Cyclists know that staying in the draft saves watts. It’s been tested outdoors, and it’s been tested on Zwift. This knowledge is important to understand and put into practice because bike racing is all about conserving energy so you have it when you need it. Drafting is the #1 way a cyclist conserves energy while remaining in contention for the win.
But how many watts do you save in the draft – and does that change based on whether you’re on a climb, a descent, or a flat road? We decided to to run a few tests in the Zwift Insider Lab to get some hard numbers.
About the Test
For our test we had two riders: Rider 1 pulling on the front, and Rider 2 drafting behind Rider 1. Both riders were 75kg in weight, 183cm tall, and riding Zwift Aero bike frames with Zwift 50mm wheelsets.
We took these riders over the Watopia Hilly KOM segments (forward and reverse) with Rider 1 pulling at 330 watts for one test, then 270 watts for the other. We also took them on Ocean Boulevard for a flat test at the same wattage. Rider 1’s job was to hold the wheel using as few watts as possible.
These tests were made with other Zwifters on the road, and required constant adjustment of Rider 2’s wattage to properly hold the wheel. This means the test results aren’t perfect. But they’re still certainly useful!

Drafting Up Climbs
On the Hilly KOM Forward climb:
- drafting a rider holding 330 watts requires 320 watts – a 3.1% savings
- drafting a rider holding 270 watts requires 264 watts – a 2.3% savings
On the Hilly KOM Reverse climb:
- drafting a rider holding 330 watts requires 292 watts – an 11.6% savings
- drafting a rider holding 270 watts requires 244 watts – a 9.7% savings
The steeper the climb, the slower the speed. The slower the speed, the less effect drafting has since drafting is all about reducing wind resistance and wind resistance isn’t an issue at slow speeds. So we see on these climbs there is certainly some savings in the draft, but it’s not nearly as significant as we’ll see on flat roads or descents.
The savings is greater on the reverse climb because that climb isn’t as steep as the forward climb. Therefore, speeds are faster on the reverse KOM, and drafting has a greater effect.
Similarly, the savings are greater when drafting the 330-watt rider compared to the 270-watt rider, because speeds are faster behind the 330-watt rider, so drafting has a greater effect.
Drafting On Descents
On the Hilly KOM Forward descent:
- drafting a rider holding 330 watts requires 207 watts – a 37.3% savings
- drafting a rider holding 270 watts requires 145 watts – a 46.3% savings
On the Hilly KOM Reverse descent:
- drafting a rider holding 330 watts requires 227 watts – a 31.3% savings
- drafting a rider holding 270 watts requires 222 watts – a 17.8% savings
Descents are much faster than climbs, and the main thing slowing riders down on a descent is air resistance. Drafting greatly reduces air resistance, so riders in a draft on a descent can sit in with very low watts even when the rider ahead is hammering.
The steeper the descent, the greater the savings from the draft. So we see the steeper forward descent gives us a greater savings than the reverse descent.
And just like on the climbs, the savings will be greater when drafting the 330-watt rider compared to the 270-watt rider, because speeds are faster behind the 330-watt rider, so drafting has a greater effect. We don’t see that exactly in the results above, because holding the wheel over changing gradients was difficult in our tests!
Drafting On Flat Roads
On Ocean Boulevard:
- drafting a rider holding 330 watts requires 241 watts – a 27% savings
- drafting a rider holding 270 watts requires 208 watts – a 23% savings
We would expect the power savings on flats to fall somewhere in between the savings we saw on the climbs and the descents, and that is indeed the case (apart from the hilly KOM reverse data at 270 watts which is probably just a testing error – like we said, these tests weren’t perfect!)
Here we see once again that riders achieve greater savings when drafting a more powerful rider because speeds are higher, so drafting has a greater effect.
Takeaways
There are several useful takeaways from these test results:
- Descents provide a wonderful opportunity to recover, if you play it smart. Stay out of the wind, surf the wheels, and apply just the power needed to stay in the pack. Of course, there’s always the supertuck as well, but that is best used on longer, steeper descents.
- Drafting is a significant factor on shallow climbs like the hilly KOM reverse. So play it smart and stay on a wheel unless you’re attacking.
- The faster you’re traveling, the bigger that “wattage window” is where you’ll stay with the rider you’re following. That’s why our descent test results were a bit messy. (The “wattage window” is the minimum and maximum wattage you can hold and without being dropped or dropping the rider you’re trying to follow.) This means you can potentially waste a lot of watts by hammering too hard–so play with it, ease up, and see how little you can work to hold the wheel on descents and even flatter roads. With practice, you’ll get a feel for it.
- If you think this savings is impressive, consider this: the draft effect is even stronger in larger groups on Zwift! This is difficult to test (requiring many computers, Zwift accounts, etc) but we know there is a stronger draft behind 4 riders than behind 1.
Your Thoughts
Do these results make sense to you? Got questions, or additional takeaways from the data? Share below!
The data are helpful, and my anecdotal experience matches the results you folks observed. Since I’m a heavier rider, however (94kg), my typical race strategy is to reserve recovery drafting to the flats. Going uphill, I’m constantly trying to just minimize time losses. I can beat other riders up very short climbs, and hang with good climbers up short climbs. But even if I hammer all the way up the Zwift forward KOM, I’m going to lose some seconds to a good climber. So in a race, I’m always hammering the downhill sections, in order to maximize the benefit my… Read more »
dude you need to drill it and get a gap at the beginning of a descent with a couple of other big guys, once you establish a gap it’s harder for the lighter guys to catch up. Shhhhh, it’s a “big guy secret”….
As one of the little guys, I’ve definitely seen people take advantage of this. I have to make sure I don’t leave it all on the climb so I’m ready to grab a wheel when the big guys come screaming down after they crest.
I believe the draft effect is less at the back of a group as opposed to in the middle of a group. Although, this could just be sticky draft playing tricks on my mind when the other riders ahead but toward the back of the pack are swapping positions, making it difficult to overtake.
Yeah, that’s a tough one to test! It’s definitely a more dangerous place to be in a race…
I think you’re probably in a better position than anyone, Eric, to recruit a bunch of people for a stop and go group ride where you measure the draft effect at various positions in the pack – everyone in a line vs one big bunch, etc… Would love to see those results!
I agree, it feels like that to me too, it feels easier up front than in the back of a group. We need some data on this…. 🙂
#doubledraft?
I’ve always wondered if the size of the rider in front influences the amount of draft the drafting rider gets? My guess is no but it would be more realistic if it did.
I would love to see this tested too, although it is obviously hard to tell the size of the riders in front of you, but I know how big I am! Eric this would be one to do pretty easily by replicating this test but with a couple different rider weights/heights up front yes?
Sure it does matter. Drafting a semi is easier than a fiat 500. Same principle with the heavier – bigger guy vs the small light one.
Any idea if the watt saving differs depending on where you are in a group? i.e. 3rd back versus right at the back?
I believe it does, but it’s hard to test so we’ve got no hard numbers. I basically treat it like riding outdoors…
As a guy who has swept group rides a couple of times, I have a bit of knowledge on the subject of the draft when it comes to group rides. I usually sweep for B and C group rides (power between 2.5w/kg and 3.5w/kg). Riders usually save about 1 to 0.5w/kg when drafting behind just me. When there is a team of riders working to get the dropped rider back, that rider sitting in the draft will save 1.5 to 1 w/kgs. These are not scientific numbers so they may not be exactly right but near enough to be useable.
What about size of rider you are drafting behind IE their height any evidence this effects it or not?
According to ZwiftHQ it does affect the draft, slightly. But it’s pretty hard to test!
Hi Eric, a couple of questions/thoughts… 1. If the forward climb is steeper than the reverse climb, doesn’t that mean the reverse descent is steeper than the forward descent? But I would agree the savings would be bigger on a steeper descent. So, is the descent section labeled the wrong way around maybe? Or that part of the test just didn’t work out as expected? 2. Generally speaking, I agree the higher the speed, the higher the savings. But because weight helps on a descent I think it complicates things a bit. The front rider needs to put in some… Read more »
Ha – yeah, I guess my hill naming could be confusing. When I say “Hilly KOM Forward” I’m referring to one particular stretch of road. Then I label that as either a climb (if you’re riding up that stretch) or a descent (if riding down it). Does that make sense?
Re point number 2, you’re right–but one challenge with checking that is, if the drafting rider stops pedaling, he’ll auto-brake and slow down, at least on descents that aren’t real steep.
ah, right, so Hilly KOM Forward descent means doing the Hilly KOM Forward climb in reverse. Got it! haha sorry.
Yes, I see the issue with the auto-braking. Not suggesting you should bother, but I’d be interested to see a whole bunch of data points at different power levels when descending.
How does trainer difficulty affect your descent speed? If a light rider had 100% and a heavy one had 10% would it be easier for the light rider to draft?
Trainer difficulty doesn’t directly affect your in-game speed anywhere. What it CAN do it make you run out of gears, which you sometimes see riders complain about on descents when they’re running Trainer Difficulty at 100%.
So going over the top of a climb without more pedaling you free wheel just as fast regardless of of your trainer difficulty?
Yes. Again: your in game speed isn’t affected by the trainer difficulty setting. Just changes what you feel on the pedals!
Hi Eric Not sure who to contact for this issue….love Zwift….noticed over the last year that when I race and cross the finish line, the times crossing the finish line and those that appear in the zwift app and zwift power results are always off by 15-20 seconds. I raced today in CP and crossed the line in 26:50. However zwiftpower has me at 27:07 even though I was way past the finish line. When I use the slider on the app it has me at 26;50 as well. Strava has me at 26:50 for the 2 laps as well.… Read more »
I think perhaps Zwift doesn’t start timing you for the race until you cross under a particular banner or live on the course. But your ride elapsed time starts counting as soon as the event starts. So I think that’s the cause. You could always contact Zwift support about it.
Here is a thought on how to test the effect of a large group draft. Golden Cheetah has a Aerolab calculator which allows a ride to be analyzed for Crr and CdA. A fit file from one of the leaders of a group ride on Tempus Fugit (like PACK or ZZRC) could be analyzed using this tool. The group leader is always in the bunch draft, you know the Crr so the CdA can be calculated on the flat course. A hilly course would also work but there is the chance a leader is not in the bunch draft on… Read more »
I wish draft downhill worked better. I have to go HARDER downhill to keep up with a group on a decent rather than easier. I guess it’s because I’m light at 70 kg, but it’s annoying the group pulls away at 2.2 w/kg while I’m up to 3.2 trying to keep up (after I had had no trouble keeping up with everyone at 3.0 up the climb)