Strava estimates are quite accurate on longer steady climbs (think 5-6% or so, that sort of thing), if you've entered all the data well.

On the flat it tends to overestimate / underestimate depending on wind, how good or crap the road is, how aero is your position, and so on, but uphill where gravity is doing most of the resistance it's pretty much in agreement with actual power meter data.

This is a tangent, so hold on, but it is related to the post by Doge

I wish bike GPS's would move away from climate based elevation estimation and use instead a slope based measure. If you have distance, time, slope...........you've got elevation.

Just have the bike GPS zero itself on level ground in your house (assuming you don't own a piece of crap that's super not level).

I feel like the granularity of a climate based elevation sensor does not pickup a lot of the little 5', 10', or even maybe 15' little oddities on some routes. It also probably tosses in weird changes in weather tossing the elevation off.

I see people with the SAME gps units on group rides of 25mi and 1500 feet arrive back with differences person to person of up to 300 feet. That's 20 freaking percent. That's silly.

As I posted above, I like the watch as the best tool for measuring cycling success. The speed and time are believable, but power - not so much. The rider here was 135lbs. Very aero and good wind and bike setup by me to deal with side winds.
Strava knows none of that.
https://cimg3.ibsrv.net/gimg/bikefor...7620d4a46d.png