Smooth Beats Surgy: How Bike Power Variability Shapes Your Run Split

“Don’t overbike” is probably the single most repeated piece of race-day advice in age-group triathlon — and it’s almost always delivered as a single number: keep your average watts below X. That advice is incomplete. Two riders can post the exact same average power for a bike leg and still get off the bike with very different legs underneath them, because one of them let power spike on every hill, gust of headwind and traffic light while the other rode close to flat. This article works through what the evidence actually says about power variability and the run split that follows it, why “protect the run” and “win the day” are not automatically the same goal, and where the popular Variability Index targets (1.02–1.05) that coaches quote actually come from — because they are useful heuristics, not laboratory-validated thresholds, and we’ll be upfront about that distinction throughout.

Context: why average watts is the wrong lens on its own

Average power tells you how hard you rode overall. It tells you nothing about how that effort was distributed across the ride. A rider who holds a dead-flat 200 W for two hours and a rider who alternates between 160 W and 260 W every few minutes can log an identical 200 W average, but they are not doing the same physiological work. The second rider is repeatedly recruiting a larger share of fast-twitch muscle fibre, generating more lactate, and drawing down glycogen faster during the surges than they replenish it during the lulls. Coaches and pacing-software vendors describe this with the Variability Index (VI): normalised power divided by average power. A VI of 1.00 would be a perfectly flat, constant-power ride; real rides are never that flat because hills, wind, corners, traffic and drafting rules on non-draft-legal courses all force power up and down. The practical question for an age-grouper is not “how do I ride a perfectly flat line” — that’s neither realistic nor, as the evidence below shows, necessarily optimal — but “how much surging is costing me on the run, and is smoothing it out worth the trade-off.”

What the evidence says

The variability penalty in the lab

The clearest direct test comes from a small but tightly controlled study: 12 well-trained male triathletes each completed one hour of cycling at the same mean power output under two conditions — constant power, and variable power that swung between 40% and 140% of maximal aerobic power (averaging 65% MAP) — followed immediately by a 9.3 km run. The variable-power ride was followed by a run that was, on average, 42 ± 37 seconds slower than after the constant-power ride, with the bulk of that decrement — 35 ± 20 seconds — concentrated in the first half of the run International Journal of Sports Physiology and Performance, 2013. That detail matters: the cost of surging doesn’t show up evenly across the run, it shows up hardest right when you’re already compromised coming off the bike.

A separate pacing review reinforces the same pattern at a different distance: 5 km running was significantly faster after a constant-intensity 20 km bike (1,118 ± 72 s) than after a variable-intensity 20 km bike (1,168 ± 73 s), and the review specifically flags that power increases late in the bike leg were detrimental to the run that followed Factors influencing pacing in triathlon, PMC review. In plain terms: attacking in the closing kilometres of the bike to “bank time” before transition is exactly the pattern the evidence says to avoid. A further study specifically in age-group and moderate-level triathletes examined cycling’s effects on subsequent running performance, stride length and muscle oxygen saturation, extending this line of research beyond the well-trained, male-only samples of the earlier work Effects of Cycling on Subsequent Running Performance, PMC.

Real-world pacing: how age-groupers actually ride

Lab findings only matter if they describe what happens on race day, and field data suggests age-groupers are riding into exactly this problem, especially over longer distances. In a study of instrumented triathletes, half-Ironman cycling was paced more stochastically — more erratically — than sprint-distance cycling Pacing strategies during sprint, Olympic and half-Ironman triathlons, Eur J Appl Physiol, 2015. That's a little counter-intuitive: you might expect athletes to settle into steadier pacing over a longer, more strategic distance, but the opposite shows up in the data, right when protecting the run matters most for total finishing time. Coaches who work with pacing files day to day describe the same thing qualitatively: “too many occasions where power goes too high” is flagged as one of the defining long-course age-group mistakes, with the recommendation to keep the Variability Index at roughly 1.05 or below, while still allowing a bit more effort uphill and into a headwind Scientific Triathlon with Ryan Cooper. Pacing-analysis platforms describe well-executed triathlon bike legs sitting around VI 1.02–1.05, with 1.10 and above flagged as erratic pacing Variability Index in Cycling and Triathlon, Best Bike Split. It's worth being direct about what these numbers are: coaching and pacing-software heuristics built from watching thousands of files, not thresholds established in a controlled trial of age-group triathletes. They're a reasonable practical target, not a scientific cutoff — use them as a guide, not a verdict.

The 2024 twist: protecting the run is not the same as winning the day

The most useful recent addition to this picture is a 2024 study that separated two outcomes usually lumped together: the run split in isolation, and the combined bike-plus-run time. In 38 triathletes (21 men, 17 women), riding a 20 km bike leg at 80% of FTP versus 90% of FTP produced genuinely different results depending which outcome you care about. Men ran their subsequent 5 km faster after the 80% FTP ride than after the 90% FTP ride (mean difference 35.1 seconds, 95% CI 2.2–68.1, p = 0.035) — the lower-intensity bike protected the run split, as the surging research above would predict. But combined bike-plus-run time was faster at 90% FTP for both men (57.7 s faster, 95% CI 26.1–89.3, p < 0.001) and women (80.9 s faster, 95% CI 29.7–132.1, p = 0.002) Cycling Intensity Effect on Running Plus Cycling Performance, Int J Sports Med, 2024. In other words: riding a bit harder cost the men time on the run split alone, but the extra time gained on the bike more than made up for it when you add both legs together. “Protect the run” and “minimise total time” are not automatically the same objective, and almost none of the popular pacing content built around a single “ride steady” rule addresses that nuance. It's a genuine trade-off, not a single right answer — and it's worth noting this study used a 20 km lab bike leg, not a 90 km or 180 km race bike leg, so treat the exact numbers as illustrative of the trade-off rather than a race-day prescription.

Why the bike leg matters so much overall

One more piece of context helps explain why this trade-off is worth thinking about carefully: across 16,611 race records from 163 professional Ironman 70.3 races, cycling was the strongest single-split predictor of overall finishing time (r = 0.85, R² = 0.8) Cycling is the most important predictive split discipline in professional Ironman 70.3 triathletes. That correlation is drawn from professional racing, so it supports the general point that the bike leg carries outsized weight in your final result — it does not tell an age-grouper what specific power number to ride at. Professional pacing dynamics, fitness levels and course tactics differ enough from age-group racing that this figure should be read as “the bike leg matters enormously,” not as a target to copy.

What “VI 1.02–1.05” really means — heuristic, not proven threshold

Pulling this together honestly: the direction of the evidence is fairly consistent across several independent studies — surging, variable power output tends to cost you time on the run relative to a smoother effort at the same average power. But the studies behind that conclusion are small (12, 8 and 38 participants across the cited trials), skew male and well-trained rather than mid-pack age-group, and use short runs (5–9.3 km) after relatively short bike efforts (20 km to one hour) rather than a half or full marathon off a 90 or 180 km bike leg. The 2024 intensity study shows the effect is outcome-dependent and sex-dependent, which is a good reason to treat this as real nuance rather than a new blanket rule. And the specific VI numbers — “keep it under 1.05,” “1.02–1.05 is well executed” — come from coaching experience and pacing-software design, not from a trial that tested different VI targets against each other in age-group athletes. Use them as a sensible practical range to aim for, understanding that nobody has proven 1.06 is meaningfully worse than 1.04 for your specific race.

Practical application

Start by finding out where you currently sit. Most bike computers and head units that record power will show normalised power (NP) and average power for a ride; dividing NP by average power gives you your Variability Index for that ride. Pull this from your last few long rides and races before changing anything — you can’t manage what you haven’t measured. If your long-ride VI is regularly sitting well above 1.10, that’s a reasonable signal that your pacing is surge-heavy rather than that anything specific is “wrong.”

Then decide what you’re actually optimising for on race day, because the 2024 finding above means this is a real choice, not a default. If the run is your strongest relative discipline — the leg where you make up ground on competitors — biasing toward a smoother, slightly more conservative bike effort to protect that run split is a defensible strategy. If the bike is where your relative strength lies, or if total finishing time (not splits) is genuinely all that matters to you, a slightly higher and still-controlled bike effort may be net faster across the full day, even though it costs you a bit on the run in isolation, mirroring the 90%-FTP finding in the study above. Neither choice is “wrong”; they're different objectives.

In training, treat pacing discipline as a rehearsed skill rather than a race-day intention. Practice holding effort by feel and by power on hilly and windy routes so that a controlled, slightly-harder-uphill, slightly-easier-on-the-flat pattern becomes automatic rather than something you have to think through mid-race. Pay particular attention to the final 20–30 minutes of long training rides: given that late-bike power spikes were specifically flagged as detrimental to the following run, that’s the exact window to practice riding controlled rather than accelerating to “finish the bike strong.” Brick sessions — bike straight into a run — are the best place to feel this trade-off directly and to test how a given VI on the bike translates into how your legs feel in the first kilometre of the run.

Common mistakes

• Chasing an average-watts target while ignoring how that power was actually distributed across the ride — two rides can share an average and have very different physiological costs.
• Surging hard on every climb and then coasting on the descent to “balance it out” — this pattern raises VI even when the average power looks exactly on target.
• Attacking in the last 10–20 km of the bike to bank time before T2 — the evidence specifically flags late-bike power increases as detrimental to the run that follows Factors influencing pacing in triathlon, PMC review.
• Treating VI 1.05 as a hard scientific rule rather than a coaching heuristic, or panicking mid-race over a single high VI reading on a genuinely hilly or gusty course.
• Pacing off an outdated or untested FTP number — a power target built on stale data is not a safe or meaningful target regardless of how smoothly you hold it.
• Copying professional pacing or split-correlation data onto an age-group race day — the finding that cycling strongly predicts finishing time in Ironman 70.3 professionals Cycling is the most important predictive split discipline in professional Ironman 70.3 triathletes says the bike leg matters, not that you should ride at a pro's power numbers.

How to apply this week

• Pull the power file from your last long ride and calculate VI (normalised power ÷ average power) to see your current baseline.
• Do one long ride with a deliberate “smoothness” focus: aim to keep VI at or below roughly 1.05 without treating it as a hard rule, and notice how the terrain forces natural variation.
• Run a brick session where you specifically hold the final 20–30 minutes of the bike controlled rather than surging toward the finish, then notice how the first kilometre of the run feels compared with a session where you didn't.
• Decide, in writing, what your next A-race priority actually is — protecting your run split or minimising total time — so the choice is made in training, not improvised mid-race.
• If you're pacing off an FTP number that's more than 8–12 weeks old, schedule a retest before you build a race pacing plan around it.

References

• Cycling attributes that enhance running performance after the cycle section in triathlon (International Journal of Sports Physiology and Performance, 2013)
• Factors influencing pacing in triathlon (PMC review)
• Cycling Intensity Effect on Running Plus Cycling Performance among Triathletes (International Journal of Sports Medicine, 2024)
• Pacing strategies during the swim, cycle and run disciplines of sprint, Olympic and half-Ironman triathlons (European Journal of Applied Physiology, 2015)
• Cycling is the most important predictive split discipline in professional Ironman 70.3 triathletes
• Effects of Cycling on Subsequent Running Performance, Stride Length and Muscle Oxygen Saturation in Triathletes (PMC)
• Optimal Bike Pacing and Best Bike Split with Ryan Cooper (Scientific Triathlon)
• Variability Index (VI) in Cycling and Triathlon (Best Bike Split)