Ran this experiment across a block of riders I coach: same 20-min FTP test protocol, same climb, two attempts two weeks apart with zero training change in between. Average swing: ±4% just from pacing alone. That's enough to put you in the wrong zone for an entire training block.

Here's what actually moves the needle on test accuracy:

1. Pace the first 5 minutes conservatively, not aggressively.

Most riders go out 8-10% too hot because the legs feel fresh. If your last 5 minutes aren't your strongest 5 minutes of the test, you paced it wrong — throw the number out.

2. The warm-up matters more than people think.

3x1min at ~105% FTP with 2min easy between, done 5-10 minutes before the test, primes your aerobic system without burning matches. Skipping this costs you real watts in minute 15-20 when fatigue hits.

3. Test on the same course/trainer setting every time.

Wind, gradient changes, and even tire pressure introduce more noise than most people's actual week-to-week fitness change. If you can't control for terrain, use a smart trainer in ERG-adjacent mode for consistency.

4. 0.95 multiplier isn't universal.

The classic FTP = 95% of 20-min power works well for trained riders but overestimates FTP for less experienced riders who fade less predictably. If you're newer to structured training, 0.90-0.93 is often more honest.

Your FTP number is the foundation every zone, every interval, and every training decision gets built on for the next 8-10 weeks. A bad number compounds into a bad training block. Test it right the first time.