Demi Vollering won the Tour de France Femmes for the second time on Sunday, and if you read this site’s preview of Stage 9, you’d have been told — correctly — that the fourth ascent of the Col d’Èze was where the race would be decided. It was, sort of. Vollering attacked with roughly 16km still to race, about 600 meters below the summit of that steeper fourth climb, and dropped Kasia Niewiadoma-Phinney on the spot. But the climb only bought her a handful of seconds. The race was actually won on what came next: 15 technical kilometers down into Nice, where an 8-second overnight lead turned into a 1:18 winning margin. Most of that gap opened while she was going downhill, not up.

That’s worth sitting with, because it’s a gap in what we’ve written about this race so far — and, honestly, in what most amateur training plans cover at all.

Quick Verdict: Descending as a Race-Winning Skill

What Happened	Detail
Where she attacked	~600m below the summit of the 4th Col d’Èze ascent, ~16km to go
Who she dropped	Kasia Niewiadoma-Phinney, defending the yellow jersey
Overnight GC gap	8 seconds, in Vollering’s favor
Where the gap actually grew	The technical descent and windswept run-in to Nice — not the climb itself
Final margin	1:18 over Niewiadoma-Phinney
Final podium	1. Vollering 2. Niewiadoma-Phinney (+1:18) 3. Longo Borghini (+4:29)

The lesson: The climb created the opportunity. The descent is what turned it into 1:18 and a second Tour title — the first repeat win in this race’s history.

What Actually Happened on the Final Climb

Our Stage 9 preview called the shape of the stage right: three honest laps of the Col d’Èze’s standard face, then a fourth ascent via the steeper Vinaigrier road, where the climbing actually happens. That part held up. Niewiadoma-Phinney came in defending an 8-second lead that Vollering had built the day before with a separate, anaerobic late attack into Nice — the kind of race-deciding move we wrote about training specifically because it’s a different physical skill than steady climbing.

On the fourth ascent, Vollering went again. Not at the base, not on the steepest pitch — with the summit already in sight, she accelerated and Niewiadoma-Phinney simply couldn’t follow. That’s a real gap and a real result. But by most accounts she crested the Col d’Èze with something like 15-20 seconds in hand, not the minute-plus that eventually separated the two riders at the finish. The stage — and the Tour — was decided on the way down.

The Descent Nobody Trained For

Here’s the part that should bother anyone who’s been following this series. We’ve now written about four distinct physical skills this race demanded: holding one honest number for an hour on Ventoux, staying switched on through a boring middle for Nice’s punchy late climbs, rationing effort across four laps of the same hill, and producing a hard anaerobic attack on legs that had no business having anything left. Every one of those is trainable with power targets, RPE, and a structured interval set. We wrote the workouts.

Descending doesn’t have a power target. That’s exactly why it gets skipped in most training plans — including, if we’re honest, most of what we’ve published about this race. It’s treated as a talent you either have or don’t, or worse, as a reward for being strong enough on the climb to get a gap in the first place. Vollering’s Stage 9 win argues the opposite. The climb gave her fifteen seconds. The descent gave her the Tour.

What makes descending trainable rather than just a talent? Descending speed comes from body position, braking timing, and line choice — all mechanical skills that improve with repetition, not raw fitness. A rider who brakes before the corner instead of during it, looks through the turn, and weights the outside pedal will consistently descend faster than a fitter rider relying on nerve alone.

We touched on the mental side of this after Nice’s Stage 8 finale — staying focused on a technical run-in when your legs are cooked. That’s still true and still matters. What we didn’t cover is that descending well isn’t just about not making mistakes when tired. At the level Vollering just rode it, it’s an offensive weapon — a way to turn a small gap into an unassailable one, on purpose.

How Do You Train to Descend as an Attacking Skill, Not Just a Survival One?

If you’re chasing time on club descents, trying to hold a group’s wheel downhill, or just tired of watching riders you can out-climb disappear from you on the way back down, training this comes down to five things:

1. Ride the same descent repeatedly, not different ones. Cycling Weekly’s descending guide is right that confidence comes from repetition on known terrain, not variety. Pick one local descent and ride it enough times that you know every camber change and blind corner before you try to go fast on unfamiliar roads.
2. Separate braking from cornering. Most of your braking should happen before you lean the bike, not during the turn. CTS’s cornering breakdown puts it plainly: braking mid-corner reduces the traction you have available exactly when you need it most. Practice this in an empty car park with cones before you take it to a real hill.
3. Fix your body position on flat ground first. Hands in the drops, weight shifted back, elbows bent and loose. These are muscle-memory habits, not things you figure out for the first time at 60 km/h. Rehearse the position on a slow, safe road until it’s automatic.
4. Practice under fatigue, not just fresh. A descent ridden on fresh legs after a warm-up teaches different reflexes than one ridden after three hours of hard riding. If your event finishes downhill — and a lot of gran fondos and club routes do — at least some of your descending practice needs to happen tired.
5. Treat it as a discipline with its own session, not an afterthought. Nobody schedules “descending intervals.” That’s part of the problem. Once every couple of weeks, make a ride’s actual purpose the descent, not the climb that gets you there.

A Session to Train This

You can’t fully replicate 15km of Mediterranean switchbacks on most local terrain, but you can train the two things that decided Stage 9: producing a hard effort right as the road tips down, then descending with focus instead of just surviving it.

Warm-up: 15 minutes progressive, finishing with a few short accelerations to sharpen reflexes before the technical work.

Skills block (do this on a closed road, car park, or quiet cul-de-sac, not in traffic):

1. 10 minutes of slow cornering drills — cones or water bottles at right angles, riding the entry-wide, apex-tight, exit-wide line in both directions. Focus on weighting the outside pedal, not just steering with the bars.
2. 10 minutes of braking practice — approach a marked point at moderate speed, brake to a near-stop before a line, then release and carry speed through it. The goal is finishing your braking before you’d need to turn, every single time.

Main set, on an actual descent you know well:

1. 8-12 minutes at 90-100% FTP (RPE 7/10) on the climb leading into your chosen descent — simulates arriving at the top with real fatigue, not fresh legs.
2. The descent itself, ridden with full attention — hands in the drops on anything technical, eyes up and through the corners, braking before you lean rather than during. Repeat the same descent 3-4 times if you can loop back to the top.
3. Between reps: 10-15 minutes easy spin back to the base, using the recovery time to think through what worked on the previous run, not to zone out.

Cool-down: 10 minutes easy.

Total time: 90-120 minutes depending on descent length and how many reps you fit in.

The test: Time yourself on the descent, but don’t chase the fastest possible run on the first attempt. Compare your third or fourth rep to your first — if you’re faster and smoother later in the session, on legs that are more tired, that’s the skill actually building. If you’re slower and sloppier, you’re descending on nerve rather than technique, and technique is the thing this session is meant to fix.

If you haven’t tested your FTP recently, do it before setting the climbing portion of this session — you want genuine fatigue going into the descent, not an undercooked effort that leaves you descending on fresh legs and learning the wrong lesson.

Why This Got Missed in Our Own Coverage

Part of the reason descending doesn’t get trained is that it doesn’t show up in the data most of us actually look at. Power meters measure watts; they don’t measure line choice or braking points. A rider can nail every interval on Zwift for a winter and still get spat out the back of a group on the first fast, technical descent of spring — because nothing about structured indoor training touches the skill that just decided a Grand Tour.

That’s not an argument against power-based training. Everything in this series about pacing the Col d’Èze’s four laps or producing a late anaerobic attack still holds — those are real, trainable, decisive skills. It’s an argument for adding a skill that doesn’t fit neatly on a power curve, because pretending it doesn’t matter is exactly how a fit climber loses a stage — or a Tour — on the way back down.

The Bottom Line

Vollering’s attack on the Col d’Èze was real and it mattered. But fifteen seconds off the front of a small group with 16km still to race isn’t a winning margin on its own — 1:18 is. The distance between those two numbers is 15 technical kilometers of descending, ridden with more focus and better lines than anyone chasing her could manage.

If you’re strong enough to create a gap on the climbs you race but keep losing it — or failing to extend it — on the way down, the fix isn’t more threshold work. It’s treating descending like the trainable skill it is: repetition on known terrain, braking before the corner instead of in it, and practice on tired legs instead of only fresh ones. That’s how a small gap becomes the kind of margin nobody closes.

---

Stage 9 result and race narrative from ProCyclingUK’s coverage of Vollering’s Col d’Èze attack. Final GC standings from Cyclingnews. Descending and cornering technique referenced from Cycling Weekly and CTS. Training targets are estimates based on power-based training principles — adjust for your own tested FTP and conditions.