Ultra Race Post-Mortem: Read Your Data
Ultra Race Post-Mortem: What Your Data Reveals About Where You Left Time¶
You crossed the finish line. Maybe you hit your goal, maybe you missed it by two hours. Either way, sitting in the medical tent with a drop bag at your feet, a question starts forming: where did I actually lose the race?
The polite answer is "the climb at mile 62." The honest answer is usually three or four compounding mistakes spread across 20 hours, and your watch recorded every one of them.
This post is about reading that recording. Not to beat yourself up — to pattern-match the next one.
You'll learn:
- The four data signatures that show up in almost every ultra post-mortem
- How to separate real time loss from aid station math
- Why heart rate tells a different story than pace (and when to trust which)
- A 48-hour debrief framework before memory smooths the edges
Why Ultra Post-Mortems Are Different From Marathon Debriefs¶
A marathon post-mortem is usually a single question: did you positive-split, and by how much? The answer is right there in your 5K splits. The fix is pacing discipline and the long runs you didn't do.
An ultra isn't that clean. Twenty hours of data contains dozens of decisions — how long you spent at mile 32, whether you ate on the climb, the forty minutes you sat in a chair at mile 78 "just to collect yourself." Some of that is cost-of-doing-business. Some of it is the race.
The trick in an ultra debrief is separating the two. You don't want to flog yourself for a six-minute stop at a crew access point if the real damage was a cardiac-drift spiral between miles 40 and 55 that your watch flagged and you ignored.
Good post-race analysis answers three questions in order:
- Where did I actually lose time? (Data question)
- What caused it? (Pattern question)
- Was it avoidable, or the price of the day? (Honest question)
Most runners jump straight to question 3. That's how you end up blaming "bad legs" for a fueling problem you can see in your heart rate trace.
The Four Signatures to Look For¶
After you've stared at enough ultra files, four patterns show up again and again. Each one has a distinct shape in the data. Train yourself to recognize them, because your next race will usually fail for the same reason your last one did.
1. The Cardiac Drift Spiral¶
What it looks like: Pace stays flat or drops slightly. Heart rate climbs 10–20 beats over two or three hours — sometimes more in heat or deep into a 20+ hour race — despite no change in effort or terrain.
What it means: You're getting dehydrated, under-fueled, or both. Two separate mechanisms compound: dehydration drops your stroke volume (so your heart beats more times to move the same blood), and heat stress sends more blood to your skin for cooling (so more beats get dedicated to thermoregulation instead of working muscle). Both push HR up at the same pace. By the time your legs feel it, the spiral is already well underway.
Some drift is normal in endurance events — research on cardiac drift confirms the thermoregulatory component happens even in well-hydrated athletes. A drift of 3–5% over a long run is just physiology. A drift of 10–15% or more over a few hours at constant pace is a problem, and in heat or at 20+ hours that gap can open further.
Where you'll see it in the file: Overlay your pace and heart rate traces. If they decouple — pace steady, HR climbing — that's the signature. The fix isn't "push harder." The fix happens 60 minutes earlier in the race, at the aid station you blew through.
2. The Positive Split Cliff¶
What it looks like: Your first 50K averages 11:30/mile. Your second 50K averages 15:00/mile. Your last 20K averages 19:00/mile on similar terrain.
What it means: You went out too fast. This is the most common ultra mistake and the most obvious in the data. Research on 100K racing shows the fastest finishers only slow about 15% across the race while slower runners drop off by 40% or more. The 100-mile curve is different in scale, but the shape holds.
Before you blame the pace drop, check the elevation profile for that quartile. A 4,000-foot climb in Q3 is supposed to look like a cliff. The signal you're looking for is a pace drop that isn't explained by terrain.
The tell isn't the slowdown itself — everyone slows. It's the shape of the slowdown on comparable ground. A gradual deceleration usually means fitness or fueling ran out. A cliff — fine one section, shattered the next, terrain-matched — almost always means you banked time you couldn't afford and the bill came due.
Where you'll see it in the file: Break the race into quartiles and compare average pace. If Q4 is more than 30% slower than Q1 on comparable terrain, you overpaced the front half. Pull the elevation profile alongside. If the quartile pace drops aren't explained by terrain, they're explained by your early splits.
3. The Aid Station Bleed¶
What it looks like: Your moving pace is actually fine. Your elapsed time is an hour and a half off plan.
What it means: You lost the race standing still. Ten minutes at mile 15, twelve at mile 28, twenty at mile 44 "getting it together," fifteen at the crew access, another ten because someone had grilled cheese. It adds up faster than most runners think.
This is the single most preventable source of time loss in a hundred-mile race, and the one most people refuse to look at because the cure feels unheroic. You didn't fail at running. You failed at leaving.
Where you'll see it in the file: Compare moving time to elapsed time. The gap is your aid station total. Then break it down by stop — your watch auto-pause data and your aid station timing chips will tell you exactly which ones ran long. The ugly truth is usually one or two catastrophic stops, not forty medium ones.
4. The Fueling Crash¶
What it looks like: Steady pace and heart rate through a long stretch, then a sudden pace collapse that doesn't match any terrain change. Often followed by a long aid station stop.
What it means: You ran out of calories. Not "felt a little low" — ran the tank to empty. Most runners execute a solid ultra on 200–300 calories per hour, and many strong 100-mile performers are closer to 250–350 per hour during the long middle stretch when moving slowly enough to absorb solid food. Under 150/hour for a 20-hour race is a crash waiting for a trigger.
The signature is distinct from cardiac drift. Drift is slow and compounding. A fueling crash is abrupt — often a single segment that's 20% slower than the ones surrounding it, with heart rate dropping instead of rising because you simply can't push anymore.
Where you'll see it in the file: Look for sharp, localized pace drops on sections without matching climbs. Because GPS can flake on technical trail, use aid-station-to-aid-station segment times rather than arbitrary 10K chunks — the chip data is clean where your watch might not be. Cross-reference against your fueling log. If you can't remember when you last ate, that's usually when.
How To Actually Run The Debrief¶
Here's the framework. Do it in this order, inside 48 hours, before the memory smooths.
Step 1: Pull the raw numbers (same day)¶
While it's fresh, get these data points written down somewhere you'll actually find them:
- Total elapsed time vs. moving time (the delta is your aid station total)
- Average pace per quartile (Q1, Q2, Q3, Q4)
- Elevation gain per quartile — without this, the pace numbers will lie to you
- Average heart rate per quartile
- Number of aid station stops over 5 minutes
- Your worst inter-aid-station segment (slowest, use chip splits not GPS)
- Your best inter-aid-station segment after the halfway mark
- Temperature range and approximate conditions for each quartile (day/night, heat, rain)
- A one-line subjective note: how cognitively functional were you at Q3 and Q4?
Don't interpret yet. Just list them. The subjective notes matter — sleep deprivation and cognitive fog at mile 75 shape every decision you made, and the data alone can't surface that.
Step 2: Lay pace and heart rate side by side (day 2)¶
This is the single most useful exercise in ultra analysis. Open your activity file and overlay the pace trace against the heart rate trace. Look for decoupling.
- Pace flat, HR rising = cardiac drift. Hydration/fueling problem.
- Pace dropping, HR flat = muscular fatigue or terrain. Fitness issue.
- Pace dropping, HR dropping = glycogen depletion. Fueling crash.
- Pace rising, HR rising = you're pushing. Was this section the move or the mistake?
One of these four is usually the race. NavRun's activity analytics plots HR and pace together and flags the decoupling point automatically — but a printout and a highlighter works too.
Step 3: Quartile the race (day 2)¶
Divide the course into four roughly equal distance chunks. For each, record average pace, average HR, elevation gain, and time spent stopped. The goal is a table that lets you see the shape of your race at a glance.
A well-executed ultra shows a modest pace slowdown (10–20%) across quartiles, rising heart rate (normal drift), and stable time-stopped. A blown race shows a pace cliff, decoupled HR, and one or two aid station black holes.
Step 4: Write three sentences (day 3)¶
By day three, the race is far enough away that you can see it. Write three sentences:
- The thing I did well.
- The thing that cost me the most (time, or intactness, or mental state — whichever your goal was).
- The thing I'll do differently next time.
Worth naming: not every ultra debrief is about going faster. If your goal was finishing intact, or getting your crew across the line with you still functional, or managing a chronic issue without blowing it up, then "the thing I'll do differently" might be "DNF earlier when the data says to" rather than "train more speed." Let the goal you actually ran for shape the conclusion.
Resist the urge to write an essay. If you can't compress the race into three sentences, you haven't understood it yet. Put the document somewhere you'll re-read it when you're planning the next one, because you will forget, and you will be tempted to repeat the mistake.
Patterns Most Runners Miss¶
A few things the data reveals that most self-debriefs skip over:
The "bad patch" wasn't always the bad patch. Runners fixate on the low point — "I was cooked at mile 70." Sometimes that's exactly right — a climb or the cumulative effort caught up. But often the data shows the damage happened 60–90 minutes earlier, somewhere that felt fine. In those cases the mile 70 collapse is the symptom; the cause is the decision you didn't flag at mile 60. Check both possibilities before committing to a story.
Your crew stop was longer than you remember. Nearly without exception. The "quick five minutes" at crew access was 14. The brain compresses stationary time.
Your night miles were often faster than you think. Many runners assume they slowed overnight, but the data frequently shows the overnight hours were their most consistent — heart rate low, pace steady, because it was too dark to see the climbs and push the descents. The slowdown often came at sunrise, not sunset. (The opposite happens too, if you're under-layered and fighting cold — which is why logging temperature by quartile matters.)
Your pacer probably helped in ways pace won't show. The obvious instinct is to compare miles 55–60 to miles 60–65 after your pacer joined. But the real contribution often shows up in moving-vs-elapsed time for that segment, not pace. A good pacer keeps you moving through the aid station where you'd have sat down for 20 minutes — that's 20 minutes of elapsed time saved at the same pace. Before deciding your pacer "didn't change anything," check whether your stop times dropped in the segments they ran with you.
Where NavRun Fits¶
This kind of analysis isn't exotic — you can do it with a free Strava account, a calculator, and an hour on a Monday night. The reason people don't is that the math is tedious and the tools aren't built for it.
NavRun is built for it. When you connect your Strava account, the activity detail view pulls your splits, heart rate, and pace into a single timeline. Analytics tracks your drift patterns across races, so you can see whether your mile-50 collapse is a one-time execution problem or a consistent training gap. The AI feedback feature reads your race file and flags the decoupling point, the quartile cliff, and the aid-station bleed in plain language.
It's not magic. It's a calculator and a highlighter that doesn't get bored. Core features are free — Pro adds the AI analysis and race-to-race pattern tracking.
See what your last race actually says →
Frequently Asked Questions¶
Q: How long after a race should I do the post-mortem?¶
Within 72 hours. Longer than that, memory smooths the edges — stops get shorter, bad patches get more dramatic, and the actual cause of the problem gets lost behind the story you tell about it. The data is objective; your memory isn't.
Q: Should I trust heart rate data in an ultra?¶
Trust the trends, not the absolute numbers. HR can be thrown off by temperature, caffeine, wrist strap slippage, and dehydration. But the shape of the HR curve — whether it drifts, decouples, or crashes — is reliable and tells you things pace alone can't.
Q: What if I had GPS dropouts or watch problems?¶
Work with what you have. Aid station timing chips give you clean segment times even if your watch flaked. Overlay chip splits with your watch trace and you can usually reconstruct the race accurately enough to find the pattern.
Q: My race was a DNF. Is a post-mortem still worth doing?¶
Especially then. A DNF has a precise moment — the decision to stop — and the 30–90 minutes before it usually contain the real story in the data. Was it cardiac drift, a fueling crash, or a mental spiral? The trace will tell you, and it'll shape whether you train differently, fuel differently, or build mental contingency plans for next time. (We've written more about when a DNF is the right call.)
Q: How is this different from analyzing a marathon?¶
Marathons have a narrower data surface, not a simpler one — good marathon debriefs still have to account for cardiac drift, glycogen depletion, hilly-course pacing variance, and aid station fumbling. Ultras just add aid station management as a dominant variable, fueling cadence over 10+ hours, and night/day temperature shifts on top of all that. There are more independent systems failing or succeeding at once in an ultra, which is why the debrief needs explicit structure instead of just "did I positive split."
Q: How do I compare two races fairly?¶
Quartile averages, not aggregate time. Two ultras with the same finish time can have completely different shapes — one runner executes evenly and one blows up twice — and the shape matters for what you train next. Race time tells you how fast you were. Quartile analysis tells you why.
Q: Should I do this for training runs too?¶
Yes, for your longest ones. A 30-mile training run with a cardiac drift spiral is a preview of race day. You'd rather see it in training than find out at mile 50 of the race.
Q: What if the data tells me something I don't want to hear?¶
That's the most useful outcome. A post-mortem that confirms what you already believe isn't doing any work. If you thought the climb killed you but the data shows you lost 40 minutes standing at aid stations, that's the one worth acting on — because aid station discipline is trainable and climbing fitness takes a year.
Key Takeaways¶
- Your ultra race file contains forensic evidence of where you lost time — splits, heart rate, and pace tell a story your memory won't.
- Four patterns show up in almost every post-mortem: cardiac drift, the positive split cliff, aid station bleed, and fueling crashes. Learn the signatures.
- Do the debrief within 72 hours. Pull quartile averages, overlay pace and HR, and write three sentences. Don't skip to conclusions.
- The "bad patch" is rarely where the damage was done. It's the symptom of a decision 60–90 minutes earlier.
The best ultrarunners aren't the ones who never have a bad race. They're the ones who debrief the bad races hard enough that the next one fails differently.
Start Running Smarter¶
Ready to see what your last race actually says about you?
NavRun connects to your Strava account, overlays your splits, heart rate, and pace into a single timeline, and flags the decoupling points and aid station bleeds automatically. Pattern-match across races and stop relearning the same lesson at mile 70.
Free forever for core features. Pro adds AI race analysis and multi-race pattern tracking.