The Moveee team
Free coaching · a real route for every ride
Your smart trainer says 240 W. Your power-meter pedals say 249 W. Which one is lying? Very possibly neither — and understanding why is the key to trusting your numbers at all. Power accuracy has two separate parts people constantly blur: how precise each device is, and where on the bike it's measuring. Two honest instruments reading different points of the same drivetrain should disagree. Here's what the accuracy claims really mean, why the disagreement is legitimate, and how to sanity-check your own setup in twenty minutes.
- Claimed accuracy (approximate): wheel-on trainers ±3–5%, mid direct drives ±2%, high-end direct drives ±1% — the direct-drive price premium partly buys measurement quality.
- Trainer vs pedals disagreeing by ~2–3% is physics, not fraud: the drivetrain loses that much between your feet and the trainer's hub, so the trainer legitimately reads lower.
- Consistency beats absolute truth: pick one power source per context, keep your zones tied to it, and stop comparing across sources mid-ride.
What the accuracy claims actually promise
A "±2%" claim means the device should report within 2% of true power under reasonable conditions — at 250 W, anywhere from 245 to 255 W. Here's how the classes stack up; all figures are manufacturers' claims and approximate, and a well-treated unit typically does better than its worst case:
| Class | Typical examples | Claimed | Error band at 250 W |
|---|---|---|---|
| Wheel-on smart trainer | Kickr Snap, Tacx Flow, Saris M2 | ±3–5% |
|
| Mid-range direct drive | Kickr Core, JetBlack Volt, Tacx Flux | ±2–2.5% |
|
| High-end direct drive | Kickr V6, Tacx Neo, Elite Justo | ±1% |
|
| Pedal / crank power meter | Assioma, Rally, Quarq, Stages | ±1–2% |
|
Wheel-on smart trainer: Power is estimated from roller speed and resistance, through a tyre interface that changes with pressure, tension and temperature. Needs regular spindowns to stay near its claim.
Mid-range direct drive: Measures at the cassette with no tyre in the way. Most units hold their claim well once warmed up; many auto-calibrate.
High-end direct drive: Effectively lab-grade for training purposes. The Neo family doesn't even need calibration — it measures rather than estimates.
Pedal / crank power meter: Measures at the pedal or crank — upstream of the drivetrain, which matters for the comparison below.
Bar width shows the relative size of each class's worst-case error band — wider means the reading can wander further from the truth.
Why the trainer and your power meter legitimately disagree
Follow one pedal stroke down the bike. Your legs put power in at the pedals; the chain carries it to the cassette; the trainer absorbs it at the hub. At every link and tooth along the way, a little of it becomes heat:
1. Pedals / crank
What your legs actually produce. Pedal and crank meters read here, before anything is lost.
2. Drivetrain
The chain, cassette and pulleys eat roughly 2–3% as heat — more with a dirty or worn chain, slightly more in cross-chained gears.
3. Trainer hub
What arrives at the cassette. Direct-drive trainers read here — honestly reporting a genuinely smaller number.
So when pedals say 250 W and the trainer says 243 W, both are right about the point they're measuring. The ~2–3% gap is your drivetrain's efficiency bill (approximate; a filthy chain charges more). This also means the direction of the gap is predictable: the trainer should read lower than a pedal or crank meter. If yours reads consistently higher, or the gap is far beyond the two accuracy claims combined, then something genuinely is off — usually calibration.
Calibration: who needs it and when
Wheel-on trainers need it, regularly. Their power is inferred through a tyre pressed against a roller, and that interface changes with tyre pressure, clamping tension and temperature. The fix is the spindown: ride ~10 minutes to warm the tyre and roller first, then run the calibration — a cold spindown bakes the wrong friction number in for the whole session. Re-do it when you change tyre pressure or re-tension the roller.
Direct drives mostly handle themselves. Mid-range units are factory-calibrated and many auto-calibrate as you ride; an occasional manual spindown after firmware updates doesn't hurt. High-end units like the Tacx Neo measure directly and have no calibration routine at all. The one habit that helps every class: give the trainer ten minutes of riding before you trust the numbers, because electronics and strain gauges drift as they warm — the same reason your first interval of the day can look strangely cheap or expensive.
Consistency beats absolute truth
Here's the mindset shift that ends most power anxiety: for training, you don't need the cosmically true number — you need the same ruler every time. A trainer that reads 3% low but does so every single ride is a perfectly good training instrument: your FTP, your zones and your progress are all measured in its units, and they move together. Problems only start when you mix rulers — testing FTP on the pedals, then doing workouts off the trainer, then wondering why threshold feels wrong.
The practical rule: pick one power source per context and keep your zones per source. If you own both, either let the pedals be the source indoors and out (one ruler everywhere — the cleanest option), or keep an indoor FTP and an outdoor FTP and never compare a Tuesday trainer ride against a Sunday road ride watt-for-watt. This, incidentally, is one honest reason indoor FTP often tests lower — sometimes it's physiology, and sometimes it's simply a different ruler.
Sanity-check your own setup
If you have two power sources, twenty minutes settles whether yours agree within reason:
Warm up for 10–15 minutes first
Strain gauges and trainer electronics drift as they warm. Compare warm gear with warm gear; if your trainer is wheel-on, do the spindown after the warm-up, not before.
Ride a steady 5-minute interval at endurance pace
Record the same interval on both sources (trainer in the app, power meter on a head unit or a second recording). Steady state is the fair test — smoothing and transmission timing make sprints meaningless for comparison.
Compare the averages, not the wiggles
Second-by-second traces will never overlay perfectly; different devices smooth and timestamp differently. Averages over the full interval are the honest comparison.
Expect the trainer to read ~2–3% lower
That's the drivetrain doing its thing. Pedals at 250 W and a trainer at 243 W agree. Worry only when the gap exceeds the two claims plus drivetrain loss — say beyond ~5–6% — or when it changes ride to ride.
Repeat once, another day
One comparison is an anecdote. Two consistent ones tell you your setup's true offset — which you can then simply live with, because it's stable.
One convenience worth knowing: because Moveee Indoor runs in the browser and pairs devices over Web Bluetooth, you can pair your trainer for control and a separate power meter as the power source in the same session — the ERG resistance comes from the trainer while your watts come from the meter you've anchored your zones to. Whichever source you choose, choose it once, calibrate it warm, and then get on with the actual training. The watts don't need to be perfect; they need to be yours.
Keep reading
How to build a cycling training plan around a full-time job
You don't need 12-hour weeks to get properly fast — you need the right 6–8 hours, placed where your life can actually hold them. Here's how to anchor a working week around two quality sessions and one long ride, what to cut first when work explodes, and why a plan that adapts beats a plan that's perfect.
How many TSS per week do you actually need?
TSS puts one number on how much training you've done — but how much is enough, and when does more stop helping? Realistic weekly ranges by rider level, the ramp-rate rule that keeps you improving without breaking, and why chasing the number corrupts the training it's meant to measure.