Sam Millan

Race · Sunday 27 September 2026 · 42.195 km

Berlin

I coach people to run to a number, so mine goes up before the race rather than after it. Everything below was set on 16 August and has not been touched since.

The target

3:08to 3:104:28 to 4:30 per km

The floor is 3:15, which is 4:37 per km. If the day goes wrong, easing to 5:00 per km still finishes around 3:15. Walking costs far more than easing off.

Marathon best going in is 3:29. So the target is a nineteen minute personal best, and I would rather write that down in public than round it off afterwards.

Where the number came from

Two of my own runs, not a race predictor. The Battersea half in July came out at 1:28:56, and the standard conversion off a half puts that at 3:05 for a marathon. Then an 8 km at 4:32 per km in mid August ran with 21 beats of heart rate still in hand, which said the pace was aerobic rather than borrowed.

3:05 is what the conversion says. 3:08 to 3:10 is what I am actually going to run, because the conversion assumes marathon endurance I have not proved yet. My longest run before this block was 20 km.

Sydney Marathon
Aug 2025 · Rolling course
3:29
Athens Marathon
Nov 2025 · Very hilly
3:41
Battersea Half
18 Jul 2026 · The run the target is built on
1:28:56

How I intend to run it

One rule sits above all of this. Nothing quicker than 4:28 per km before 32 km, whatever it feels like at the time.

0 to 5 km4:35 to 4:38

Deliberately slow. The start is congested and every minute banked here is paid back double after 32 km.

Heart rate under 150

5 to 21 km4:30

Settle. Comfortable at 20 km is the trap, not the signal.

Heart rate under 160

21 to 32 km4:30

Hold it. Nothing changes here.

Heart rate ceiling 162

32 to 42 km4:25, or hold 4:30

Only lift if the heart rate is still under 165 and the legs are there.

Heart rate under 165 to lift

Fuelling is a gel every 30 minutes from 40 minutes in, six or seven of them, water at every station. The gels are not optional at 20 km when everything feels fine. That is exactly when the problem at 35 km gets built.

What could still change it

The target is confirmed or adjusted by one session, a 32 km on 6 September with the last 12 at race pace. If that holds, the number stands. If it drifts past 4:35 in the race-pace block, the race opens slower and 3:11 or 3:12 is the honest read.

It does not get raised in race week. Feeling fresh during a taper is the taper working, not new fitness arriving, and that is the most common way a good build turns into a bad race.

This is the same thing I do with the people I coach. The plan is a hypothesis, the sessions are the test, and the number moves when the data says so rather than when the mood does. That is the coaching →

Result

Empty until 27 September. Whatever the clock says goes here, including if it goes badly. A target published in advance is only worth anything if the miss gets published with the same weight as the hit.

The live training feed →|What I am focused on →|Home →

mainTypeScript2026-09-13