You open Pacewright. It says “Easy Run — 40 minutes, RPE 3-4.” Tomorrow it’ll say something different. Next week the whole pattern shifts.
This isn’t random, and it isn’t pulled from a fixed template. Every day, the engine first works out what’s safe and appropriate for you right now, then chooses the workout that best moves you toward your goal within those limits. Here’s exactly how it works.
Layer 1: What Week Is It?
The first question is structural: where are you in the training cycle?
Build weeks are when fitness develops. Volume and intensity increase according to your plan’s periodization pattern — a run of build weeks, then a week that backs off. Beginners cycle back sooner than intermediate and advanced runners, because they need the consolidation more often.
Recovery weeks are when adaptation happens. Volume drops well below your normal load. Some intensity is maintained — a shorter tempo or a reduced interval session — because complete deload isn’t necessary or optimal. The purpose is fatigue reduction, not fitness reduction.
Taper weeks happen before races. Volume comes down substantially over the last week or two, while your intensity and the number of days you run stay where they are. Your fitness doesn’t disappear in 1-2 weeks, but accumulated fatigue does. That’s the whole point of tapering.
Race week is race week. Everything is oriented around performing on race day.
Layer 2: Are You Recovered?
Before prescribing anything, the algorithm checks whether your body has had enough time since your last hard effort.
| Last Workout Type | Hours Before Next Hard Session |
|---|---|
| Easy run | 0-24 hours |
| Long run | 48-72 hours |
| Tempo / Threshold | 48 hours |
| VO2max intervals | 48-72 hours |
| Fartlek | 24-48 hours |
If you did a VO2max session yesterday, today is not a tempo run. Recovery is a hard limit, so a second hard day in a row is ruled out before the engine even compares options. The hard session moves or becomes an easy day.
This is one of the most common modifications the algorithm makes, and one of the most important. Back-to-back hard days give your musculoskeletal system less recovery than it needs, and it recovers more slowly than your cardiovascular system does. Your legs might feel ready. Your tendons might not be.
Layer 3: Is It Safe?
Three hard safety limits bound every option the engine considers, and your own feedback sits on top of them.
Spike guardrail. Would this run jump far beyond your longest run in the past 30 days? If yes, it gets capped. Even if your weekly volume is fine, a single abnormally long run creates disproportionate injury risk — and this is the limit with the strongest runner evidence behind it. In the largest study of runners to date, covering 5,205 runners and 588,071 recorded sessions, a session that exceeded the runner’s longest run of the previous 30 days by more than 10% was associated with overuse injury, in every band of excess they measured.2
Long run limit. Would your long run take too large a share of your weekly volume? The cap is goal-aware: marathon and ultra training earns the long run the biggest slice of the week (there, the long run is the workout), while PT test preparation earns it the smallest, because short-distance speed matters more than long aerobic capacity. If your proposed long run exceeds the cap for your goal, either it gets shorter or other runs in the week need to increase first.
Volume increase check. Would this week’s total volume climb faster than your current mileage allows? The limit scales to where you are: the more you’re already running, the smaller the share you can add on top.
Any option that fails a check is removed from consideration, and you see exactly which limit applied and why.
Your own reports sit alongside all three. If you log a run as harder than it should have been, or report soreness, that changes what the engine is willing to prescribe next — regardless of what the limits alone would permit. The numbers set the boundaries; what you tell the app can pull them in.
Layer 4: Environmental Conditions
If weather data is available, pace targets are adjusted before the workout is presented:
- Heat: Based on dew point, adjustments range from 0% to 9% slower across six tiers. The adjustment shrinks as you acclimatize — Pacewright tracks your recent runs in heat and reduces the bump once you’ve accumulated enough exposure for your body to adapt.
- Altitude: 1.5% slower per 1,000 feet above 3,000 feet
These adjustments ensure that your effort targets remain accurate regardless of conditions. A “Tempo at RPE 6-7” stays the same effort whether it’s 55°F or 85°F — but the pace that produces that effort changes.
Layer 5: The Best Workout
Now the engine has a set of options that are all safe and appropriate for today. From that set it chooses the one that scores highest for your goal: the workout that best builds the kind of fitness your goal needs (your goal-weighted Run Fitness Index), balanced against freshness, variety, and how close you are to a race. This is the optimization step, and it’s where Pacewright differs from a fixed plan. Two runners with the same safety limits but different goals get different “best” workouts on the same day.
Each workout type exists for a specific physiological reason:
- Easy runs build aerobic base and promote recovery. RPE 3-4. This is where most of your running should live.
- Long runs build endurance and teach your body to burn fat efficiently. RPE 4-5. Longer than other runs but slower than everything except recovery.
- Tempo runs improve lactate threshold — the pace you can sustain for roughly an hour. RPE 6-7. Comfortably hard.
- Intervals improve VO2max and running economy. RPE 8-9. Short bursts of hard effort with recovery between.
- Recovery runs are deliberately easy active recovery. RPE 2-3. Shorter and slower than easy runs.
The mix of these workout types follows the 80/20 principle — roughly 80% of your running at easy effort, 20% at moderate or hard effort.
Why a Transparent Optimizer, Not a Black Box
Pacewright optimizes, but it’s not a black box. Every decision follows from a model built on published training science (Daniels, Pfitzinger, and the load-management research), and every choice can be traced back to it.
This is a deliberate contrast with opaque AI. A black-box model produces an output that works but can’t tell you why. Pacewright can show you exactly what bounded the decision and what won: which safety limits applied, which option scored highest, and what would need to change for a different answer. Where the engine adapts to you over time, it does so on a documented model you can inspect, never an opaque one. That transparency is the entire point.
When Pacewright says “Easy Run, 40 minutes, RPE 3-4,” it can show you the full chain: you’re in a build week, you ran intervals yesterday so a hard day is ruled out (a safety limit), today’s distance sits comfortably inside your longest recent run, and among the allowed easy options this one best fits your week’s needs. Every link is visible. Nothing is hidden.