Why your training plan just changed (and why that's the point)

It's Wednesday. The plan says 4 × 8 minutes at threshold. You slept badly twice, a work deadline ate Monday's swim, and your legs feel like someone let the air out of them. Do you do the session anyway, move it to Friday, or write the week off?

Every self-coached athlete has stood in that exact spot, and the plan on the page cannot help, because it was written before any of this happened. That is the argument for an adaptive training plan in one sentence: a plan is a forecast, and forecasts age. The one you wrote in January made assumptions about a week in July — your fitness then, your free evenings, how the intervening six months would go — and a decent number of them are already wrong.

What follows is the checklist: what a plan should react to during a season, what a good response looks like in each case, and the one change a plan should never make by itself. None of it requires software. It is what a coach does on a Sunday evening with your week in front of them, written down so you can do it yourself.

The parts of a plan that age well, and the parts that don't

A season plan has two layers, and they go stale at completely different rates.

The top layer is the shape: race date, how many weeks you have, roughly where the hardest block sits, when the taper starts, which weakness you are spending the season on. That layer is built backward from race day, and it should barely move. If you are rewriting it every three weeks you don't have a plan, you have a mood.

The bottom layer is the detail: which session lands on which Wednesday, how many intervals, how long the long ride is. Ten weeks out, that detail is a guess dressed up as a schedule. It was generated from the fitness you had the day the plan was written and a calendar that assumed nothing would happen.

Most self-coached seasons go wrong by mixing the two layers up, in both directions at once. Athletes defend Wednesday's intervals as though the season depends on them, then abandon the season shape entirely the first time they catch a cold. It is exactly backward. The shape is the thing you defend. The details are supposed to move.

That split is why PaceBeats builds the season plan backward from your race date and then generates each week separately — from your recent training, the hours you actually have, and how much fatigue you are carrying — instead of stamping out a template twelve weeks in advance. The skeleton stays put. The week gets written last, when there is real information to write it from.

What a plan should react to

Six triggers, and the difference between handling them well and handling them badly.

The rest of this article takes the first four one at a time, because they are the ones that come up every month.

Trigger 1: you missed sessions

The most common reason a week gets rewritten, and the one most athletes handle worst.

The instinct is debt. Two sessions missed, two sessions owed, stack them onto the weekend. That instinct is wrong on the arithmetic before it is wrong on the physiology. Fitness in every model worth using is a rolling average over roughly six weeks, so one missed session moves it by well under a point. What actually hurts you is the repayment: three hard days in a row that you didn't build up to, tacked onto the end of a week that was already compromised by whatever caused the misses in the first place.

A good plan treats the days you have left as a fresh, smaller week. It asks what the block was for, keeps the one session that carries that purpose, lets the rest go, and does not make Saturday pay for Tuesday. When PaceBeats rebalances a week around missed sessions, that is the rule it follows — the remaining days get re-planned, not loaded up.

One exception worth naming: a pattern of misses is a completely different signal from a one-off. Missing your Tuesday run four weeks running is not a scheduling accident, it is your calendar telling you that Tuesday evenings don't exist. That should change the plan permanently rather than get rebalanced away every seven days. There is more on how to triage a missed week in what to do when you miss workouts.

Trigger 2: the load is catching up with you

The second trigger is the one you are least likely to act on yourself, because acting on it means doing less when you feel like you should be doing more.

The measurable version is ramp rate: how fast your accumulated load is climbing week over week. Three to five points of fitness per week is a build most athletes can absorb. Sustained climbs above eight points a week are where illness and injury cluster in practice. That is accumulated coaching convention rather than a number from a controlled trial, but it is consistent enough to plan around, and it is the kind of thing software notices before you do. Training load, explained covers how those numbers are built and where they lie.

A good plan responds by taking load out before the wheels come off, not after. That is what PaceBeats does when accumulated load says you are digging a hole — the coming week gets smaller, and the reason is attached to the change.

Here is the honest limit, and it matters more than the feature. A plan built on training data can only see training. It cannot see that you slept five hours a night for a week, that work is on fire, or that you are coming down with something. Those things cost you real recovery capacity and appear nowhere in the model. The numbers are a rear-view mirror with a wide-angle lens: good at telling you a three-week ramp has been too steep, useless at telling you whether today should be intervals or coffee.

So the load trigger works in both directions, and only one of them is automatic. Software can pull load out when the arithmetic says to. Pulling load out because you feel wrong is still your call, and you should make it earlier than feels comfortable. The early signs of overtraining are worth knowing by heart for precisely that reason.

Trigger 3: your week changed shape

A work trip, a five-day conference, childcare falling through: your training week just went from ten hours to four.

The bad answer is to keep the plan and squeeze it in, which produces a week of rushed sessions and a Sunday of guilt. The second-worst is to write the week off entirely, which turns four available hours into zero.

The right answer is to rebuild the week around the hours you genuinely have and to be honest that a four-hour week is a four-hour week — not a ten-hour week performed quickly. Decide what the block was for, keep the session that serves it, make the rest maintenance, and accept a lower total. A four-hour week inside a twenty-week build costs you almost nothing. Four hours of panic and a strained calf cost you three weeks.

This is the one place where saying so in advance genuinely pays. A week built from four available hours is a coherent four-hour week; a ten-hour week you dismantle on Wednesday is neither. Declare the constraint on Sunday, whether you are telling software or a spreadsheet. How many hours a week triathlon training actually takes is a useful reality check on what a given number of hours can and cannot buy.

Trigger 4: you had a strong week — and nothing should escalate on its own

This is the trigger where most adaptive systems, and most enthusiastic self-coached athletes, get it wrong.

You completed everything. The numbers looked good. The obvious inference is that you can take more, so next week should be bigger. Plenty of software makes that inference automatically, and it is the wrong default for three reasons.

The metric cannot tell how the week felt. Completion and load look identical whether you floated through the sessions or gritted your teeth through every one of them. A week you barely survived and a week you sailed through produce the same green ticks.

A strong week is often the week before the crash. Feeling great is what a well-absorbed block feels like, and it is also what the top of a ramp feels like right before you get sick. The data alone cannot separate those two states, and the cost of guessing wrong is asymmetric: an easy week you didn't need costs you a few days, an overreach you didn't need costs you three weeks.

Escalation compounds silently. If every good week raises the next one and no week ever lowers it, the ramp only points one way. Add five percent enough times without anyone deciding to, and you arrive somewhere nobody chose.

So the default should be: progress is offered, not imposed. PaceBeats will tell you the week went well and offer a step up. It does not raise your load behind your back. You are the one who knows whether that strong week felt strong or merely looked strong, and that information exists only in your head.

The same rule works without software. When a block goes well, write down the increase you are considering and why, then decide it deliberately on Sunday. Progression you chose is progression you can also un-choose.

Every change should come with a why

A change you don't understand is a change you won't trust, and a change you don't trust is one you will quietly ignore. That is the whole difference between a plan that adapts and a plan that is merely unstable.

A useful explanation has three parts, and any of them missing makes the change harder to follow:

• What changed. Thursday's threshold run became 40 minutes easy.
• What triggered it. You missed Monday and Tuesday, and the week's remaining days can't carry the original load without three hard days in a row.
• What it costs. This week's total drops by about 15 percent; the block's key session on Saturday is unchanged.

Without the trigger, adaptation is indistinguishable from randomness. Without the cost, you cannot judge whether to accept it. The reason does not have to be right every time — it has to be readable, because a reason you can read is a reason you can argue with.

If you plan your own season, the cheapest version is a one-line note in your log every time you deviate: what you changed and why. Three months later it is the only record of what actually happened.

What an algorithm cannot know

Being straight about the limits is the point of this section, not a disclaimer at the bottom of it.

An adaptive plan sees your completed sessions, your stated availability, and the load arithmetic those produce. That is a genuinely useful view and it is a narrow one. It cannot feel your legs. It cannot tell the difference between a niggle that needs two easy days and an injury that needs a physio. It cannot see the argument you had, the flight you are about to take, or the fact that you have been living on four hours of sleep. It does not know that the bad session on Tuesday was a bad burrito rather than a fitness problem.

The measuring stick has its own wobble too. Session load is priced against your threshold, and if that threshold is stale, every number downstream is off — which is worth reading up on if your zones came straight off a watch, in what your watch's FTP and threshold estimates are really telling you.

So the override belongs to you, permanently and without argument. Every session in PaceBeats can be edited or replaced by hand, and the plan takes what you actually did as the new truth. That is not a safety valve bolted on the side. It is the correct division of labor: the algorithm is better than you at arithmetic over six weeks of data and at noticing a ramp that has been too steep since June. You are better than it at everything happening in your body today.

Back to Wednesday

The plan says 4 × 8 at threshold. You slept badly and missed Monday.

Work through it in order. Is the shape of the season still right? Almost certainly yes — one bad week doesn't move race day. Is this the key session of the block, or a supporting one? If it is supporting, replacing it with 40 easy minutes costs you close to nothing. Have you missed enough this week that the remaining days can't carry the load without three hard days in a row? Then make the week smaller and stop trying to repay it. And if this is the fourth Wednesday in a row you have had this conversation with yourself, the answer isn't about today's session at all — it's that Wednesday needs to change permanently.

That is the work. A plan that changes when the facts change isn't broken discipline, it is the plan doing its job. The one that never moves isn't strict. It is just no longer describing your life.