How to test your FTP without a 20-minute test

The 20-minute test has one real virtue and one real problem. The virtue is that almost everyone uses it, so your number is comparable with everyone else's. The problem is that it is genuinely unpleasant, which means riders postpone it, ride it half-committed, or quietly stop testing altogether and let their zones drift for a season.

There are several other ways to arrive at a threshold number, and some of them are shorter, easier to execute, or require no dedicated test at all. What follows is what each one actually measures, and — the part usually skipped — how far each sits from the thing FTP is nominally defined as: the power you could hold for about an hour. If you want the concept itself rather than the testing, start with FTP explained.

The protocols, side by side

Each of these produces a number people call FTP. They are not measuring quite the same thing, and they do not agree with each other on an individual basis.

Why the ramp test is the easiest and the least trustworthy

A ramp is the friendliest test ever designed: you sit on the trainer, the resistance climbs by a fixed increment every minute, and you stop when you cannot turn the pedals. There is no pacing decision to get wrong, and it is over in ten minutes or so.

The catch is the arithmetic afterwards. Turning a ramp result into a threshold number means multiplying your peak ramp power by a fixed percentage, and that percentage is a software convention rather than a physiological constant. Reviews of exercise-intensity methods have been blunt about this whole family of fixed-percentage approaches: the relationship between a maximal aerobic power number and an individual's actual metabolic thresholds varies substantially between riders. Two people with identical ramp results can have genuinely different sustainable power.

Practically, the ramp flatters riders with a large anaerobic contribution — they keep producing power well past the point where their aerobic system has topped out, so the peak is high and the derived threshold is too. Diesel-engined riders get the opposite treatment. If your zones from a ramp test feel systematically too hard or too easy, this is usually why.

Critical Power: three efforts instead of one

Rather than picking one duration and applying a correction factor, the Critical Power approach asks you for several maximal efforts of different lengths — a common set is roughly 12, 7 and 3 minutes, with long recoveries between — and fits a curve through them. CP falls out as the asymptote that curve settles towards.

The appeal is that CP corresponds to a real boundary in physiology rather than a convention, and the fit also gives you W′, the finite amount of work you can do above it. The cost is that all three efforts have to be honestly maximal, and a single conservative one skews the fit in a direction that is not visible in the output.

When both were measured in the same trained riders, CP came out slightly higher than 20-minute-derived FTP — about 7 W on average — but the limits of agreement spanned roughly −19 to +33 W. On average, nearly the same number; for any individual rider, not interchangeable. We go into that gap properly in Critical Power vs FTP.

The three-minute all-out test

The shortest credible protocol asks for one three-minute effort ridden flat out from the first pedal stroke, with no pacing whatsoever. Power collapses over the first two minutes and then plateaus; that end-test power approximates CP, and the work done above it approximates W′.

In the original validation, end-test power was statistically indistinguishable from independently measured CP, and the work above it landed close to the separately determined W′. That is a genuinely good result for a single three-minute test. The practical difficulty is behavioural: almost nobody goes all-out from second one on the first attempt. The instinct to save something is strong, and any pacing at all invalidates the result. Expect to need a practice run you throw away.

No test at all: fitting a curve to your rides

If you have months of power files, the information is arguably already there. Take your best effort at every duration, plot them, and fit the same power–duration relationship the CP protocol fits — only using efforts you produced when you were genuinely motivated, which for most riders means races, group rides and long climbs rather than a scheduled test on a Tuesday.

This is how Moveee's power profile works: it derives an estimated FTP and a Critical Power from your own best efforts across durations, and updates as you ride. It is worth being clear-eyed about the limitation, though. The model can only see efforts you actually produced. A rider whose winter contains no hard 5–12 minute efforts will get an estimate that is confidently too low, because nothing in the data contradicts it. If your numbers have drifted down during a base block, the honest interpretation is usually "no recent maximal efforts", not "lost fitness".

The failure modes that ruin any of them

Protocol choice matters far less than execution. These four account for most bad numbers.

So which one should you use?

• If you will actually do it, the 20-minute test is still fine. It has the most validation behind it and the widest comparability. Its weakness is adherence, not physiology.
• If you dread testing, use the ramp — but treat it as an anchor. Take the number, ride a few weeks of threshold work, and adjust based on whether those sessions felt like the right kind of hard. That correction is more informative than the test was.
• If you want a number with a physiological meaning, fit Critical Power. Either from three efforts or from your ride history. It also hands you W′, which is useful in its own right.
• Whatever you pick, keep the conditions identical. Same trainer or same power meter, same room, comparable freshness, same time of day if you can. Consistency of method beats accuracy of method for anyone training rather than publishing.

A threshold number is a means of setting training intensities, and it only has to be good enough to do that job. Pick the protocol you will repeat, repeat it the same way, and spend the energy you save on the training itself.