An FTP that hasn't moved in eighteen months is not bad luck. It is a structural pattern, and there are roughly five of them. The fix is rarely "train harder." It is almost always to find the input you have been holding constant while the rest of your life moved on, and change that one.
This is the diagnostic framework we use with riders applying to the Not Done Yet coaching community. Most plateaued cyclists arrive convinced they need a bigger plan. About one in twenty actually does. The other nineteen need to look at one of the five causes below.
Cause one: too much grey-zone training
The single most common cause. The rider trains four or five times a week. None of it is genuinely easy and none of it is genuinely hard. Everything sits at zone 3 — RPE 6-7, breathing perceptibly heavier than easy, perceptibly lighter than threshold.
This is the comfort zone of the trained amateur. It feels productive. It produces the satisfying ache of a session you'd describe to your group as "a good ride." It also crowds out the two intensities that drive adaptation: the patient zone 2 work that builds aerobic base, and the focused threshold and VO2 work that pushes the ceiling up.
Prof. Stephen Seiler's polarised-training research is the durable signal here. Across twenty years of studying trained endurance athletes, the pattern is consistent: 80% of training time at low intensity, 20% at high intensity, almost nothing in the middle. Trained amateurs who shift from heavy zone 3 to a polarised distribution typically gain 5-10% on FTP within a season, with no increase in total hours.
The diagnostic question. Pull the last four weeks of training files. What percentage of time is spent at heart rate above 70% of maximum but below your lactate threshold? If the answer is more than 25%, you have the grey-zone problem. The number is often closer to 40% in plateaued riders.
The fix. Cap intensity on easy days at 70% of max heart rate or RPE 4. If you cannot have a full conversation, you are riding too hard for the easy day. Move two sessions a week to deliberate hard — sweet spot, threshold, or VO2 — and let the rest sit clearly easy. The week starts to look polarised. The fitness usually starts moving in 6-8 weeks.
Cause two: chronic under-recovery
The training looks correct on paper. The intensity distribution is right. The hours are reasonable. The FTP still won't move.
Look at recovery. The thing about being thirty-eight with a job and three kids is that the recovery profile underneath the training is not the recovery profile of the athlete the plan was written for. Chronic under-recovery is the silent killer of trained amateurs.
It shows up in three places.
Sleep. Less than seven hours, three or more nights a week. Or seven hours of broken sleep that looks fine on a Garmin but reads as fragmented when an Oura or Whoop is honest about it. The growth-hormone window for muscle adaptation is heavily concentrated in the deep-sleep stages of the first four hours. Lose those and the work doesn't bank.
Deload cadence. A 3:1 cycle for a 35-year-old, a 2:1 cycle if life is heavy, is roughly the right pattern. Riders who run 4:1 or 5:1 because the legs feel fine are usually the ones whose FTP plateaus six months in, when the cumulative fatigue catches up with the form curve.
Life-stress load. A heavy work quarter or a prolonged family stretch will compete for the same cortisol bandwidth as your training. A rider who suddenly stalls when nothing in the plan changed has often had something change outside it.
The diagnostic question. Has anything outside training changed in the last six months? Job stress, sleep duration, kids' schedule, alcohol intake, travel? If the answer is yes and the FTP plateau coincided, you found the cause.
The fix. Bring the deload cadence in to 2:1 for two cycles. Take the week's volume off by 30-40%, hold the intensity but reduce the volume of each hard session, and protect sleep aggressively for those weeks. Most riders see the fitness move in the second cycle.
Cause three: base aerobic underdevelopment
This one is invisible until you stop and look. The rider has been doing intervals for years. Lots of intervals. The threshold sessions are sharp, the VO2 ceiling looks fine. But the base — the durability that lets you absorb training load week after week — is thin.
The signal is durability under fatigue. The rider can hit big numbers fresh, but power-at-heart-rate drops sharply after two hours. The third hour is a different file. Long sportives or stage races collapse in hour four. The watts are there in a vacuum; they are not there at the end of a real ride.
This is the masters cyclist who has been doing TrainerRoad workouts for five years and has never spent a winter doing a deliberate base block.
The diagnostic question. Is your power on a four-hour ride at hour four meaningfully lower than your power at hour one for the same heart rate? A drift of 5-7% is normal; 10-15% is base underdevelopment.
The fix. Twelve weeks of patient zone-2 base, with one threshold session per week and no VO2 work. Most experienced amateurs hate this prescription because it feels slow. The riders who do it almost always come out the other side with both a higher absolute FTP and a much higher durable FTP. Joe Friel's framework on this question — that the base is not a phase you graduate from but the substrate every other quality is built on — is durably correct.
Cause four: under-fuelling
The cyclist who arrives at intervals already in glycogen debt cannot adapt to the intervals. The body in caloric deficit, with insufficient carbohydrate available, has the metabolic priorities of survival, not adaptation. You cannot train your way out of an under-fuelled body, and most amateur cyclists are slightly under-fuelled most of the time.
This shows up at two scales.
Across the day. Total daily energy availability — the calories left over after exercise, divided by lean body mass — is often well below the 45 kcal/kg/day floor that endurance research uses as a sustainable target. Energy availability below 30 kcal/kg/day for any sustained period crashes hormones, suppresses adaptation, and is associated with the relative-energy-deficiency-in-sport (REDs) signature now well-documented in cycling.
In-ride. Eating only when the rider feels hungry, on rides over 90 minutes, leaves them dropping into glycogen debt before the body asks for fuel. The recovery cost is paid the next day in a session that should have been hard but isn't.
The diagnostic question. Are you eating 90-120g of carbohydrate per hour from the first hour of any ride over 90 minutes? Are you eating more, not less, on hard training days?
The fix. Front-load the day's fuelling around the workout. 1.0-1.5g of carbs per kg of body weight in the two hours before a hard session. 90-120g/hour during a ride longer than 90 minutes. 1.6-2.0g of protein per kg of body weight per day, distributed evenly across meals. The plateau often moves within a fortnight when this changes.
Cause five: measurement noise
The least romantic cause, and the most common. The FTP didn't actually plateau. The measurement did.
Every step of the test protocol is a place for drift to enter. The trainer was calibrated last month. The power meter zero-offset hasn't been done in six rides. The 20-minute test was done on a different course. The ramp test was done at the end of a heavy week last time and at the start of a recovery week this time. The room was warmer or colder. The tyre pressure was different.
Any one of these is small. Two or three of them together produce a 3-5% drift, which is more than enough to make a real improvement look like a plateau, or to mask a real decline.
The diagnostic question. Have you used the same test protocol, the same equipment, the same conditions, and the same physical state for the last three FTP tests?
The fix. Standardise. Run the same test, on the same trainer or the same loop, with calibrated equipment, at the same time of day, well-rested, after the same warm-up. Most riders find that two of the last four "stalled" tests were actually fine, and one of the apparent improvements wasn't real. The signal becomes interpretable.
Putting it together
Most plateaued riders have not one of these causes but two. The grey-zone problem and chronic under-recovery cluster together — the rider who never goes easy is also the rider who never properly rests. Base underdevelopment and under-fuelling cluster together too — the rider who never built the aerobic floor is often the rider who skips breakfast and starts hard rides with three espressos and a bag of fig rolls.
The diagnostic order matters. Start with measurement noise, because if the test isn't reliable you cannot evaluate any of the other interventions. Then check intensity distribution — that's the single biggest lever. Then look at recovery. Then base. Then fuelling.
Three months of the wrong fix doesn't break a plateau. Three months of the right fix usually does. The Plateau Diagnostic at /plateau is a four-minute version of the same five-cause framework, with a specific recommendation at the end based on which pattern matches your situation. The framework above is the longer version of the same logic.
Most plateaus look like fitness ceilings. They are almost always structural. Find the structural answer, change it, hold the change for two months, and the watts move. The riders who never break a plateau are not the riders who can't get faster. They're the riders who keep training harder against the same broken structure.