Why am I not getting faster?

Before you rebuild your programme, check whether the improvement you're looking for is even large enough for your stopwatch to see.

A plateau usually gets diagnosed before it gets verified. The athlete decides the problem is their squat, or their block clearance, or the fact that they haven't done enough plyos this year.

The coach usually decides something similar. Everyone starts at "what should I change?" and skips the step before it: are we actually sure nothing is changing?

That sounds like a dodge. It isn't. It's the most useful thing you can establish before you touch the programme, and almost nobody does it.

The research and the track tell you two different stories

If you read the intervention literature, sprint speed looks pretty trainable. Seitz and colleagues pooled 15 studies and found that increases in lower body strength transferred to sprint performance with an average improvement of about 3.1%.

If you took that at face value, a well-programmed year should knock a chunk off your 100m.

Now look at what actually happens to sprinters. In their 2019 review of elite sprint development, Haugen and colleagues report that for world top 100 sprinters in their early 20s, mean annual improvements sit in the range of 0.1 to 0.2%. Over a 10.50 second 100m, 0.2% is 0.02 seconds.

They go further: for elite athletes, most annual within-athlete performance differences are smaller than the typical variation of the measurement, smaller than the smallest worthwhile change, and smaller than the influence of external conditions like wind, temperature, altitude and timing method.

Those two pictures look contradictory. They aren't.

Haugen offers two explanations for the mismatch, and I think both are correct. The first is publication bias in favour of positive findings.

The second, and the bigger one, is subject training status bias: most experimental data comes from untrained or moderately trained subjects.

A 3% improvement in a group of PE students who had never sprinted properly before tells you almost nothing about what happens to someone in year six of structured speed work.

So the first honest answer to "why am I not getting faster" is often: you are, at a rate you were always going to be improving at, and it's a rate you have no ability to detect.

Your measurement is noisier than your adaptation

Haugen and Buchheit's monitoring review is the reference here, and the numbers are sobering. In a well-trained young footballer, the typical error was 1.6% for 10m time and 2.9% for maximal sprinting speed.

The smallest worthwhile change for team sport athletes works out around 1.5% at 5m and roughly 1% from 10 to 40m.

Read that again. Your noise is larger than the change you're trying to detect. And that's before conditions get involved.

Different combinations of start procedure and triggering device can produce time differences many times larger than the performance change caused by years of conditioning.

Wind, altitude, temperature, barometric pressure and humidity can each shift short sprint times. Manual timing and single-beam gates are effectively useless over 10 to 20m because the absolute errors are so large.

Here's a test you can run this month, and I'd run it before your next testing block. Take a flying 10m. Same runway, same build-up, same timing setup, same warm-up, same spikes, three separate sessions inside a fortnight, three reps each.

Take your best from each day.

The spread between those three days is your noise floor. It's not inconsistency, it's the resolution limit of your testing. Any change smaller than that spread is a change you cannot claim.

If your noise floor is 0.04 and your athlete "went from 1.02 to 1.00," you don't have an improvement. You have a good day inside your normal spread.

One data point above a noisy baseline is a coin flip. Four points trending in one direction across a season is evidence.

But some of you genuinely aren't getting faster

Measurement noise explains a lot of perceived plateaus. It doesn't explain all of them. And when there's a real plateau in a trained athlete, the cause is usually specificity rather than volume, effort or programme design.

Go back to the mechanics. Weyand and colleagues tested 33 runners spanning top speeds from 6.2 to 11.1 m/s, and found that faster runners weren't repositioning their limbs any quicker.

Swing time didn't differ meaningfully across the whole range. What differed was the mass-specific force they applied to the ground during a contact lasting roughly a tenth of a second.

That's the variable. Not leg turnover, not effort, not how hard you feel like you're pushing. Force into the ground, in about 0.10 seconds.

Now ask yourself how much of your week is spent in a state where that variable is actually being loaded.

Because the honest answer for most athletes is: not much. Haugen's group recommends a polarised approach where sprinting intensity is either at or above 95% of maximal velocity, or below 70%.

Above 95% you're training the thing. Below 70% you're recovering so you can train the thing. The zone in between is where an enormous amount of amateur speed work lives, and it does neither job well.

Count it honestly.

In a normal week, how many reps does your athlete run at 95% or above, on fresh legs, with full recovery, without accumulated fatigue from the session before?

For a lot of people the number is one or two. For some it's zero, because everything is done tired and nothing reaches the intensity that drives the adaptation.

Recovery is the other half of that.

Intensive sprint sessions need at least 48 hours, and a genuinely maximal session often needs 72. Three "speed days" stacked on consecutive days doesn't give you three speed sessions.

It gives you one, followed by two sessions of tempo running that you've mislabelled.

The proxies move before the times do

If your 100m time can't resolve a year of training, monitor things that can. Flying 10 or flying 20 splits under fixed conditions.

Ground contact time at max velocity. Jump testing. Your strength markers. They have their own error, but they're closer to the mechanism and they move faster than a race result.

They also protect you from the worst failure mode in coaching, which is abandoning a programme that's working because a noisy number went the wrong way once.

So, to sum up: