Methodology
Every workout and plan on this site is generated from written rules, not from a model. This page is those rules, in summary. If something here looks wrong to you, I'd genuinely like to hear it — that's more useful to me than a compliment.
Last revised · August 2026
The zones
Everything is anchored to FTP — the power you could hold for roughly an hour. Zone boundaries follow the standard Coggan model, which is also what Intervals.icu uses, so the numbers you see here match the numbers you see there.
🌱 Recovery (≤55% FTP)
Blood flow and lactate clearance without adding load
🚴 Endurance (≤75% FTP)
Mitochondrial adaptation. Slow, cumulative, needs months
⚙️ Tempo (≤90% FTP)
Aerobic stimulus at manageable recovery cost
🔶 Threshold (≤105% FTP)
Raising the ceiling you can sustain
🔥 VO2max (≤120% FTP)
Aerobic power
⚡ Anaerobic (≤150% FTP)
Sprint and repeated-effort capacity
💥 Neuromuscular (150%+ FTP)
Peak power and recruitment, not metabolic
Sweet spot sits at 88–94%, deliberately straddling the tempo/threshold boundary. That overlap isn't sloppiness on my part — the literature treats sweet spot as its own concept rather than a slice of either neighbour.
Building a session
One target per phase
Almost every validated interval protocol in the literature is a single fixed combination — target power, work duration, rest duration, repeats — repeated consistently. Not a sliding scale of shifting targets. The researchers behind the 30/15 protocol are explicit that intensity must stay identical on every repeat.
So sessions are built as one to three clearly separated phases, each with a fixed target. What gets avoided is fragmentation: lots of short segments with different targets and no recognisable pattern.
Total dose matters more than repeat count
Each zone has a session-level target rather than a number of intervals:
Tempo
8–10 min × 3–4 — 24–40 min in zone
Sweet spot
10–15 min — Recovery 30–50% of work
Threshold
8–20 min — 20–60 min at threshold
VO2max
3–5 min, or 30/15 — 10–20 min above 90% VO2max
Anaerobic
20–30 s, 2–4 min rest — ~15 min total session
The VO2max figure is the important one. It's minutes actually spent above 90% of VO2max, not total interval time — a meta-analysis puts the minimum at around 10 cumulative minutes. That's why the site avoids classic sprint interval training with long recoveries: it feels brutal and accumulates very little time where it counts.
Warm-up and cool-down
Every session has both. The more explosive the main effort, the longer the ramp into it. Cool-downs are typically shorter, since they only need to bring heart rate down rather than drive an adaptation.
Weeks and blocks
Hard sessions are capped by what the recovery literature supports: threshold and VO2max once or twice a week, sweet spot and tempo up to two or three times, endurance and recovery with no real ceiling as long as the intensity genuinely stays low. The most common mistake in that last category is easy rides quietly creeping into the grey zone.
Blocks run three progressive weeks followed by one lighter week. After two or three of those in a row — roughly eight to twelve weeks — a longer reset of two to four weeks is scheduled, because progressive blocks stacked indefinitely stop producing anything.
Within a block, week-on-week load increase is capped, and the recovery week drops to a set fraction of the preceding week rather than being left to chance.
Where the research disagrees
Three places where I could have picked a winner and chosen not to. Anywhere you see a site claim consensus on these, be a little suspicious.
VO2max intervals: short or long?
Short wins
Rønnestad and colleagues found that 30/15 protocols produced greater gains in FTP and VO2max-related performance than classic 5×5 min in well-trained cyclists, with more time spent near VO2max despite the shorter blocks.
Long wins
In highly trained middle-distance runners, 3-minute intervals produced significantly more time above 90% VO2max than 30-second intervals — the exact opposite result, despite higher heart rates in the short protocol.
How it's handled. Both are offered. The deciding factor doesn't appear to be interval length but how efficiently the protocol as a whole — recovery duration especially — accumulates time at high intensity. That evidently differs by population, so neither is treated as default.
Polarized or sweet spot?
Polarized
Seiler's work shows elite endurance athletes across different sports independently converge on roughly 80% low intensity and 20% high, with little time in between. Stöggl & Sperlich (2014) found greater gains from polarized than from threshold-focused training.
Sweet spot
Below roughly 6–8 hours a week, zone 1–2 sessions are simply too short to produce the deep aerobic effect polarized relies on. Sweet spot extracts more stimulus per available hour.
How it's handled. Weekly hours decide it. Under about 8 hours, sweet spot and tempo get a larger share. Above 10–12 hours, the split moves toward 80/20 with minimal time in the middle. Worth noting the polarized studies were run on already well-trained athletes, which doesn't generalise cleanly to a recreational rider on six hours a week.
FTP might not mean what we all assume
Wong & Burnley (2022) found the average time riders could actually sustain FTP was 33.7 minutes, not the assumed 60. Since every zone on this site is a percentage of FTP, that means the whole scale can be slightly miscalibrated for any given rider.
How it's handled. Honestly, it can't be fully solved without lab testing. What it does mean is that a current FTP figure matters more than a precise one, which is why the site nudges you toward retesting rather than treating your number as permanent.
What changes your day
Dynamic Plans reads your last 42 days from Intervals.icu each morning and compares your current form and HRV against what the plan expects at this point in the block.
Most mornings nothing happens, and that's the intended behaviour — a plan that changes constantly isn't a plan. When your form is deeper than the block anticipated and your HRV has sat below your own baseline, the hard session moves to a later day and you get something of the same duration at a lower intensity. The week's volume stays intact; only the placement of the load changes.
The logic is deterministic. Same inputs, same output, every time. There is no model involved and nothing about your training is used to train anything.
What this doesn't do. It can't see that you slept badly if your HRV didn't register it, that you're coming down with something, or that you've got a hard week at work. Nothing here replaces knowing when to stop. If a session feels wrong, it is wrong — skip it.
Sources
The two books do most of the work. The papers cover the places where the books are either silent or out of date.
The Cyclist's Training Bible
Joe Friel
Training and Racing with a Power Meter
Coggan & Allen
Short vs long intervals in well-trained cyclists
Rønnestad et al.
Polarized vs threshold training distribution
Seiler; Stöggl & Sperlich 2014
Time to exhaustion at FTP
Wong & Burnley 2022
Lactate clearance in WorldTour riders
San Millán
Each rule in the underlying rule set carries its source and an evidence level, so I can tell the difference between something well established and something I've extrapolated. Where I've had to assume rather than cite, it's marked as an assumption.