You just paid for a bike fit. Two days later you run a 20-minute FTP test, and the number is worse than before.
That drop is probably not real. It’s either normal test noise or a body that hasn’t learned its new position yet. A little arithmetic tells you which.
Why Your FTP Number Moves After a Bike Fit
A fit changes your saddle height, your reach, maybe your cleat position or your whole aero shape. Your legs notice before your fitness does.
Think of your pedal stroke as a worn path through the woods. You’ve walked it thousands of times. Move the trailhead a few feet and your legs still want the old route. They fire in the pattern built for the old bike. That mismatch costs a few watts until your nervous system learns the new one.
That relearning has nothing to do with whether the fit itself is good. A great fit can still test low on day one.
An early bad number doesn’t mean the fit failed.
The Two Reasons a Retest FTP After a Bike Fit Can Lie to You
Two separate things can make a post-fit retest untrustworthy. Almost nobody separates them.
Reason one: the test itself is noisy. Step on a bathroom scale twice in a row. You might see two slightly different numbers, even though your weight didn’t change. FTP tests work the same way. Researchers had 8 trained cyclists run the 20-minute test twice, days apart, with nothing else changed. Average power came out 342 watts one day and 344 watts the next. The typical error between runs was 4.6 watts. The correlation was .99 out of 1.0 (MacInnis et al. 2019). Translation: even a clean, controlled retest bounces around by a few watts from pacing and chance alone, not fitness.
Separate research on threshold-power testing looked at the smallest change worth caring about. That number is roughly 1% of FTP (Karsten et al. 2020). Below that, you can’t tell a real shift from statistical noise.
Put both numbers together and you get a simple rule. A retest needs to move by more than roughly 1.5% to 3% of your FTP before you treat it as real. On a 300-watt FTP, that’s about 5 to 10 watts.
A 3-watt drop after a fit change isn’t a verdict. It’s the test shrugging.
Reason two: your body hasn’t adapted yet. Even a genuinely better position asks your muscles to recruit in a slightly new order. That takes rides, not minutes. Test too early and you’re measuring an unlearned skill, not the fit.
Noise and adaptation lag are different problems with different fixes. Confuse them and you might undo a good fit chasing a number that was never real.
What the Saddle Height Research Actually Shows
Saddle height is the one fit variable with real research behind it. A classic 1977 study moved 10 trained cyclists’ saddles 5% above and below the optimal height. Researchers measured oxygen cost at a fixed workload. Oxygen use climbed from 1.61 L/min at the optimal height. It rose to 1.69 and 1.74 L/min at the two off-heights (Nordeen-Snyder 1977).
More recent work found the same pattern at smaller changes. A 2% saddle-height shift measurably changed gross efficiency (19.9% vs. 20.4%) and oxygen cost within a single ride. That happened before any adaptation could kick in (Ferrer-Roca et al. 2014). Below about 4% of leg length, though, the effect is small enough to call trivial.
| Fit variable | Change tested | Measured effect | Adaptation-dependent? |
|---|---|---|---|
| Saddle height | ±2% of leg length | Efficiency and oxygen cost shift within one ride | Acute, same session |
| Saddle height | ≥4% of leg length | Larger, consistent efficiency and joint-load change | Acute effect, likely needs riding time to settle |
| Crank length | ±5mm from preferred | No significant HR or efficiency difference | Not applicable, no real effect found |
| Cleat fore-aft position | 5mm forward vs. backward | %VO2max used shifts (65.0% vs. 59.7%) | Acute, economy shift |
| Aero/TT position | Moderate vs. aggressive tuck | Roughly 2-4% (moderate) to 4-7% (aggressive) power cost vs. upright | Fixed position tax, not something you train away |
A 2024 systematic review covered 47 studies. It confirmed saddle height is the only position variable with anything close to research consensus (Husband et al. 2024). Crank length, bar height, Q-factor, and cleat position all lack a clear, agreed-upon performance link. That review graded most of the underlying studies “poor” or “fair” quality. Be honest with yourself about how much certainty any single fit change deserves.
Crank Length, Cleats, and Aero Position: What Actually Moves Power
Riders assume every fit change should move FTP. Most probably don’t.
A 2017 study tested crank lengths 5mm shorter and 5mm longer than each rider’s preferred length. Heart rate and gross efficiency barely moved across all three lengths (144-145 bpm, 20.1-20.4% efficiency). Hip and knee range of motion clearly changed anyway (Ferrer-Roca et al. 2017). In short: your joints feel a crank-length change. Your engine mostly doesn’t.
Cleat position tells a different story. Moving a cleat 5mm forward of the first metatarsal head raised oxygen cost to 65.0% of VO2max. A position 5mm back used only 59.7%, at matched output (Millour et al. 2018). Your economy shifted. Your ceiling didn’t necessarily.
Aero position is the honest exception. A moderate time-trial tuck costs roughly 2-4% of your seated road power. An aggressive tuck costs 4-7%, according to aggregated fit-industry figures rather than a single controlled trial. If your retest drops after moving to an aggressive aero position, some of that drop is real. It’s a permanent trade for speed, not a fitness loss to chase down.
How Long to Wait Before You Retest FTP After a Bike Fit
Here’s the part no study has actually measured. No published trial tracked cyclists week by week after a real-world fit change to see when FTP stabilizes. What follows is professional judgment, not a research finding, and it should be labeled that way.
Steve Hogg Bike Fitting recommends about 3 weeks of easy-to-moderate riding before judging a new position or adding intensity. Coach Jim Rutberg at CTS recommends at least 2 weeks or 10 rides for small saddle adjustments. For a full position overhaul, like a new aero setup, he pushes that out to 6 months before a goal race.
Both sources describe the same rough pattern. A good first ride comes first. Then a dip in comfort and power over the next few days. Then gradual recovery by week 2 to 3.
Test on day 3 and you catch the dip. Test in week 4 and you catch the real number.
A Rider Who Almost Reversed a Good Fit
Take a rider I’ll call Derek. He’s 34 and races Cat 3. He’d just had a full bike fit done, a new saddle height and a more forward cleat position. His FTP going in was 285W.
Two days after the fit, itching to check the work, Derek ran a 20-minute test. It came back at 274W, an 11-watt drop. He nearly called the fitter to undo the changes.
His coach talked him into 3 weeks of easy, consistent riding first, no judgment, no changes. In week 4 he retested. 291W, a 6-watt gain over his original number and a real 17-watt swing from the panic test. The fit was working. He’d just tested it during the dip.
How to Retest FTP After a Fit Change Without Fooling Yourself
A few rules keep you from chasing ghosts.
Wait at least 2 to 3 weeks, or 8 to 10 rides, before a real retest. For a major aero overhaul, give it longer.
Ignore any change smaller than roughly 5 to 10 watts on an average FTP. That’s inside the test’s own noise band, not a real signal.
Use the same protocol every time. The 20-minute test carries a 4.6-watt typical error. The 4-minute max-power test carries about 8.1 watts. Don’t compare across protocols and call the difference “the fit.”
Track the trend, not one ride. Your FTP history matters more than any single test day. The same logic applies to a single aerobic decoupling reading against its eight-week trend.
AthleteOS lets you log the fit change itself, a new saddle height, cleats, an aero position, right on your workout calendar as a dated event. When you run a retest, it shows how many weeks and rides sit between the fit change and the test. It also flags any retest taken inside the usual adaptation window or inside the test’s own noise band. Pair that with your fitness, fatigue, and form scores and you can tell a bad fit from bad timing before you touch a wrench.
If you want your next retest read in context instead of in isolation, sign up for AthleteOS and connect your Garmin or Strava account. And before you retest at all, know what a good FTP actually looks like for your category, so you’re not chasing someone else’s number either.