Most people don't realise this about indoor training. The trainer itself is the least interesting part. The interesting part is what you do on it. But — and this is a big but — a bad trainer makes good training harder than it needs to be. Calibration drift, power spikes, tyre slippage, noise that wakes the kids at 5:30 AM. These aren't minor annoyances. They're friction that chips away at consistency, and consistency is the only thing that actually works indoors.
I've spent the winter testing seven trainers across two pain caves, running structured sessions on TrainerRoad and free-riding on Zwift. This isn't a spec-sheet comparison. This is what I found after hundreds of hours of actual use, cross-referenced with what coaches like Dan Lorang and Tim Kerrison prescribe for their World Tour riders when they bring them indoors.
Here's what actually matters.
Why indoor training matters more than you think
There's a reason the coaches behind Pogacar, Froome, and Roglic prescribe indoor work even when their riders have access to perfect roads. Power accuracy indoors is better. Environmental variables — wind, traffic, junction stops — disappear. Every interval is exactly the interval it's supposed to be.
Tim Kerrison built Team Sky's marginal gains philosophy partly on the back of indoor power data. When you eliminate the noise from outdoor riding, you can see the signal more clearly. Did threshold actually move 3 W, or was it just a tailwind? Indoors, you know.
Dan Lorang has talked on the podcast about how he uses indoor sessions to prescribe very specific intensity targets for Roglic. Not "ride hard for 4 minutes." More like "hold 412 W for 4 minutes at 95 RPM." That level of precision requires a trainer you can trust.
The good news is that in 2026, even mid-range trainers deliver accuracy that would have been unthinkable five years ago. But the differences still matter, and they matter most at the moments that count — the last 30 seconds of a threshold interval, the ERG mode response when you shift cadence, the road feel that tells your legs they're actually riding.
Direct drive vs wheel-on: the real difference
Let me be really clear about this. Direct drive is better. It's not a marginal difference. It's a category difference.
A wheel-on trainer presses a roller against your rear tyre. The power reading comes from measuring the force at that contact point. The problem: tyre pressure changes with temperature. Your tyre wears down over weeks. The contact patch shifts when you rock the bike. All of these introduce error, and the error isn't consistent — it drifts session to session.
A direct drive trainer removes the rear wheel entirely. Your chain drives the trainer's cassette directly. Power is measured at the resistance unit. No tyre. No slip. No drift. The result is +/-1% accuracy on the best units, compared with +/-2-5% on wheel-on trainers.
Here's where it gets really interesting. That 2-5% error on a wheel-on trainer isn't random. It's systematic and it shifts with conditions. So your FTP test might read 5-12 W differently on a cold Monday morning versus a warm Thursday evening. Your training zones drift with it. Your intervals hit different targets depending on when you last pumped your tyre. This is fixable. You fix it by removing the tyre from the equation.
The other differences — noise, road feel, gradient simulation — are all better on direct drive too. But the power accuracy alone justifies the price difference for anyone doing structured training more than once or twice a week.
The trainers: what I found
Wahoo KICKR V6 — the one most people should buy
Price: ~$1,300 / EUR1,300 | Accuracy: +/-1% | Max power: 2,200 W | Max gradient: 20% | Noise: ~58 dB at 250 W
The KICKR V6 is not the cheapest and not the most feature-rich. It's the one that gets the most things right at the same time.
Power accuracy is +/-1%. ERG mode response is fast but not jerky — when you drop cadence, it adjusts resistance smoothly rather than slamming you with a wall of torque. The road feel is the second-best on this list. Zwift, TrainerRoad, Rouvy, IndieVelo — it connects to everything via ANT+ FE-C, Bluetooth FTMS, and WiFi Direct.
The KICKR AXIS feet, included in the V6, allow the bike to rock laterally by a few degrees. It sounds gimmicky. It isn't. After 90 minutes in the saddle, the ability to shift your weight naturally reduces saddle pressure and lower-back fatigue in a way you actually notice. Small change, big difference on longer sessions.
What I don't love: the price. At $1,300, it's a serious investment. And the fan noise at high wattage — above 400 W, you can hear the flywheel working. For most of us, that's only during short VO2max intervals and it's not a dealbreaker.
Wahoo KICKR CORE — the smart budget pick
Price: ~$800 / EUR800 | Accuracy: +/-2% | Max power: 1,800 W | Max gradient: 16% | Noise: ~62 dB at 250 W
The CORE shares the KICKR V6's basic architecture but trims the flywheel weight and drops the AXIS feet. The result is a trainer that's 85% as good for 60% of the price.
Power accuracy is +/-2%, which is worse than the V6 but still substantially better than any wheel-on trainer. Max wattage of 1,800 W is more than enough unless you're a track sprinter. The 16% gradient cap means you won't simulate the steepest Zwift climbs with full realism, but structured training rarely needs more than 10%.
This is the trainer I'd recommend to anyone who's moving from wheel-on to direct drive for the first time. The jump in experience from wheel-on to CORE is much bigger than the jump from CORE to KICKR V6.
Tacx NEO 3 — the best road feel money can buy
Price: ~$1,500 / EUR1,500 | Accuracy: +/-1% | Max power: 2,200 W | Max gradient: 25% | Noise: ~55 dB at 250 W
The NEO 3 is the trainer for people who care about how indoor riding feels, not just whether the numbers are right. Its motorised resistance unit — no belt, no physical flywheel contact — means it can simulate road surfaces. Cobblestones on Zwift's Paris course actually feel like cobblestones. It's a subtle vibration through the pedals, not a gimmick.
The NEO 3 is also the quietest trainer I've tested. At conversational pace, it's barely audible. At threshold, it's quieter than most desk fans. If you train early mornings in a flat or a house with thin walls, this matters.
Power accuracy is +/-1%, matching the KICKR V6. Gradient simulation goes to 25%, which is overkill for structured training but really useful if you ride Zwift's steep mountain routes and want the resistance to match. No calibration is ever required — the NEO 3 self-calibrates continuously.
The downside is price. At $1,500, it's the most expensive trainer here. And at 21.5 kg, it's heavy — not something you'll move between rooms casually.
Tacx NEO 2T — still a contender
Price: ~$1,100 / EUR1,100 | Accuracy: +/-1% | Max power: 2,200 W | Max gradient: 25% | Noise: ~57 dB at 250 W
The NEO 2T is the previous generation, still widely available, and still a brilliant trainer. It shares the NEO 3's motorised resistance design and road surface simulation. The differences are marginal: slightly louder, slightly heavier, slightly less refined pedal feel at very low cadences.
At $400 less than the NEO 3, the 2T is arguably the best value in the premium tier. If you find one on clearance, buy it.
Saris H3 / H4 — the quiet achiever
Price: H3 ~$700, H4 ~$900 / EUR700-900 | Accuracy: +/-2% | Max power: 2,000 W | Max gradient: 20% | Noise: ~60 dB at 250 W
Saris doesn't get the marketing attention of Wahoo or Tacx, and that's a shame because the H3 and H4 are seriously good trainers. The H4 uses a larger electromagnetic resistance unit than the H3, delivering smoother ERG mode transitions and slightly better accuracy at low wattage.
The H3 was one of the first direct drive trainers to get the basics completely right — good accuracy, low noise, solid build quality — and then Saris just kept selling it without much fanfare. The H4 refines the formula. Neither will give you the road feel of a NEO 3 or the polish of a KICKR V6, but both deliver the data accuracy and reliability that structured training demands.
If you're the kind of rider who puts the head down, stares at the TrainerRoad screen, and grinds through intervals without caring whether cobblestones feel realistic, a Saris will do everything you need.
Elite Suito-T — best value in the category
Price: ~$700 / EUR650 | Accuracy: +/-1.5% | Max power: 2,050 W | Max gradient: 15% | Noise: ~61 dB at 250 W
The Elite Suito-T is the trainer I'd point anyone towards who says "I want direct drive but I can't spend over $800." It ships with a cassette pre-installed, which saves you another $40-60 and fifteen minutes of fiddling. Unbox, mount, ride.
Accuracy at +/-1.5% sits between the premium units and the KICKR CORE. Max gradient of 15% is the weakest on this list, but again — structured training rarely asks for more than 10%. ERG mode is stable and responsive. Build quality is solid, and Elite's companion app (My E-Training) is free, which is a nice bonus even if you primarily use Zwift or TrainerRoad.
The Suito-T won't make you feel like you're riding Paris-Roubaix in your spare bedroom. It will give you accurate power data, stable resistance, and quiet enough operation that a closed door is sufficient soundproofing. For the price, that's hard to beat.
JetBlack VOLT — the outsider worth knowing
Price: ~$900 / EUR850 | Accuracy: +/-1% | Max power: 2,200 W | Max gradient: 20% | Noise: ~59 dB at 250 W
JetBlack is an Australian brand that most European and American riders haven't heard of, and that's going to change. The VOLT hits +/-1% accuracy, 2,200 W max power, and 20% gradient simulation at a price point $400 below the KICKR V6.
The build is heavy — 22 kg — and the flywheel delivers a slightly different pedal feel compared to the Wahoo and Tacx units. It's not worse, just different. Some riders prefer the heavier inertia simulation. ERG mode response is smooth. Connectivity covers ANT+ FE-C and Bluetooth FTMS. Zwift, TrainerRoad, everything works.
The main concern with JetBlack is service and parts availability outside Australia. If you're in Europe or North America, check that warranty support and replacement parts are accessible before buying. The trainer itself is excellent.
What actually matters: a framework for choosing
After all those specs, here's where it gets really interesting. The specs matter less than three practical questions:
1. How often will you ride indoors? Once a week or less: a wheel-on trainer or the cheapest direct drive (Elite Suito-T) is fine. Three or more times a week: invest in a proper direct drive. The quality-of-life difference compounds with every session.
2. How much do you care about noise? If you train while others sleep, noise is the first filter. The Tacx NEO 3 wins. The KICKR V6 is close behind. Everything else is acceptable behind a closed door but audible from the next room.
3. Is this for structured training or free-riding? Structured training rewards power accuracy above all else. Free-riding on Zwift rewards road feel and gradient simulation. The NEO 3 and KICKR V6 do both well. The Saris H4 and Elite Suito-T prioritise data accuracy over ride feel, which is the right trade-off for pure interval work.
The trainer is the enabler, not the magic
I keep coming back to this. The best indoor trainer in the world won't make you faster if you don't ride it with purpose. And a decent trainer ridden four times a week with structured sessions will produce more gains than an expensive one gathering dust.
What the pros understand — and what Dan Lorang has talked about on the podcast — is that indoor training works because it removes excuses. No weather delays. No traffic. No freewheel descents that break up the interval. You clip in, you do the work, you get off. The quality of each session is higher because the precision is higher.
If you're serious about your indoor training and want the structure and accountability to make it count, come join us in the Roadman Cycling community on Skool. We share indoor training setups, session breakdowns, and real-world results from riders who are doing this every week. No gatekeeping. Just serious cyclists who are not done yet.
Whatever trainer you pick, pick one that removes friction from your training rather than adding it. Accuracy matters. Noise matters. Everything else is preference. The work is what makes you faster.