You own a power meter. You know your FTP matters. You have probably already tested it at least once — maybe on Zwift, maybe following a plan your mate sent you, maybe just going full gas for twenty minutes on a Tuesday evening and multiplying by 0.95 because someone on a forum said to.
Here's the thing nobody tells you: the protocol matters less than how you execute it. A perfectly paced 20-minute test beats a badly paced one by 15-20 watts. A ramp test done with collapsing cadence in the final minutes gives you a number that belongs to someone else's legs. And every protocol has a failure mode — a specific scenario where it hands you a number that will make your training either too hard or too easy for the next six weeks.
This is the practical guide. Not which test is "best" — that argument is circular and mostly pointless — but how to actually do each one properly, what preparation looks like in the 48 hours beforehand, and how to tell when the number you got is lying to you.
Before You Test: The 48-Hour Protocol
Let me break this down, because preparation explains more variance between test results than protocol choice.
Two Days Out
Reduce volume by 40-50%. If your normal Tuesday is a 90-minute endurance ride, do 50 minutes easy. No intervals, no group rides, no "just one more sprint for the town sign." The goal is glycogen replenishment without complete detraining. You are banking energy, not building fitness.
Eat normally but increase carbohydrate intake slightly — an extra 50-75g across the day from rice, pasta, bread, whatever sits well. This is not carb-loading for a race. It is making sure your fuel stores are topped off rather than three-quarters full from a hard training week.
Sleep matters more than anything else on this list. If you slept five hours the night before a test, the result will be 3-5% lower than your actual capacity. That is 8-15 watts for most riders. There is no supplement, no warm-up protocol, and no motivational playlist that compensates for poor sleep.
Test Day Morning
Eat your normal pre-ride meal 2-3 hours before the test. Whatever you eat before a hard Sunday ride — that. Do not experiment. Do not skip breakfast because you read something about fasted training. This is a maximal effort; your body needs fuel.
Caffeine: if you normally have coffee, have coffee. Same amount, same timing. If you do not normally use caffeine, test day is not the day to start. Caffeine at 3-5mg per kilogram of bodyweight does improve sustained power output by roughly 2-3% in the research (Ganio et al., 2009), but the effect is most reliable in habitual users who maintain their normal intake. Doubling your espresso because it is test day introduces jitters, a higher resting heart rate, and an unreliable baseline for future tests.
Hydrate in the hours before. 500ml of water with the pre-ride meal, another 250-500ml in the hour before you start warming up. Dehydration of even 2% bodyweight reduces sustained power output.
Environment Standardisation
This is the part most people skip. Then they wonder why their FTP went "down" when they tested in August versus March.
Indoor testing: Same room. Same fan setup — and fan positioning really matters, because core temperature affects performance more than most riders realise. Point a strong fan directly at your torso and face. If you used two fans last time, use two fans this time. Same bike, same tyre pressure on the trainer (if using a wheel-on), same firmware on the smart trainer. Wahoo KICKR units can read differently after a firmware update — perform a spindown calibration before every test.
Outdoor testing: Same road or climb. Same time of day. Wind direction and temperature should be comparable, which realistically means you cannot perfectly standardise outdoor tests. That is why most coaches — and certainly anyone tracking trends seriously — test indoors.
Record the ambient temperature. A 10°C increase in room temperature can cost 2-4% of sustained power output through thermoregulatory strain.
The Standardised Warm-Up
Use this warm-up for every protocol. Every time. No variation. It takes 20 minutes and it is not optional.
Warm-Up Script (20 Minutes)
Minutes 0-5: Easy spinning, 50-55% FTP. Cadence 85-95 rpm. This gets blood flowing to working muscles and starts raising core temperature. It should feel like nothing.
Minutes 5-10: Progressive build from 55% to 70% FTP. Increase power every 60 seconds. Cadence stays at 85-95. By minute 10, the effort should feel "comfortably firm" — you could hold a conversation, but you would rather not.
Minutes 10-12: First opener. 30 seconds at 95-100% FTP, then 60 seconds easy at 50% FTP. This is not an effort — it is a primer. You are telling your neuromuscular system that hard work is coming.
Minutes 12-14: Second opener. 30 seconds at 100-105% FTP, then 60 seconds easy. Slightly harder than the first.
Minutes 14-16: Third opener. 30 seconds at 105-110% FTP, then 60 seconds easy. This one should feel sharp. Your heart rate will climb. Good.
Minutes 16-20: Easy spinning, 50% FTP. Let your heart rate settle. Breathe. Mentally rehearse the first two minutes of the test. This recovery window is critical — skip it and you carry residual fatigue into the test.
At minute 20, you start. Not minute 18 because you feel ready. Not minute 23 because you got distracted checking your phone. Minute 20.
Protocol 1: The Ramp Test
What It Measures
Your maximal aerobic power (MAP) — the highest sustainable one-minute power before voluntary exhaustion. FTP is then estimated at 75% of that number. The protocol was popularised by Zwift, TrainerRoad, and Wahoo, and it is now the most commonly performed FTP test in the world.
Exact Execution
Starting power: 100 watts (adjust to 75W for lighter riders or 125W for riders who know their FTP exceeds 300W).
Step increment: 20 watts per minute for ERG-mode trainers, or 25 watts per minute on platforms that use that step size. Check your platform's protocol — Zwift uses steps based on rider weight, TrainerRoad uses a fixed percentage increase. Know what your software does before test day.
Cadence target: 85-95 rpm from the first step to the last. This is the single most important execution detail in the ramp test. As the wattage climbs past your comfortable range, your instinct will be to drop cadence — push bigger gear, grind harder. Do not. The moment you drop below 80 rpm, you are recruiting muscle fibres in a way that inflates MAP beyond what your aerobic system alone can produce. Consider the test over when you cannot maintain 80 rpm, regardless of whether you could force out another 30 seconds of grinding.
Breathing: Let it go. Do not try to control your breathing in the final minutes. Deep, ragged breaths are the correct response to maximal aerobic stress. Trying to stay composed constrains oxygen delivery.
When to stop: When you cannot maintain the target wattage for the current step despite full effort. On ERG-mode trainers, this happens automatically — the resistance climbs beyond what you can produce at any cadence. On non-ERG setups, stop when cadence drops below 80 rpm for more than 10 seconds.
The maths: Your platform typically calculates MAP as the best one-minute power in the final stage. FTP = MAP × 0.75. If your best final minute was 360 watts, FTP = 270.
Where It Lies
Here's where it gets really interesting. The 75% conversion factor assumes your FTP-to-MAP ratio matches the population average. It often does not.
Riders with strong anaerobic capacity (former footballers, rugby players, track sprinters, or anyone who can produce disproportionate power for 30-90 seconds) push deeper into the ramp on anaerobic reserves. Their MAP reads high. The 75% calculation sets an FTP their aerobic engine cannot sustain. If your sweet spot intervals feel like threshold work after a ramp test, this is why.
Diesel riders — big aerobic engines, modest sprint — hit failure earlier because they lack the anaerobic punch to push through the high-wattage steps. The 75% calculation understates their true sustainable power. Their training zones feel too easy.
The fix: After a ramp test, do a 30-minute ride at your calculated FTP within the next week. If you can hold it comfortably for 30 minutes, the number is probably close. If you blow up at 15-20 minutes, reduce by 3-5%. If 30 minutes feels like endurance pace, increase by 3-5%.
Protocol 2: The 20-Minute Test
What It Measures
Your best sustained power over 20 minutes, from which FTP is estimated at 95% of the average. This is the Coggan and Allen protocol from Training and Racing with a Power Meter, and it remains the most validated field test against actual 60-minute efforts. If you can only master one test, master this one.
Exact Execution
Pre-test blow-out (important): Before the warm-up, or as part of an extended warm-up, do a 5-minute hard effort at approximately 105-110% of your expected FTP. Wait — that sounds counterintuitive. The purpose is to burn off the upper layer of your anaerobic capacity before the 20-minute effort. Without this blow-out, your anaerobic system contributes more to the first few minutes than the 95% factor accounts for, inflating your result. Some coaches skip this; Coggan includes it. If you skip it, your number will read slightly higher but may overestimate your true threshold.
After the blow-out, do 10 minutes of easy spinning, then complete the standardised warm-up above (starting from the openers at minute 10). This means your total warm-up with blow-out takes about 35 minutes. Yes, it is long. No, you should not skip it.
The first 2 minutes: Target 3-5 watts below where you expect to average. This is the hardest instruction to follow. Adrenaline, test-day anxiety, and the desire to "get ahead" will push you above your target. Resist. The physiological cost of starting too hard is non-linear — every watt above threshold in the early minutes costs more than one watt in the closing minutes. A rider who starts 15W too high for five minutes typically loses 20-25W over the final ten.
Minutes 2-15: Settle into a rhythm. Your target power should feel "hard but repeatable" — the kind of effort where you would not choose to talk, but you could force out a sentence. Monitor your power in 30-second averages, not second-by-second. Second-by-second numbers cause reactive adjustments that waste energy. Heart rate should climb gradually and plateau somewhere in the 90-95% of max range by minute 10.
Minutes 15-18: This is the valley. The effort is properly painful, and the finish is close but not close enough to sprint. Hold your power. Do not look at the clock every five seconds. Pick a cadence and hold it. If you have been pacing well, your power will naturally want to drift down 5-8 watts here. Let it drift by 3-4 at most. This is where most riders lose the test.
Minutes 18-20: Everything left. Increase cadence slightly — 3-5 rpm above your cruising cadence — and let power rise. You are not sprinting. You are lifting from a sustainable pace to a maximally sustainable pace. The last 30 seconds can be an all-out push.
The maths: Average power for 20 minutes × 0.95 = FTP. If you averaged 285W, FTP = 271.
Where It Lies
Pacing errors compound. A 20-minute test with a 300W first half and a 260W second half averages 280W — but it is not the same physiological test as an even 280W effort. The uneven version overestimates because the 95% factor was calibrated against evenly paced efforts.
The 5% factor is a population average. Some riders sustain 96-97% of their 20-minute power for a full hour (diesel types). Others sustain only 92-93% (strong anaerobic contribution over 20 minutes). If your threshold intervals consistently feel too easy or too hard after a 20-minute test, your personal correction factor is not 0.95.
Protocol 3: The 2×8-Minute Test
What It Measures
Your best sustained power across two 8-minute maximal efforts, with recovery between. FTP is estimated at 90% of the higher of the two efforts (or the mean, depending on the version). This protocol was popularised by Carmichael Training Systems (CTS) for riders who find the 20-minute test mentally overwhelming.
Exact Execution
Warm-up: Use the standard 20-minute warm-up above. No need for the 5-minute anaerobic blow-out — the dual-effort structure handles that problem differently.
Effort 1 (8 minutes): Start at what feels like your 20-minute pace — slightly conservative. This is not the effort where you go for broke. The first effort serves two purposes: it produces a data point, and it burns off your anaerobic top end for the second effort. Cadence at 90-95 rpm. Build over the 8 minutes — first minute at 95% of target, last minute at 105% of target.
Recovery (10 minutes): Easy spinning at 40-50% FTP. Drink. Breathe. Do not stop pedalling. Do not analyse your numbers from the first effort. The recovery interval must be standardised — same duration, same intensity, every time you test. If you ride harder during recovery because you feel good, you compromise the second effort and lose repeatability.
Effort 2 (8 minutes): Now you go for it. Your anaerobic reserves are partially depleted from effort 1, so this second effort is a cleaner read of sustained aerobic power. Target the same average as effort 1 or slightly higher. A good test produces two efforts within 5 watts of each other. If effort 1 is 15+ watts higher than effort 2, you started too hard. If effort 2 is 15+ watts higher, you were too conservative on effort 1.
The maths: Take the higher of the two average powers. Multiply by 0.90. If effort 1 was 290W and effort 2 was 295W, FTP = 295 × 0.90 = 266W.
Where It Lies
Eight minutes is a short enough window that anaerobic capacity still contributes meaningfully, even on the second effort. The 90% conversion factor was calibrated on a narrower population than the 20-minute protocol, and it tends to overestimate for strong time-triallists who produce disproportionate power over 8 minutes relative to their 60-minute capacity.
The recovery interval introduces a second variable. How hard you rode during those 10 minutes, whether you stayed on the bike, whether you rehydrated — all of these affect effort 2. One test done with disciplined 50% FTP recovery and another done with 65% FTP recovery are not the same test, even if the hard efforts look identical.
Protocol 4: The Hour of Power
What It Measures
Your FTP. Directly. No conversion factor. You ride as hard as you can sustain for 60 minutes and your average power is your functional threshold power by definition.
Exact Execution
Warm-up: Standard 20-minute warm-up. No blow-out needed — the effort is long enough that anaerobic contribution is negligible.
Pacing strategy: Start at 95% of your expected FTP for the first 10 minutes. Yes, deliberately conservative. The margin between "slightly too hard" and "blew up at 40 minutes" is thinner than in any other protocol. If you feel comfortable at minute 10, increase by 2-3%. Hold. Reassess at minute 30. If you still feel controlled — not comfortable, but controlled — hold. If you feel like you are fading, do not increase. The final 10 minutes is where most people lose watts, so preserving them through conservative early pacing pays compound interest.
Cadence: Self-selected, but consistent. Do not shift cadence mid-test unless terrain forces it. Every cadence change costs a tiny amount of energy through neuromuscular adjustment.
The maths: Average power for 60 minutes = FTP. No multiplication. No estimation.
Where It Lies
Pacing error. That is essentially the only failure mode, but it is a catastrophic one. Even professional time triallists find it difficult to perfectly pace a 60-minute maximal effort. Most amateur riders who attempt this produce a lower average than a well-paced 20-minute test would have given them, because they either start too hard and fade badly or start too conservatively and leave 10-15 watts on the table.
The fatigue cost is also significant. This is a race-day level effort that requires 48-72 hours of recovery and disrupts your training week.
For most riders, this test is academically interesting and practically unnecessary. Use it once for calibration if you want to know your personal correction factor for the 20-minute test. Do not use it for regular retesting.
Cost-Benefit Analysis: All Four Protocols
| | Ramp Test | 20-Minute Test | 2×8-Minute Test | 60-Minute Test | |---|---|---|---|---| | Duration (including warm-up) | 35-45 min | 55-65 min | 50-55 min | 80-90 min | | Pacing skill required | None | High | Moderate | Very high | | Conversion factor | MAP × 0.75 | Avg power × 0.95 | Higher effort × 0.90 | None (direct) | | Best for | Beginners, time-pressed riders | Experienced riders with pacing discipline | Riders who blow up in 20-min tests | One-off calibration | | Biggest failure mode | Overestimates anaerobically strong riders | Pacing errors compound non-linearly | 90% factor less validated | Almost impossible to pace correctly | | Recovery needed | 24 hours | 24-36 hours | 24-36 hours | 48-72 hours | | Repeatability | High (no pacing variable) | Moderate (pacing must be consistent) | Moderate (recovery interval adds variability) | Low (pacing errors dominate) | | Indoor suitability | Excellent (ERG mode ideal) | Good | Good | Mentally brutal indoors |
Interpreting Your Result
You have a number. Now what.
The Validation Ride
Within 3-5 days of testing, do a sustained effort at your new FTP — 20-30 minutes on a steady climb or flat road. This is not a second test. It is a sanity check. You are asking one question: can I hold this power for a meaningful duration without falling apart?
If you hold 100% of your tested FTP for 25 minutes and feel like you could have done another 10, your number is conservative. Consider adding 2-3%.
If you blow up at 15 minutes, the test overestimated. Drop by 3-5% and revalidate.
If you hold it for 20-25 minutes at an RPE of 8-9 out of 10, the number is probably right.
Cross-Validation Against Real Rides
Here's the thing nobody tells you about single-protocol testing: every protocol has a systematic bias, and the only way to catch it is to compare your test result against real-world sustained efforts.
Look at your best 40-60 minute power from the last month of riding — a hard group ride, a local time trial, a sustained climb. Your FTP should sit within 3-5% of that best effort. If your tested FTP is 15 watts above your best real-world sustained effort, the test overestimated. If it is 15 watts below, either the test underestimated or you have not had an opportunity to produce that effort in a real ride.
Andrew Coggan himself has said that the best FTP estimate comes from triangulating multiple data sources — test results, race data, training file analysis — rather than relying on any single protocol.
When Your Zones Feel Wrong
This is the most practical signal that your FTP is off. If you consistently cannot complete prescribed sweet spot intervals (88-93% FTP) for the target duration, your FTP is set too high. If your threshold intervals (95-105% FTP) feel like tempo work, your FTP is set too low.
Trust the feel of the training over the number from the test. The test is a starting point. The training response is the feedback loop.
Indoor Versus Outdoor: The 5-8% Problem
Most riders produce less power indoors than outdoors. The magnitude varies — 5-8% is the commonly cited range, supported by research from Mieras et al. (2014) — but the direction is consistent.
The reasons stack up. Indoor riding lacks inertial load variations — outdoors, micro-changes in gradient and wind create constant small shifts in effort that keep muscle fibres cycling through recruitment patterns. On a trainer, you are grinding the same pattern over and over. Core temperature rises faster indoors, even with fans, because there is no 30-40 km/h airflow across your body. And there is a psychological component — staring at a wall or a screen for 20 minutes of maximal effort is harder to motivate than riding a road you know.
The practical implication: set separate indoor and outdoor FTP values. Do not inflate your indoor result to match your outdoor number. Do not deflate your outdoor result to match your indoor one. If you train indoors, test indoors and use that number for indoor zones. If you race outdoors, validate against outdoor efforts and set outdoor zones accordingly.
Mixing indoor test results with outdoor training zones — or vice versa — is one of the most common reasons riders feel like their zones are wrong when the zones are actually fine. The zones are correct for the context they were tested in. Apply them in the wrong context and they stop working.
Recovery Between Tests
Testing is a maximal effort. Treat it like one.
After a ramp test: 24 hours of easy spinning or rest. You can resume structured training the following day.
After a 20-minute or 8-minute test: 24-36 hours of easy spinning. The sustained threshold effort creates more metabolic stress than the ramp test's brief peak. A validation ride 3-5 days later is fine; a hard interval session the next morning is not.
After a 60-minute test: 48-72 hours of recovery. This is a race-level effort. Plan your test week accordingly — test on Saturday, easy spin Sunday, back to structured training Tuesday.
Do not test during a build block. Test at the end of a recovery week, when glycogen stores are full, fatigue is low, and the result reflects your actual capacity rather than the accumulated cost of the last three weeks of training.
When to Retest (And When Not To)
Every 6-8 weeks. End of a training block. Start of a new phase. That cadence gives your body enough time to produce measurable adaptation and gives you enough separation between data points to distinguish real change from noise.
Testing every two or three weeks is worse than not testing at all. The physiological change over a fortnight is smaller than the day-to-day variation in sleep, nutrition, stress, and glycogen status. You are measuring weather, not climate. And each test adds a maximal effort to your training calendar that produces fatigue without producing information.
If you suspect your FTP has shifted mid-block — intervals feel noticeably easier, heart rate at threshold power has dropped by 5+ beats, RPE at current zones has changed — adjust zones by 2-3% rather than scheduling a full retest. You can validate with the FTP Zone Calculator and recalibrate your sessions without disrupting the training week.
The psychological trap is real. Testing becomes a goal in itself. You start chasing a number instead of training the system that produces the number. The rider who tests every three weeks and celebrates a 5-watt increase is often just measuring recovery status, not adaptation. The rider who tests every eight weeks and sees a 12-watt increase over three blocks is tracking genuine fitness development.
The Protocol Decision Tree
Still not sure which test to use? Let me break this down.
Are you new to structured training or power meters? Start with the ramp test. No pacing required, short duration, gives you a starting number to build from. Accept that it might be 5-10 watts off in either direction. That is fine for setting initial zones.
Can you hold a steady effort for 20 minutes without pacing errors? Use the 20-minute test. It is the most validated, the most widely used, and the one most coaching platforms — including TrainingPeaks — standardise on.
Do you consistently blow up in the first half of 20-minute efforts? The 2×8-minute test removes that failure mode. Two shorter efforts with recovery between. The 90% conversion is less validated, but a well-executed 8-minute test beats a badly executed 20-minute test every time.
Do you want to calibrate your personal correction factor? Do the 60-minute test once. Compare the result to your most recent 20-minute test × 0.95. The difference tells you whether 0.95 is right for your physiology. Adjust accordingly and move on.
Whichever you choose: commit. Run the same protocol, with the same warm-up, in the same conditions, every 6-8 weeks. The trend is the signal. The individual number is just a waypoint.
We talk about testing, training zones, pacing, and the mistakes that cost masters cyclists years of wasted effort inside the Roadman Cycling community on Skool. If you are done guessing and ready to train with a plan — and the support of riders who have already made the mistakes — that is where you start.