Your winter watts are lying to you. Every spring it plays out the same way: riders crawl out of the trainer cave, hit the road for the first real ride, and watch their numbers sag 20, 30, sometimes 40 watts under what the smart trainer swore was true all winter, at the exact same heart rate and the exact same effort. What follows is a do-it-this-weekend protocol for re-zeroing your power meter, retesting FTP the modern way, and resetting your Coggan zones so indoor and outdoor finally tell the same story. It's current for 2026, and it folds in the latest firmware and tools, including Favero's February 14, 2026 battery update with its mandatory re-zero, Garmin's new Smart Calibration, and TrainerRoad's AI FTP Detection, into one clean changeover checklist.

Key takeaways

- The typical indoor-to-outdoor power gap is 3–10%. Some of it is real (cooling, drivetrain losses) and some of it is fixable (calibration, device matching).

- What your head unit calls "calibration" is almost always just a zero offset, taring the baseline, like pressing tare on a kitchen scale.

- Modern meters auto-zero and temperature-compensate, but a manual zero is still mandatory after firmware updates, transport, battery swaps, reinstalling pedals, and big temperature or altitude changes.

- A 10–20°C temperature swing can inject 1–3% error, and cold-garage-to-warm-road setups can drift 3–5% in the first 10–20 minutes.

- Retest FTP on your reference device, freshly zeroed, then rebuild your seven Coggan zones from that single number. Use PowerMatch so you never carry two conflicting sets of zones again.

This guide is for data-driven riders who already train with power, already understand FTP and zones, and just want clean, comparable numbers as the season turns. There's no "what is a watt" preamble here. Just the protocol and the reasoning behind every step.

Zero offset vs calibration: what your head unit is actually doing

Let's kill the most common confusion in power training first. "Calibrate" and "zero" are not the same thing, even though your head unit jams them together under one button. Getting this straight is the foundation for everything else, because if you don't know what the meter is actually doing when you tap "Calibrate," you can't tell when it has worked or when it has quietly failed on you.

Power is simple physics: Power = Torque (N·m) × Angular velocity (rad/s). Your strain gauges measure torque, meaning how much the crank, spider, or pedal spindle flexes under load, and the unit pairs that with cadence to compute watts. Every calibration concept you'll ever touch comes down to two questions about that torque measurement. What reading equals no torque? And how many strain counts equal one N·m?

So you actually have three distinct concepts here, and they're not interchangeable:

When your Garmin or Wahoo head unit flashes "Calibration successful" with a number like −12 or 1043, it almost always just did a zero offset. It tared the baseline. It did not check whether your scale factor is correct. And that distinction matters, because a meter can zero perfectly and still read 3% high if its slope has drifted, and a field zero will never catch that.

If you want to go a step further, static calibration is now a consumer feature. Favero's Assioma app v4.0.6, released May 26, 2025, added an Advanced "static weight test": you hang a known weight (the docs use a 20 kg example) and confirm the pedals report the correct torque, or compare left-versus-right balance even without a calibrated weight. It's the home version of the hanging-weight check pros have leaned on for years.

Decision rule: For 95% of rides, a clean zero offset is all you need and all your head unit will let you do anyway. Reach for a static weight test only after a crash, a reinstall that felt off, or a stubborn disagreement between two devices you trust.

What's new in 2026: firmware, auto-zero, and smart calibration

The biggest calibration story of 2026 isn't a new sensor. It's software. On February 14, 2026, Favero released a firmware update that more than doubled claimed battery life on its Pro RS and Pro MX power-meter pedals, from 60 hours to 160 hours, with no hardware change at all. They pulled it off purely by rewriting the power-consumption algorithm. The update takes about 3 minutes 20 seconds, and here's the part that matters for this guide: it forces a mandatory manual zero offset at the end, with the crank arm held vertical for a few seconds.

That forced re-zero is the perfect teaching moment. Even a battery firmware update shifts the meter's internal state enough that the manufacturer won't let you ride until you re-tare. Watch the scope, though. The update only applies to the newer Pro-spindle pedals (Pro MX-1/MX-2 and Pro RS-1/RS-2), not the older Assioma Duo/Uno, which run different hardware. Minimum firmware is 7.09, and the update unit shipped on 7.11. The payoff is genuinely dramatic: 15 minutes of charging now buys you roughly 15 hours of ride time.

After that update, the pedal battery-life pecking order looks like this:

The one genuinely new platform is the Garmin Rally X10, leaked in July 2025 with expected availability around August or September 2025. It adds "Smart Calibration" (Pedal IQ), a guided routine that blends temperature and context data instead of performing a single static zero, plus expanded per-pedal Force Data analytics. This is where the whole category is heading: calibration that's aware of conditions rather than a blind tare.

Favero has been quietly building toward this for years. Auto-zero by default arrived back in firmware 4.75 (2021): the Assioma pedals auto-zero after roughly 2 seconds of freewheeling, so a manual zero is only strictly required after install or after moving pedals between bikes. The latest Assioma firmware, 6.24, shipped November 6, 2025. And the Favero Pro spindle is immune to PCO (Platform Center Offset) changes, a known accuracy issue on Look and SRM pedals, so mid-ride zeroing on the Pro won't shift your baseline readings.

The 2026 takeaway: modern gear auto-zeroes and temperature-compensates more aggressively than ever, which means fewer mandatory manual zeros during normal riding. But the two events that still demand a manual zero haven't gone anywhere. Firmware updates (like the Feb 2026 Favero one) and big seasonal temperature swings (like your first warm spring ride). Auto-zero handles the easy 95%. This protocol handles the 5% that actually moves your numbers.

Why temperature is the enemy of clean spring data

If there's a single villain in the spring power story, it's temperature. Strain gauges work because metal flexes predictably under load, but how much it flexes depends on how warm that metal is. Both the electrical resistance of the strain gauges and the elastic modulus of the crank or spider shift with heat, and that drags the no-load baseline around. Which is exactly why a meter zeroed in a cold garage can read meaningfully wrong once it warms up to road temperature.

Modern meters fight this with automatic temperature compensation across roughly −20°C to 50°C, targeting ±1% accuracy across that whole band. For most rides in stable conditions, it works well enough that you never think about it. The problem is transitions. And spring is nothing but transitions.

Here's the magnitude you're dealing with. A 10–20°C swing can produce roughly 1–3% error if the meter is uncompensated or simply not re-zeroed at the new temperature. Older or cheaper designs are worse. Going from a 5°C garage to a 25°C road can drift 3–5%, and the worst of that drift lands in the first 10–20 minutes before the unit equilibrates. Which is, of course, the exact window in which most people do their warm-up and, fatefully, their FTP test efforts.

Picture a concrete scenario. You store your bike in an unheated garage at 6°C overnight. Saturday morning you carry it out, clip in immediately, and start a 20-minute FTP test on a 22°C road. For the first 15 minutes your meter is chasing a moving baseline. Its strain gauges and crank are still equilibrating, and your numbers may read several percent off before they settle. The test is contaminated before you've even hit your stride.

The fix is procedural, not technical, and it costs you five minutes:

1. Bring the bike to ambient temperature by letting it sit outside (or in whatever conditions you'll ride in) for 1–5 minutes so the meter equilibrates.
2. Back-spin the cranks a few times to wake the unit and let auto-zero engage.
3. Trigger a manual zero from your head unit or app.
4. Only then start your warm-up and effort.

Pro tip: The bigger the temperature delta between storage and riding, the longer you should let the bike sit before zeroing, and the more reason to avoid an all-out effort in the first 15 minutes. In spring specifically, treat every cold-garage start as a re-zero trigger. No exceptions.

The 5-minute spring zero-offset protocol

This is the section to bookmark. Run this checklist on your first outdoor ride of the season, and again after any of the trigger events below, and your baseline will be honest. It takes about five minutes and removes the single most common source of "why are my numbers weird" frustration.

The core protocol, in order:

1. Equilibrate. Bring the bike to the temperature you'll ride in and let it sit 1–5 minutes. The colder the storage, the longer you wait.
2. Back-spin. Spin the cranks backward a few rotations to engage auto-zero and confirm the meter is awake and connected.
3. Zero. Trigger the manual zero offset from your head unit's sensor menu or the manufacturer's app. Hold the crank arm in the position the device asks for (often vertical) and keep the bike still.
4. Sanity-check the number. Note the returned offset value. A wildly different number than usual can flag a loose pedal, low battery, or a damaged unit, so investigate before you ride.
5. Ride. Save your first easy ride as a baseline you can trust.

Re-zero whenever any of these happen. This is your trigger list for the whole season:

• Transporting the bike (car rack, travel case, flight)
• Swapping or recharging the battery
• Reinstalling pedals or moving them between bikes
• Installing a firmware update (e.g., the Feb 2026 Favero update forces this)
• A big temperature or altitude change mid-ride

Device-specific notes:

• Garmin Rally: Use the head unit's Calibrate menu (Sensors → power meter → Calibrate). On the new Rally X10, lean on Smart Calibration / Pedal IQ, which blends temperature and context rather than a single static zero.
• Favero Assioma / Pro: The pedals auto-zero after ~2 seconds of freewheeling, so day-to-day you can often skip the manual step. Do a manual zero after install, after moving pedals between bikes, and after any firmware update. And remember the Pro spindle's PCO immunity means mid-ride zeroing won't shift your baseline.
• Smart trainers: Wheel-on and mid-range trainers need a periodic spindown (a coast-down friction check after a 10–15 minute warm-up). Premium units like the Tacx Neo, Wahoo Kickr v6/Move, and Elite Justo are marketed as "factory-calibrated / no spindown," but they still self-calibrate via internal temperature sensors and auto-zero firmware, so let them warm up before a hard effort regardless.

Decision rule for "do I need to zero right now?": If any trigger event happened since your last ride, yes. If nothing changed and your gear auto-zeroes, a manual zero is optional but free. And on a spring morning with a big temperature swing, just treat it as required.

Why your outdoor power is lower than indoor (and how much is real)

Now the question every spring rider actually types into Google: why is my outdoor power lower than indoor at the same heart rate and effort? The honest answer is that some of the gap is a measurement artifact you can fix, and some of it is real physiology you can't. Learning to separate the two is what keeps you from chasing ghosts.

Indoor-versus-outdoor discrepancies of 3–10% are common and mostly normal. As a rough rule of thumb: 3–5% is expected even with two perfectly in-spec devices; 5–8% points to an uncalibrated trainer, a missed spindown, or cooling and posture differences; and anything over 10% usually signals a fault, like a wrong crank length entered, a mis-set trainer, or a genuinely failing device.

Here's the gap broken into its real causes and how much each one contributes:

Two of these are worth a closer look. The measurement-location effect is pure geometry. A smart trainer reads power at the hub or axle, after the chain, cassette, and pulleys have already eaten 2–4% in drivetrain losses, while a crank or pedal meter reads upstream of all that. So the same legs producing identical power will show higher numbers on the pedals than on the trainer. That's correct, not a malfunction.

The device error stack is the sneaky one. Say your trainer reads +2% and your pedals read −2%. Both sit comfortably within their stated tolerances, yet they'll disagree by about 4%. Neither is broken. This is exactly why you can't resolve a trainer-versus-meter dispute by deciding one is "right." You resolve it by picking one as your reference (more on that below).

The scenario to internalize: You held 250 W all winter on a Kickr Core (±2%). Your first outdoor ride on Favero pedals (±1%) shows 240 W at the same heart rate. That's a 4% gap, entirely explainable by drivetrain losses plus the error stack, with no fault anywhere. Don't "fix" it by training harder. Fix it by matching your devices.

Gear check: accuracy and 2026 prices for popular meters and trainers

Before you rebuild your zones, it's worth knowing exactly what your hardware claims and what it would cost to upgrade. The error-stack math above only makes sense once you know each device's tolerance, and spring, with its sales and its new-season motivation, is when a lot of riders start eyeing a meter or trainer change. Pricing in 2026 is broadly flat versus 2025, within about ±$50 across the board.

Power meters, claimed accuracy and 2026 USD price:

Smart trainers, claimed accuracy:

The implication you have to carry into zone-setting: a ±1% meter paired with a ±2% trainer can legitimately sit 3–4% apart with neither device at fault. If your pedals are tighter than your trainer (the common case), that's your cue to make the pedals your reference device and slave the trainer to them. Which is exactly what PowerMatch does.

Should you upgrade this spring? Upgrade if your current device is wider than ±2% and you race on power, or if you can't get two devices within about 5% even after clean zeros, or if you're moving to a second bike and want one consistent meter across both. Otherwise, calibrate what you own. A clean zero on a ±1.5% meter beats a sloppy setup on a ±1% meter every time.

Retesting FTP for the new season: ramp vs 20-minute vs AI FTP

With a trustworthy, freshly zeroed meter, it's finally time to retest FTP, because the number you carried out of winter is almost certainly stale. A winter of indoor base work, or a few weeks off the bike, shifts your threshold enough that building zones on an old FTP will mis-target every workout for months. The right time to retest is after any detraining has stabilized and before you start building structured zones for the season.

You've got three modern routes, and they're not equivalent.

1. The 20-minute test. Ride an all-out, evenly paced 20 minutes after a thorough warm-up, then take FTP ≈ 95% of your 20-minute average power. A 300 W average gives you a 285 W FTP. It's demanding and it punishes bad pacing, but it's well-understood and easy to repeat identically each season.

2. The ramp test. Ride steadily increasing power until you can't hold on; your FTP ≈ 75% of your one-minute peak (MAP). A 400 W MAP implies a 300 W FTP. It's shorter and less pacing-dependent, but the fixed 75% multiplier fits some riders poorly. Riders with big anaerobic capacity tend to come out with their FTP overestimated.

3. AI FTP detection (the 2026 default for a lot of riders). TrainerRoad's AI FTP Detection drops the fixed multiplier entirely. It reads your raw power and heart rate, picks the next best wattage, then back-calculates FTP so that a Threshold workout lands at about Level 3. It targets a ~2% hard-zone failure rate, roughly 33% better than the old ramp-surrogate approach. The internal logic is readable too: a Threshold level below 3 means your FTP gets cut; 3–4 leaves it unchanged; 4–6 nudges it up; 7+ triggers a big increase. There's also AI FTP Prediction, which estimates a future FTP with no all-out effort. Ideal when you're returning from time off and a max test would just flatten you.

The other platforms have drifted the same way. Zwift revamped its automatic FTP detection in 2025 and now pulls in outdoor rides for training metrics, proposing a new FTP whenever your recent ~20–60 minute power exceeds your current value (no published multiplier). Garmin's auto-FTP uses proprietary Firstbeat-style performance modeling and simply prompts "New FTP detected," again with no public formula.

The non-negotiable rule: test on your reference device, freshly zeroed, using the same protocol and same conditions each season. An FTP set on an indoor trainer and an FTP set on outdoor pedals are not the same currency, which is the whole reason zone-setting and PowerMatch come next.

Reset your power zones from the new FTP

A fresh FTP is useless until it propagates into your zones. The seven-zone Coggan model expresses each zone as a percentage of FTP, so the moment your FTP changes, every zone boundary shifts with it. Rebuilding them is mechanical, but you actually have to do it, because a single stale number silently mis-targets every interval you ride.

Here are the canonical Coggan percentages with a worked example for a 250 W FTP:

Notice that Z7 (Neuromuscular) isn't expressed as a percentage of FTP. It covers short, maximal sprint efforts that FTP simply doesn't describe, so don't try to back a watt number into it from your threshold.

The worked scenario: Your AI FTP Detection bumps you from 240 W to 250 W after a strong spring block. Your Z2 endurance ceiling moves from 180 W to 188 W; your Z4 threshold band shifts from 218–252 W to 228–263 W. Ride your old zones and every "endurance" ride creeps slightly into tempo, and every threshold interval undershoots. Small errors that compound across a whole training block. Updating the single FTP value in your app cascades all seven zones automatically.

That raises the spring rider's recurring dilemma: should you keep separate indoor and outdoor FTP values and zones? You can. Plenty of riders run a lower indoor FTP to account for cooling and a higher outdoor FTP measured on pedals. But maintaining two sets of zones is a maintenance headache and a constant source of "which number am I training to today?" confusion. The cleaner answer for most riders is one FTP, one set of zones, and a tool that makes both environments report the same watts, which is exactly the job of PowerMatch.

Checklist, rebuilding zones:

• [ ] Enter the new FTP in your training platform (zones recalculate automatically)
• [ ] Confirm the percentages match the Coggan table above
• [ ] Decide: one FTP for both environments (recommended) or separate values
• [ ] If separate, label them clearly so you never train to the wrong set
• [ ] Re-check zones after every retest, not just in spring

Make indoor and outdoor agree: PowerMatch and one set of numbers

This is the step that ties the whole protocol together, and the reason you can finally stop running two sets of zones. The core idea is simple. Instead of letting your trainer and your meter report two different numbers, you pick one reference device and make everything else defer to it.

The natural reference is the meter you race and ride outdoors with, usually your pedals or crank meter, because it travels with you everywhere and it's what your competitive results are built on. Once you've chosen it, you configure your indoor app to read watts from that device and slave the trainer's resistance to match:

• In TrainerRoad: enable PowerMatch. The app reads watts from your on-bike power meter and adjusts the trainer's resistance to hit those targets, so indoor and outdoor data come from the same device. The trainer becomes a dumb actuator; your meter is the single source of truth.
• In Zwift: set your power meter as the Power Source and the trainer as the Controllable. Same effect: the meter reports, the trainer obeys.

There's one important exception. Stages Smart Bike users should disable PowerMatch, because the bike rebroadcasts its own integrated meter as the power source, and layering PowerMatch on top would have it trying to match a device to itself. If your "trainer" is actually a smart bike with a built-in meter, let it be the reference and skip the matching.

Why this matters for your zones: with PowerMatch active, the FTP you tested on your pedals and the zones you built from it are valid both indoors and outdoors, because both environments now report your pedals' watts. The drivetrain-loss and error-stack gaps from earlier don't vanish from physics, but they vanish from your training decisions. Every workout, indoor or out, is measured against the same ruler.

The closing scenario: You set FTP to 250 W on your Favero pedals, built your Coggan zones from it, and enabled PowerMatch in TrainerRoad. Tuesday's indoor threshold intervals and Saturday's outdoor tempo ride now reference the identical 250 W. No mental math, no "add 5% for indoors," no second set of zones. Just one trustworthy number carrying you through the whole outdoor season. That's the payoff for one careful weekend of calibration.

Final decision rule: One reference device, one FTP, one set of Coggan zones, PowerMatch on (unless you ride a smart bike). Re-zero on every trigger event, and retest FTP each season on that same reference. Do this and your data stops lying to you.

Frequently asked questions

Q: How often should I calibrate (zero-offset) my power meter, before every ride? A: Zeroing before every ride is best practice and costs under a minute. At minimum, you must re-zero after transport, battery swaps, reinstalling pedals or cranks, firmware updates, and big temperature or altitude changes. Modern meters auto-zero during freewheeling, but a manual zero on a cold spring morning is always worth it.

Q: What is the difference between a zero offset and a full calibration? A: A zero offset tares the no-load baseline. It tells the meter what "no torque" reads, like pressing tare on a kitchen scale. A full (static) calibration verifies the scale factor by hanging a known weight to confirm the meter reports the correct torque. What your head unit labels "Calibrate" is almost always just a zero offset.

Q: Do I need to re-zero my power meter when I move from indoor trainer to outdoor riding? A: Yes, especially in spring. The big temperature change between a climate-controlled trainer room and outdoor air shifts the strain-gauge baseline, and transporting or reinstalling the bike are themselves re-zero triggers. Let the bike equilibrate 1–5 minutes at outdoor temperature, back-spin the cranks, then zero before your first effort.

Q: Why is my outdoor power lower than my indoor power at the same heart rate and effort? A: A 3–10% gap is normal. Trainers measure at the hub after 2–4% drivetrain losses, while pedals measure upstream, so pedals read higher. Add the device error stack (a +2% trainer vs a −2% meter sits ~4% apart in spec) plus cooling and position differences. Under 5% is expected; over 10% suggests a fault.

Q: Should I use separate FTP values and power zones for indoor and outdoor? A: You can, but it's usually unnecessary and it creates confusion. The cleaner approach is one FTP and one set of zones, with PowerMatch making both environments report the same device's watts. Reserve separate values only if you have a specific, consistent reason, and label them clearly so you never train to the wrong set.

Q: How soon after a winter of indoor training should I retest my FTP outside? A: Retest once any detraining or fitness change has stabilized and before you start building structured zones for the season, typically in early spring as you transition outdoors. Test on your reference device, freshly zeroed, using the same protocol you'll repeat each season so the numbers stay comparable year over year.

Q: Which should I trust when my smart trainer and power meter disagree? A: Neither is automatically "right." A ±1% meter and a ±2% trainer can legitimately sit 3–4% apart. Instead of picking a winner, choose one reference device (usually the outdoor meter you race with) and use PowerMatch so the trainer slaves to it. Consistency matters more than which absolute number is "true."

Q: Does cold or hot weather affect power-meter accuracy, and should I recalibrate in spring? A: Yes. Strain-gauge resistance and crank elasticity shift with temperature, drifting the baseline. A 10–20°C swing can add 1–3% error, and cold-garage-to-warm-road starts can drift 3–5% in the first 10–20 minutes. Always re-zero in spring after letting the bike reach ambient temperature.

Q: What is the best FTP test in 2026, ramp, 20-minute, or AI FTP detection? A: It depends on you. The 20-minute test (FTP = 95% of average) is the most repeatable benchmark; the ramp test (FTP = 75% of MAP) is quicker but overestimates high-anaerobic riders. TrainerRoad's AI FTP Detection back-calculates from real workout data to a ~2% failure rate and is the best low-effort option if you train in-app regularly.

Q: What are the Coggan power-zone percentages and how do I set them from my FTP? A: As a percentage of FTP: Z1 <55%, Z2 56–75%, Z3 76–90%, Z4 91–105%, Z5 106–120%, Z6 121–150%, and Z7 (neuromuscular) is short maximal efforts not tied to FTP. Enter your FTP in your training app and the zones recalculate automatically, then re-check after every retest.

Q: Do auto-zero and temperature compensation mean I can skip manual calibration? A: For routine rides in stable conditions, largely yes. Auto-zero handles the easy 95%. But a manual zero is still mandatory after firmware updates (the Feb 2026 Favero update forces one), transport, battery swaps, reinstalls, and big seasonal temperature swings. Auto-zero is a convenience, not a replacement for the trigger checklist.