Your FTP number means nothing if the test that produced it was taken at the wrong time. That might sound extreme, but I've watched it play out hundreds of times — in the community, through coaching conversations, on emails that land in our inbox. A rider finishes a hard training block, feels strong, decides to test, and the number comes back lower than six weeks ago. Confidence craters. They drop their zones. The next month of training becomes too easy, and the fitness they were building starts to erode. All because they tested on the wrong Tuesday.
The number itself is just a reference point. It sets your training zones, anchors your interval targets, and gives you a way to measure progress over time. But the conditions under which you capture that number determine whether it's useful or actively destructive. Timing is the variable that separates a good test from a bad one.
Contents
- The Standard Rhythm: Every 6-8 Weeks
- When Testing Actually Helps
- When Testing Does More Harm Than Good
- Ramp Test vs 20-Minute Test vs 8-Minute Test
- Signs Your FTP Has Moved Without a Formal Test
- Auto-Detection: Where the Software Fits
- Getting the Timing Right
The Standard Rhythm: Every 6-8 Weeks
The textbook recommendation exists for a reason. Six to eight weeks is roughly the time it takes for meaningful physiological adaptation to show up in a sustained-power test. Your body needs stimulus, time to absorb that stimulus, and enough repeated exposure for the adaptation to become stable and measurable.
Test too frequently — every two or three weeks — and you're measuring noise. Day-to-day variation in sleep, hydration, nutrition, stress, and muscle glycogen can shift your result by 5-10 watts in either direction. That's not fitness change. That's Tuesday versus Thursday.
Test too infrequently — every four or five months — and your training zones quietly drift out of alignment. You're doing sweet spot intervals at 88-93% of a number that's now 15 watts too low. The sessions feel comfortable. You tick the boxes. But you're not getting the stimulus you think you are. Months pass before you realise the work has been slightly too easy the entire time.
Six to eight weeks lands in the productive middle. It's long enough for real adaptation to accumulate and short enough to keep your zones honest. But — and this is where most people get it wrong — the interval only tells you how often. It says nothing about when within that window. That's where context takes over.
When Testing Actually Helps
Three moments stand out as productive times to test.
At the transition between training phases. If you've just finished a 12-week base block and you're about to move into build-phase intensity, testing gives you an updated anchor for the harder work ahead. Your zones will be set correctly for the intervals that matter most. Joe Friel made this point on the podcast — if you're about to stack race-specific intensity on top of your base, you want to know exactly where threshold sits so those efforts land in the right place.
After a recovery week. This is the single best time to capture an accurate number. You've absorbed the training from the previous block, your glycogen stores are topped up, accumulated fatigue has dissipated, and your nervous system is fresh. The result reflects your real, current fitness rather than your fitness minus the fatigue you've been carrying. If your plan already includes a recovery week every fourth or fifth week, slot the test at the end of it. You're already rested. The disruption to your schedule is minimal.
At the start of a structured training plan. Before you begin any new programme, you need a baseline. Without it, every zone-based prescription in the plan is a guess. The test doesn't need to be perfect — it just needs to be done under controlled conditions so that the numbers you train from are grounded in something real.
The common thread is freshness. Every productive test happens when you're rested, fuelled, and physically capable of producing a genuine maximal effort. Without those conditions, the result is fiction.
When Testing Does More Harm Than Good
This is where most of the damage happens, and it's almost always driven by curiosity or anxiety rather than any structured plan.
Mid-build block. You're three weeks into a five-week VO2max block. The sessions are hard. You can feel yourself getting fitter. You want to know if the number has moved. So you test.
Two problems. First, you're fatigued. Three weeks of accumulated high-intensity work has loaded your system. Your acute performance is suppressed even though your underlying fitness is improving. The test result will almost certainly understate your real threshold. Second, the test itself is a maximal effort that disrupts the carefully constructed training stimulus. Your coach — or your plan — has prescribed specific work in a specific sequence for a reason. Dropping a maximal threshold effort into the middle of that sequence throws off the recovery-adaptation cycle for everything around it. You've inserted a wrecking ball into a structure that was doing its job perfectly well.
When carrying significant fatigue. This extends beyond mid-block testing. If you had a terrible night's sleep, if you're stressed about work, if you rode hard yesterday, if you're slightly dehydrated — the test result will be lower than your actual fitness. And that lower number will set your zones too low for the weeks that follow. You'll train below the intended stimulus, wonder why you're not progressing, and potentially test again out of frustration. It becomes a cycle of bad data producing bad training producing bad tests.
Professor Stephen Seiler has spoken about the difference between your capacity on a good day and your capacity on a normal day. A proper FTP test should capture something close to your good-day capacity, because that's what your interval targets need to be pegged to. Testing on a bad day gives you a bad-day number, and you don't want to train from a bad-day number.
During race season. If you're racing regularly, the races themselves are giving you data. A hard road race or time trial produces power files that tell you far more about your current fitness than any indoor test. Your 20-minute, 40-minute, and 60-minute power from racing — where motivation and competition push you harder than any solo effort in a pain cave — is a more honest reflection of where you stand. Testing mid-season adds fatigue without adding information. Use your race data instead.
Ramp Test vs 20-Minute Test vs 8-Minute Test
The protocol matters less than most people think. What matters more is consistency — using the same test, under the same conditions, every time. That said, each protocol has strengths and weaknesses worth understanding.
The ramp test starts easy and increases wattage by a fixed amount every minute until you can no longer hold the target. It's quick — usually 10 to 20 minutes from warm-up to failure. Your FTP is estimated from the final completed stage, typically as 75% of your one-minute peak power during the test.
The ramp test suits riders who are new to testing. It requires no pacing skill. You just keep going until you can't. The result is highly repeatable because the protocol removes the pacing variable entirely. But it has a significant weakness: it rewards anaerobic capacity disproportionately. If you have a big sprint and strong anaerobic system, the ramp test will overestimate your FTP because your one-minute peak is inflated relative to your sustainable threshold. Conversely, if you're a diesel — built for long, steady efforts with a modest kick — the ramp test may underestimate your threshold.
The 20-minute test is the classic. You ride as hard as you can sustain for 20 minutes, and FTP is estimated at 95% of the average power. It requires pacing discipline — go out too hard and you'll fade; go out too conservatively and you'll leave watts on the table. Most experienced cyclists need three or four attempts before they learn how to pace it properly.
The 20-minute protocol captures sustained aerobic capacity more accurately than the ramp test because it actually requires you to hold a threshold-range effort. It's the closest approximation to what FTP represents — the maximum power you can sustain for roughly an hour. Its weakness is that pacing errors introduce variability. Two identical athletes with different pacing skills will produce different results, and that difference has nothing to do with fitness. For riders who know how to pace, it remains the most reliable option.
The 8-minute test splits the difference. Two 8-minute maximal efforts separated by a recovery period. FTP is calculated from the average of the two efforts. It's shorter and slightly easier to pace than a 20-minute block, which makes it more accessible for riders who mentally struggle with the long grind of a 20-minute protocol.
The trade-off is that a shorter effort leans more heavily on anaerobic contribution, much like the ramp test. The dual-effort structure helps mitigate this — if you go out too hard on the first effort, you'll pay for it on the second — but it's still not as pure a measure of threshold as 20 minutes of sustained work.
Which one should you pick? If you're experienced, disciplined at pacing, and want the most accurate number, the 20-minute test is hard to beat. If you're new to testing, want something quick and repeatable, or test frequently as a check-in, the ramp test works well — just be aware of the anaerobic bias. If you struggle psychologically with the 20-minute block but want something more than a ramp, try the 8-minute protocol. Whichever you choose, stick with it. Switching protocols between tests makes comparison meaningless.
Signs Your FTP Has Moved Without a Formal Test
You don't always need a test day to know your threshold has shifted. If you're paying attention to the right signals, your regular training data will tell you.
Workout compliance at lower RPE. This is the most reliable indicator. You're completing the same interval sessions — same wattage, same duration, same rest — but the perceived effort is noticeably lower. Three weeks ago, 4x8 minutes at 270 watts felt like an 8 out of 10. Now the same session sits at a 6 or 7. Your body has adapted. The work hasn't changed; your capacity has.
Heart rate drift at given power. If your average heart rate during a threshold or sweet spot session is trending lower over successive weeks — while the power target stays the same — that's a clear sign of improved aerobic efficiency. Your cardiovascular system is delivering the same power output with less effort. When I had Professor Seiler on the podcast, he talked about heart rate as a window into the cost of producing a given power output. When that cost drops, fitness has moved.
HR:power decoupling on long rides. On a 3-4 hour endurance ride at steady power, watch how your heart rate behaves. Early in a training block, heart rate will drift upward relative to power as the ride progresses — a sign of cardiovascular fatigue. As you get fitter, that drift diminishes. If you used to see a 10% decoupling between the first and second half of a long ride and now it's down to 3-4%, your aerobic engine has strengthened.
Interval extension. You were prescribed 4x10 minutes at sweet spot and found yourself able to push a fifth interval without collapsing the structure. Or you held the power target comfortably for 12 minutes instead of 10 on the last rep. That capacity to do more work at the same intensity is a proxy for increased threshold, even if you haven't pinned an exact watt number to it.
Strava and training platform PRs during regular rides. Not KOMs — those depend on too many variables. But if your power curve is quietly shifting upward across multiple durations during normal training rides, the data is telling you something your legs already know.
None of these signals replace a formal test. But they reduce the urgency of testing and help you avoid the trap of testing too often or at the wrong time. If three of these indicators are pointing upward, you can be reasonably confident that your FTP has improved — and you can wait for the right moment to confirm the exact number.
Auto-Detection: Where the Software Fits
Algorithmic FTP detection has removed the need for regular test days for a lot of riders, and the tools are getting meaningfully better.
TrainerRoad's AI FTP Detection analyses your workout history — power data, workout completion rates, how your performance tracks against predicted targets — and updates your FTP estimate without requiring a test. It's surprisingly accurate for riders who train consistently on the platform, because the algorithm has a rich dataset of your actual performance across hundreds of intervals. For TrainerRoad users who follow their plans closely, this has largely replaced manual testing for routine zone adjustments.
Xert's Fitness Signature takes a different approach. Rather than estimating a single FTP number, Xert models your entire power-duration curve and derives a threshold estimate from where that curve crosses into the sustainable range. It updates continuously based on any "breakthrough" effort — a ride where you pushed beyond your current model. The advantage is that it doesn't require a dedicated test; any hard effort feeds the model. The disadvantage is that if you never ride hard enough to produce a breakthrough, the model stagnates.
WKO5's mFTP (modelled FTP) uses a power-duration model built from your 90-day power curve to estimate threshold. It's considered one of the more robust algorithmic approaches because it accounts for the full shape of your power curve rather than relying on a single test duration. It works best for riders with a diverse training profile — if all you do is 90-minute endurance rides, the model won't have enough data points in the threshold and above-threshold range to estimate accurately.
Where auto-detection fits in practice. Think of these tools as a running estimate between formal tests. They're good at tracking trends — if AI FTP Detection says you've gained 8 watts over the past six weeks, that trend is almost certainly real even if the absolute number is off by a few watts. They reduce the need for disruptive test days, especially mid-block when testing would do more harm than good.
But they benefit from periodic calibration against a manual test. Every 12-16 weeks, do a proper test — rested, fuelled, controlled conditions — and compare the result against what the algorithm says. If they're within 3-5 watts, trust the software between tests. If the gap is larger, recalibrate and investigate why the model drifted.
Getting the Timing Right
The fixable mistake here isn't that people don't test. It's that they test reactively rather than proactively. They feel good after a big ride and decide to test on a whim. They're anxious about whether training is working and test to soothe the anxiety. They see a new test protocol trending on YouTube and want to try it mid-block. Every one of these impulses produces a test that's either inaccurate, disruptive, or both.
Build your testing into your training calendar the same way you build recovery weeks and race preparation. Know the date before you start the block. Taper into it with a couple of easy days. Control the conditions — same time of day, same fuelling, same warm-up, same equipment. The test becomes a data point you can trust rather than a number you have to second-guess.
And if you're between tests and you can feel the fitness building — RPE dropping, heart rate settling, intervals feeling more manageable — trust that signal. Not every improvement needs a number attached to it immediately. Sometimes the smartest thing you can do is keep your head down, keep doing the work, and test when the time is right.
If you want to talk through any of this with riders who are working through the same questions, the Roadman Cycling community on Skool is where those conversations happen daily. FTP testing, zone calibration, when to adjust and when to hold — it's the kind of thing that's easier to sort out with people who've been through it.