bottles. 
• Lastly, Strava will need trusted elevation data for your activity from a device with a built-in barometric altimeter or automatically calculated from Strava's elevation basemap."

Obviously, the accuracy of the estimated power is going to be highly dependant on the accuracy of this information, so I was careful to accurately input these weights and used a Wahoo Elemnt Roam with built in barometer to log the rides.

2025 strava watts test Wahoo Roam power data (credit: road.cc)

> Best cycling computers

So, why do I think Strava underestimated my power? I think it was because I had a headwind. What that means is that I was travelling slower than I would have been if it was much calmer or I had the wind behind me. Strava doesn’t know that I had the weather against me, so thinks I was just putting less effort in.

Let's look at another scenario... 

Example 2

2025 strava watts test long ride loop (credit: road.cc)

To try and minimise the effects of the weather, I next went on a ride that was a particularly still day, much longer and a round loop rather than heading in just one direction.

At the end of the ride my power meter informed me that I had averaged 153 watts (including zeros).

Repeating the process from example 1, I found that Strava estimated a wattage of 159 watts, a huge amount closer to the real figure. In fact, the difference of just 3.7% is not too far off the claimed accuracy of some power meters.

Finding trends

2025 strava watts test Jamie (credit: road.cc)

Of course, we’re going to need more than a few rides to find some trends, so I spent a rather tedious morning stripping ride uploads of their power data, and repeatedly getting told that my ride was a duplicate. 

To get a better idea of the capabilities of Strava's algorithm, I took other peoples' rides too. Some slower riders, some faster riders, lighter, heavier, flat rides, lumpy rides etc, to see if that makes any difference.

So, are Strava watts a threat to power meters? Uhhhmm, not really… firstly, a power meter can be used to inform your training while you’re out on the road or trails, and at the moment Strava doesn’t give any live power estimates on a head unit or mobile device.

Rouvy vid Jamie riding on a flat road (credit: road.cc)

> Cycling tech innovations that never took off

Even so, many people don’t actually care about the number while they're riding. They ride as hard as they like, maybe give it a dig up a hill and then only care about the numbers when they get back, and want to compare their efforts to previous rides.

For this, Strava watts undoubtedly has a great deal of value. The estimated wattage from solo rides on regular routes did seem to be pretty accurate, and although I would never take that number as gospel, I do believe that it gives a good indication of whether one ride was harder than another on a like-for-like setup.

2024 midrange bike vs superbike climbing road.cc kit jamie van rysel rcr (credit: road.cc)

In fact, we found this number to generally be within 4-5% of the actual figure, which is nearly as accurate as some of the cheaper power meters out there.

Where Strava's estimated power a little bit undone, even though the app's algorithm is undoubtedly clever, is that it doesn't properly take variables into account. It doesn’t know the state of the road surface, what tyres you’re using, how aero you are, or how billowy your clothing is. Taking this estimate as anything more than that, an estimate, is risky business. I wouldn’t for example, want to take my FTP or set my training zones based on this number.

Strava app icon in iPhone (credit: Focal Foto on Flickr)

In Strava's own words: "Average power is most accurately measured by a power meter, though if you don't have a power meter we give a rough approximation through our power estimator." 

That said, in some cases, I’ve been thoroughly impressed by how accurate the Strava estimated power data is. On longer, steeper climbs the numbers look quite similar, probably due to the fact that how fast you get up a hill is based far more on your watts per kilogram (W/kg) and helps to reduce other variables such as wind and drafting. The more you sway away from rides that rely on W/kg, the less accurate the Strava estimated power data seems to be.

2025 strava watts test Jamie riding road.cc kit 4 (credit: road.cc)

> How to improve your power-to-weight ratio 

So, how accurate is Strava watts? Well, as with most cases, it depends. On most still, solo rides, we’ve found it to be about +/-5% from the actual power figure.

This is great for comparing a regular route, for example a regular post-work ride to see if you’re getting stronger. It isn’t, however, a rival to the power meter just yet. A good power meter will be accurate no matter the weather conditions and no matter who you’re drafting. It can be reliably used during your ride to better gauge your efforts.

The future

2025 strava watts test Jamie riding road.cc kit 2 (credit: road.cc)

Given the fact that the Strava algorithm is already doing pretty well in some cases, it got me thinking on how it could be improved upon in the future. I think that it would be impossible or at least impractical for it to take variables such as drafting, road condition and clothing in to account more than it already does, but I would love to see weather data impacting the estimate more.

I also reckon that Strava could somehow use power data from other people that are on the same ride to further help it improve the accuracy.

Do you use Strava watts, and how have your experiences been? Let us know in the comments below.