If you actually know the actual, Bob's your uncle. But the actual is going to be different every day depending how much air you put in your tires. If you look it up in the table that came with the sensor, it'll probably be wrong because the same tire will be different sizes on different rims. Auto basically does a rollout test using GPS to measure the wheel size.

I recently realized my Wahoo Speed Sensor's battery was dead; possibly it's been dead for months/ thousands of miles. Didn't matter. I'd been using the AUTO setting, rather than tire circumference, which then defaults to GPS. My speed and distance results have been spot on. BTW, when I did use tire circumference setting, I got very skewed results over a 40 mile ride. I would have to enter a false circumference measurement to get results to match up with reality. (reality being riding alongside known exact distances). So, at times when relying on GPS under heavy tree cover for instance, I may not get an accurate display, but at end of ride my speed and distance results are more accurate without using a speed sensor.

I agree that the Auto setting is the best way to get accurate, calibrated results from the speed sensor. But I don't agree that the setting means you're defaulting to GPS for measurement. All Auto does is use GPS data to periodically calibrate the wheel sensor, which continues to provide the distance and speed data.

I'm not really sure why anyone wouldn't use one.

Is it absolutely necessary? Of course not. But if you've purchased a $200+ GPS unit to track your rides and provide real-time on-bike data, spending another $25 to attach a sensor to the rear hub to help it work better seems like a no-brainer.

Depends on use case.

Oddly, I just ran across an example: someone was looking at the aerodynamics of handlebar bags and claimed that the bags could act as a fairing to reduce drag rather than as a parachute to increase drag. To support his claim, he coasted down a couple of hills with and without the handlebar bag. He reported the maximum speed attained on each run, and observed that max speed with the bag was quite a bit higher than max speed without the bag. It turns out that he was using GPS for speed and in the handlebar bag run, the speed was anomalous: it jumped up by ~10 km/h in one second, stayed high for 3 seconds then dropped back down in one second right after. On the run without handlebar bag, the speed was smooth. I think if speed increased by 10 km/h in one second, then a couple of seconds later it dropped by 10 km/h in one second, that would be pretty noticeable. I suggested that if he really wanted to know how aerodynamic his handlebar bag is, he should use a dedicated speed sensor.

But not everyone has this particular use case.

It isn't absolutely necessary, but it does provide faster, more accurate data, especially on grades. I think GPS only measures the horizontal distance between two points. The wheel sensor will measure actual distance on the road.

I only use the speed sensor for trainer rides, outdoors I just let the GPS do it's thing, haven't had any issues. If your ride is "ruined" because your data is slightly off, then you prob need to chill out a bit.

I can understand someone just not bothering to ever buy the sensor, but if you already own it and use it, why would you take it off when you ride outside? Different wheelset?

Two bikes with different wheelsets, I don't bother moving it, if I happen to have the training wheelset on, then it's there.

As some previous posts mentioned, a rear wheel speed sensor is useful.
1. for an indoor rear wheel trainer.
2. your recorded speed goes to zero as soon as you stop, with none of the usual GPS drift. I used to get some 1-2 mph GPS readings when stopped for lunch.
3. more accurate grade percentages, especially with GPS units that have barometers for elevation. Slight distance errors when climbing will affect the grade quite a bit.

Speed sensors are really good at telling you the magnetic ore content within the bridge you're riding over.

Because time is what actually matters for performance? I.e., the lower the time for a particular distance, the higher the placing, whether in races, group rides, or KOMs.

I have a GPS computer because that's what I use to see all of the things that do matter in my rides: notably power and time. When racing, it's to record the data to look at afterwards.

Speed as an absolute in and of itself simply doesn't matter that much to me, though of course I do check the speed provided by the GPS fairly routinely and all.

As others have said, the sensor that you can strap around the rear wheel hub is useful on a trainer or when using the rollers with a fork mount. Also, you can save some with some of the lower priced ANT+/BLE units that Amazon sells. Believe mine was around $22 and there were some under $20. It works flawless with the Wahoo fitness app and with the Cateye app (I don’t even have a Wahoo unit, just a Cateye Padrone Smart).

One interesting thing regarding GPS drift. If I have GPS on when riding the rollers, the map shows me wiggling up and down the street near my house, so that gives some idea of the inaccuracies when used to measure speed.

This is the one I have been using on my rear hub indoors.

https://www.amazon.com/moofit-Blueto.../dp/B085ZGKDDT

I regularly ride a rail trail that’s is almost all canopy and neither the wahoo or garmin can maintain an accurate speed it’s not even close. On the regular road it’s fine but the speed sensor seems a good bit more accurate on accelerations and stopping even then. I like it, for as much money as I spend on that stupid bike the speed sensor pretty cheap cost wise

Whoa. That's a little big, and pretty ugly.

Sorry, it is way out of scale. Unit is actually about an inch square.

Noting that when the battery goes out on the speed sensor, it starts sending all kinds of crazy data to the head unit. This happened to me today so I unpaired it and continued on my ride with the GPS speed data.

And I had that exact same experience. Once you notice unexpected speed results, or even a stuck speed, you'll begin to lose part of your ride if you don't disconnect the sensor ASAP.

Crossing metal decked bridges, the sensor can get inaccurate.

Device dependent. I’ve noticed in Garmin’s when not using a speed sensor, the at that second speed reading is sometimes 4-5 seconds behind actual speed. I know this as if I speed up a bit, the speed display lags. This goes away when using a sensor. A sensor is useful when on a group road ride and trying to minting a steady speed.

Four years later... Are people still using speed sensors when there is GPS? A couple years ago I picked up a computer and it came with the cadence/speed sensors and I was confused why a speed sensor was a thing. I suppose its primary use case is for indoor training?

Primary use is still for those who wish to see accurate real time speed on their display. GPS speed is still pretty laggy on my Garmin 530.

In general, this thread has not been about old school non GPS cycling computers that use a magnet/sensor on the wheel and fork. It’s about speed sensors used in conjunction with a GPS device. As noted in the assorted posts, GPS can generate errors due to poor signal in wooded areas, tall buildings, etc..,, the speed reading can lag with the basic GPS signal, and you can use the sensor on a rear wheel on a wind trainer. Etc….,

Speed sensor is also useful if you ride in areas with poor GPS signal, like the coastal redwoods. Without a speed sensor, my Garmin auto-stops (and beeps) until it reacquires a signal.

It's also useful if you display grade or VAM, which can go whacky when the GPS signal is weak or inaccurate.

The speed sensor is also a "distance traveled" sensor.