Critical Power Cycling: Test CP and Use W′ Smarter

What Critical Power Means for Cyclists

Critical power cycling is useful because it separates two things riders often blur together: the power you can repeatedly stabilize around, and the limited amount of work you can do when you ride above that line.

Critical Power, usually shortened to CP, is the threshold-like output produced by a cycling power-duration model. It estimates the boundary between hard work you can keep under control and severe work that pushes you toward exhaustion.

The practical value is not the number by itself. It is the decision-making structure around it. Below CP, fatigue still accumulates, but the system is more stable. Above CP, you are spending W′, the model's estimate of your above-threshold work capacity.

That distinction explains why two riders with similar threshold numbers can race very differently. One may be better at sitting close to threshold without wasting energy. Another may have more capacity for repeated surges. CP and W′ give those differences a usable framework.

Critical Power vs FTP: Where Each Metric Helps

FTP and CP are closely related, but they are not the same tool.

FTP is usually used as a training anchor. It helps set zones, compare changes in aerobic fitness, and organize steady-state work. If you are building workouts around endurance, tempo, threshold, and over-under sessions, FTP is still a practical reference point.

CP comes from a model built from multiple maximal efforts. Instead of asking what you can sustain in one test format, it describes the shape of your power-duration curve across different effort lengths.

That makes CP especially useful when the question is not only whether your threshold improved, but how your power falls away as duration changes. Did you gain steadier long-effort durability, sharper short power, or a better balance between the two? FTP may show the general shift. CP helps explain the shape of it.

For pacing, the distinction matters. If you are riding a steady climb or time trial, FTP can be a clean starting point. If the race is full of attacks, corners, climbs, accelerations, and repeated surges, CP plus W′ gives you a better language for what is actually happening.

The mistake is treating one metric as the winner and the other as obsolete. FTP is useful because it is simple and broadly understood. CP is useful because it models the cost of hard work above threshold. The more variable the event, the more valuable the CP model becomes.

How W′ Explains Hard Efforts Above Threshold

W′ is the model's estimate of anaerobic work capacity. In plain terms, it describes how much work you can do above CP before that reserve is gone.

This is where W prime cycling analysis becomes practical. When you surge above CP, the cost is not only that the effort feels hard. You are drawing down a limited reserve. The further above CP you ride, the faster that reserve is spent. Recovery below CP can restore some capacity, but it is not instant, complete, or independent of context.

That explains a familiar race pattern. A rider feels comfortable near threshold, follows one hard move, recovers a little, follows another, and then suddenly has nothing left. The collapse feels abrupt because the limiter was not only aerobic strain. The rider had used too much above-CP capacity and did not get enough meaningful recovery before the next demand.

This is also why pacing above CP has to be intentional. A small early mistake can matter later if the route or race situation demands another acceleration. The model does not tell you which move is tactically correct, but it clarifies the cost of making it.

A Practical Field-Testing Protocol for CP

A critical power model is only as good as the maximal efforts feeding it. The goal is not to collect one heroic ride file. The goal is to capture different parts of your power-duration relationship with clean, repeatable inputs.

A practical starting protocol is to test short, medium, and longer maximal efforts across a small testing block. This is not a universal rule, but it is a useful way to get cleaner data than relying on random group ride files.

Example field-testing block

• Day 1: maximal 3-minute effort
• Day 2: easy riding or rest
• Day 3: maximal 12-minute effort
• Day 4: easy riding or rest
• Day 5 or Day 6: maximal 20-minute effort

If life, weather, or fatigue makes that layout unrealistic, keep the principle and adjust the exact spacing. The key is that each effort should be fresh enough to represent a true maximum, not just the best you could do while carrying fatigue from the previous test.

Some riders combine a short and medium effort in one session, but that is a compromise. If you do it, put the shorter effort first, take full low-intensity recovery, and only continue if you can start the next effort with stable breathing, stable power, and no obvious residual deadness from the first test. If the second effort is clearly compromised, it is better to reschedule than to feed the model bad data.

How to ride each test

Use terrain or an indoor setup that allows uninterrupted pressure on the pedals. Avoid routes with traffic, braking, forced coasting, sharp corners, steep gradient changes, or obvious drafting effects. Use the same power meter setup throughout the block, and warm up in a way that is familiar enough not to become another variable.

The effort should be maximal, but not chaotic. Start hard enough that the file reflects real capacity, but not so hard that the test becomes an avoidable detonation. A usable file usually shows continuous power, a clear maximal commitment, and no external interruption. The final section should feel like you are emptying what is left, not discovering that you saved too much.

What makes a test file usable or unusable

A file is more usable when:

• The effort is continuous from start to finish
• Power comes from the same trusted meter setup
• There is no meaningful coasting, braking, traffic, or cornering interruption
• The effort is clearly maximal for that duration
• Pacing is imperfect but not self-sabotaging
• Conditions and position are similar enough to the riding you want to model

A file becomes suspect when:

• You stop pedaling or soft-pedal during the target effort
• A descent, corner, vehicle, or group dynamic changes the power demand
• The power meter setup changes or calibration is questionable
• You finish with too much left because the effort was underpaced
• You explode so early that the result says more about execution than capacity
• Heat, fueling, illness, or fatigue obviously suppresses the output

Bad files are not useless forever, but they should not be treated as clean maximal inputs. If the model is going to guide race pacing, the testing needs to be cleaner than a hard workout that happened to hurt.

Race files can help validate the model, especially when motivation is higher than in solo testing. The problem is that races rarely produce clean standalone efforts at exactly the durations the model needs. A race surge interrupted by drafting, cornering, or soft pedaling may be tactically decisive without being a good test file.

The best approach is to use structured testing to build the model, then use racing to check whether the model explains what actually happened.

Common Data Problems That Skew Critical Power

Critical power can look precise because software draws a clean curve. That does not mean the inputs are clean.

The most common problem is stale data. If the model includes old best efforts from a different training phase, different body mass, different equipment setup, or different fitness level, it may describe a past version of you. That is historical clutter, not current analysis.

Another problem is non-maximal data. A hard group ride pull is not the same as a maximal test. A climb ridden at controlled race pace is not the same as emptying the tank. If the data never forces the model to see your actual limit, CP can be underestimated and W′ can be misleading.

Power meter consistency matters as well. Swapping between devices, mixing indoor and outdoor files without understanding the difference, or riding with poor calibration habits can create a curve that looks physiological but is partly mechanical.

Pacing errors also matter. A maximal effort that starts recklessly and collapses may understate what you could do with better distribution. A maximal effort that finishes too fresh may understate your true capacity. The model cannot always know whether the limitation came from physiology or execution.

Finally, do not ignore context. A power number produced in a comfortable upright position may not transfer perfectly to an aggressive race position. Heat, fatigue, fueling, and terrain all change the expression of fitness. CP is a model of capacity, not a guarantee that capacity will appear under every condition.

Using Critical Power for Race and Interval Pacing

The best use of CP is not to stare at it mid-race like a speed limit. It is to understand what different decisions cost.

Imagine a rolling race with a decisive climb late. Early in the race, a small gap opens after a corner. You can jump well above CP to close it immediately, or you can hold a steadier effort, use the draft as the group stretches, and close it without a full commitment. If the move is not decisive, the second choice usually preserves more tactical flexibility.

Now put the same rider on the decisive climb. If the selection is forming and the race is being made, going above CP can be the correct decision because the cost has a purpose. The difference is not simply hard versus easy. It is whether you are spending W′ to change the race, or spending it to solve a problem that patience and positioning might have solved more cheaply.

That is the actionable shift CP and W′ provide. Non-decisive surges should be questioned. Decisive surges should be planned for. If the course has a known selection point, avoid drifting above CP too often before it unless the tactical return is obvious.

For intervals, the same logic applies. If the goal is to improve repeatability above threshold, the workout should intentionally draw on W′ while allowing partial recovery. If the goal is steadier threshold control, the work should sit closer to CP and avoid turning every repetition into an anaerobic withdrawal.

A practical interval example: on hard-start climb repeats, the opening surge above CP can be deliberate, then the rider settles near CP rather than continuing to chase the highest possible average power. That teaches the body and the rider to absorb an acceleration and regain control. If every repeat starts with an uncontrolled surge and ends in survival mode, the session has changed from targeted pacing practice into unmanaged depletion.

Useful applications include:

• Setting hard-start efforts where the opening surge is planned, not accidental
• Designing over-under work that teaches control around CP
• Pacing race simulations by deciding where W′ should be spent
• Reviewing race files to identify whether failure came from sustained CP strain or repeated above-CP cost
• Comparing athletes with similar threshold power but different ability to repeat surges

This is the advantage over simpler pacing rules. Average power can hide the cost of variability. Summary metrics can describe load, but they do not show the above-CP reserve as clearly. CP and W′ help explain why the same average output can feel controlled in one ride and impossible in another.

When to Retest and How to Track Changes Over Time

Retesting should happen when the model is likely to be wrong, not just when curiosity spikes.

Good times to retest include after a focused training block, after a meaningful change in fitness, after time away from structured work, or before a goal event where pacing precision matters. You should also retest if recent race or workout data repeatedly disagrees with the model. If efforts above CP feel too easy, the model may be stale. If workouts based on the model are consistently impossible despite good recovery and fueling, the inputs may be too optimistic.

Track CP and W′ together. Looking at CP alone can miss the story. A rider may raise CP while W′ falls, which could be a reasonable tradeoff for steady climbing or time trialing. Another rider may maintain CP while improving W′, which could be more valuable for punchy racing. The right adaptation depends on the event demands.

Also track the quality of the data behind the number. A model built from recent, clean, maximal efforts is more actionable than one built from a scattered set of files. When reviewing progress, ask:

• Are the efforts recent enough to describe current fitness?
• Were they truly maximal?
• Did they cover different durations?
• Were they produced with the same power meter setup?
• Do race outcomes support what the model suggests?

The goal is not to chase a perfect curve. The goal is to make better decisions. Critical power cycling analysis should help you pace climbs, judge attacks, structure intervals, and understand why you cracked or survived.

Used well, CP and W′ turn power data into a tactical map. CP tells you where stability starts to break down. W′ tells you how much above-threshold work you can afford. The race still decides when to spend it, but the model helps you stop pretending every hard effort costs the same.