Coros Pace Pro – GNSS Accuracy Test

The results in this article are from a single GPS/GNSS accuracy test on the Coros Pace Pro using a standard methodology I’ve applied to almost every sports watch produced in the last decade. Obviously, I also perform other kinds of extensive GPS tests. This particular test interests me as I score watches on an almost like-for-like basis, and the test conditions over 10 miles represent many real-world conditions you’ll encounter. It’s a good test.

I used to record DOP/HDOP/VDOP and satellite availability, but that factor is no longer an issue in determining the accuracy of modern watches. Whilst satellite availability does vary minute-by-minute, multi-constellation watches now have tens of satellites to choose from – there used to be issues linked to occasionally, say, only having 6 GPS satellites available.

The test uses the first production firmware, ie v3.0708.0 (GPS: v165.20.24-8178). For comparison, the PACE 3 GPS test I performed a year ago used v3.11 GPS3.0.3. This confirms that a different Airoha chipset or algorithm is used in the Pace PRO.

Buy Coros Pace Pro: USA – $349, UK – £349, EU – 399€

Test Results

Distance: The distance recorded was exactly 100% of the median distance of all tests. All that means is the over- and under-recording at specific points on the route perfectly cancel each other out.

Pace: The responsiveness to instant pace displayed live on the watch was 12 seconds. That sounds bad, but other watches can be over 30 seconds, so it’s OK but not as good as the true near-real-time from a footpod like Stryd.

The high-level overview of the route looks as good as all the best previous tracks, and all appear to follow my actual route perfectly. For your pretty track on Strava, almost all sports watches are OK. Pace Pro is no exception.

I use the satellite image as the human-made map can sometimes put trails and roads in the wrong place, especially if there is tree cover. Indeed that IS the case on this route.

Points of Interest

Here, I run a sharp bend near a building in Twickenham, home of World Rugby and the Cabbage Patch 10-miler, then under a shop’s awning. Pace Pro was relatively poor here, taking over 30 metres to recover to a more accurate position toward the bottom of the image.

Again, this next section is poor, where I run in the middle of a minor road with 3-metre walls on either side and tree cover in places. Watches are usually better than this, and I consider errors of over 5m poor.

Next is a gently curving riverside path under tall, relatively dense tree cover. Zoomed out, Coros seems OK, but when zoomed in a little, it’s not quite as good as I’d like.

Crossing Richmond Bridge gives a clear sky for PACE to ready itself for a tricky loop down steps and back under the bridge. As you can see, something weird happens on the easy bit before turning right and running next to a wall where, surprisingly, it’s OK.

We are now on the rather pretty Richmond Riverside. We’re next to Elon Musk’s old PayPal offices and tens of metres away from Ted Lasso country. I loop around a tree at the top of the image, but had I continued for 40 metres; I would be at the site of the old Royal Palace of Richmond where Queen Elizabeth the First waited anxiously for the Armada to be defeated (or not…it was…she lived). The point of the GPS track’s precision here is not to cross over the outward track on the return leg. Pace Pro does do this slightly, but it’s OK. It then has to return under Richmond Bridge, and clearly show it has taken the leftmost tunnel, which it does. It did well here. History lesson over.

Next comes a straightforward section where Coros Pace Pro should excel…and does excel. Running past Petersham Meadow, you may see some cows grazing and, in the distance beyond them, Petersham nurseries, where the Richmonderatti go to buy overpriced objects for their gardens or very overrated food (in my experience), albeit in a lovely setting.

Next is an avenue of trees next to the German School and Ham Polo Club. If you want your kids to be educated alongside all the German kids in SW London, this is the place to go. If not, you can watch grown men and women playing hockey (polo) on horseback. I was once chased by a badger here at night; luckily, I was much quicker.

I’d give Pace PRO full marks here, even though it’s not perfect.

Then we come to Cut Throat Alley. You can probably guess why it was called that, but, like me, your guess would be wrong as its name is a corruption of Cut THROUGH Alley, where historically, the farm workers could cut through the Ham House lands and avoid a detour. Anyway, I chose this alley over a decade ago for the route as I was convinced no GPS device would ever get it right. In more recent years, they have started to nail it. Pace Pro doesn’t quite nail the second part of the alley, but I’ll be generous and give it full marks, knowing that it has already slipped up on more manageable sections elsewhere so that the overall score won’t be too severely impacted.

The sweeping curve of Riverside Drive shouldn’t be hard for watches to perfect. Often, watches used to struggle with long curves, but Pace Pro nails this, as it should.

Oddly, Coros nails the next more difficult piece on the trail under the trees but is slightly out in the easier, more open section. I’ll have to dock a mark here. Very strange.

The following section shows a good performance under some quite serious tree cover.

This section is impossibly hard. I run under an awning next to a tall building with other tall buildings on the other side of the road. After a 90-degree turn, I ran very close to the multi-story John Lewis building and almost made a 180-degree turn close to the same building and into a tunnel under it. The tunnel has no GPS reception and must be 60 or 70 metres long. So the test here is many-fold: to see how reflected signals are handled off tall buildings and if the accelerometer can make any sense of a lack of signal. After the tunnel, we see how quickly the watch can recover to the correct position. From memory, only one watch ever fluked this section. As I was lenient in the previous section, I’ll have to score harshly here as the track doesn’t look good.

The final section I score on is almost 4km long and represents easy suburban running. I expect a perfect score here but surprisingly rarely get an ideal track; often, for no reason, watches show me running on the wrong side of the road – i.e. an error of over 5m. Coros does NOT do that and produces a great track except at the start of the section, with some close buildings to navigate. This section is near the well-known Sigma Sports bike store and reliably indicates how dual frequency chipsets handle (or not) reflected signals. The whole point of the Dual-frequency chipsets that the brands sell you at premium prices is that the two frequencies produce two paths that should be the same. If they are not the same, the watch should discard that particular signal as it has either been reflected or refracted, which means it’s giving an incorrect reading. It is VERY CLEAR to me that MANY dual-frequency chipsets are not presently doing that. Coros is doing better than most here, not perfectly, but pretty well. You can see some of the other greyed tracks from leading watches on the same image and how they veer into the road. However, even doing the max zoom-in on the bottom corner, you can still see an error of 2-3m – as of 2024, I think this is as good as you can expect.

Take out

Coros Pace Pro scored a creditable 81% compared to Coros Pace 3, which scored 92% – the latter is the joint best-ever result. Pace Pro’s score oddly uffered in the medium-difficulty sections of my test.

Other reviewers have commented that Pace Pro has improved accuracy over Pace 3. It hasn’t; perhaps Coros is still bedding in a new GPS chipset. Another reviewer (I can’t recall which) also said that a post-launch firmware change lowered the GPS accuracy.

This sounds odd to me. Whilst the antenna and algorithms are essential to accuracy, the power available to the chip is also an important factor in realising precision capabilities. I genuinely hope Coros is not guilty of forcing more power through the firmware some reviewers received to fool them into reporting high accuracy levels before launch. Only then to lower the power and quality to boost battery life.

Anyway. I can only test what I have when I have it. My whole suite of GPS and HR tests is far from trivial to complete. If every brand changed firmware after my test results were initially published, testing the watches would be a full-time job. So they need to get it right before launch.

I’ll ask Coros to comment below if they want to.

Buy Coros Pace Pro: USA – $349, UK – £349, EU – 399€