The rules behind the plan
Every number in the outline comes from the protocol I give first-time marathoners. The full version, with the workouts and race day, is my first marathon guide.
- Give yourself 18 to 20 weeks from a base of about 15 miles a week. Sixteen weeks is a little shorter, so you want some running behind you heading in. Twelve is too aggressive for a first marathon.
- Never raise weekly mileage more than 10% over the week before, and take a cutback every fourth week, dropping volume by 15 to 20%.
- Build the long run 1 to 2 miles at a time, not 4, and hold it flat now and then. It tops out around 20 miles.
- Run easy days genuinely easy. If your easy pace creeps up to moderate, workouts collapse and long runs hurt for the wrong reasons.
- One threshold session and one quality long run a week, plus strides twice. That's the whole recipe for a first marathon.
- Practice race fueling on every long run over 90 minutes, and lift twice a week for 20 minutes.
- Taper three weeks, dropping volume and keeping a little intensity, then run the first 10 miles slower than feels right.
How many days a week
Three days works: the long run, the threshold session once the build starts, and one easy run with strides. Four is better, and it's what my first-marathon protocol assumes: add a second easy run. Five or six days spread the same weekly miles across more runs, so each one is shorter and easier to recover from. Whatever the number, the long run and the threshold day are the two you protect.
If you're using ChatGPT for the plan
I've had a few athletes come to me with a plan ChatGPT wrote. The bones are usually fine. It knows the textbook: build the long run gradually, mostly easy miles with a couple of quality days, a cutback week every few weeks, a taper at the end. If you squint, it looks like a real plan. What's missing is the athlete. The paces are usually built off the goal time, not current fitness, so every workout is a little too fast and people end up forcing it instead of building.
If you're going to use it anyway, give it real inputs, not just a goal: your current weekly mileage, your longest recent run, a recent race or time trial, how many days a week you can actually run, your injury history and your race date. Then ask it to:
- Set paces from your recent race, not your goal (or run that race through the pace calculator yourself).
- Keep weekly mileage increases small, with a cutback week every 3 to 4 weeks.
- Explain the purpose of each workout, so you know what you're skipping if you miss one.
- Give you a plan B for a missed or sick week.
Never trust it on pain or injury. It can't tell shin splints from a stress reaction, and "should I run through this" is a question for a PT or doctor. And don't trust it to tell you to back off. It tends to go along with whatever you ask, so if you ask for an aggressive plan, you'll get one. The plan is maybe 30% of coaching. The other 70% is adjusting it when real life hits. More on that in a coach versus an app or an AI plan.
Where this fits
If you have more than 20 weeks, the outline starts 20 weeks out; run easy three or four days a week until then. If you're coming from under 15 miles a week, it adds a few weeks of easy running first. For a runner with a solid base who wants more structure, the 20-week marathon plan goes deeper, and the three-week taper covers the end of the block. Once you have a goal time, the pace calculator sets your easy and threshold paces.