A benchmark workout only works as a benchmark if the second attempt is the same test as the first. Your finish time tells you very little if the load changed, the pull-up variation changed, or a cramped home setup added ten seconds to every transition. This guide covers what to write down when you repeat a mixed benchmark, and how to label scaled versions so you never compare them against the standard by mistake.
Why a time on its own can't be compared
CrossFit describes Fran as a task-priority workout: the work is fixed (21-15-9 reps of thrusters and pull-ups) and you try to finish it as fast as possible while keeping full range of motion and sound mechanics. When you repeat it, CrossFit counts two results as improvement: finishing faster with the same weight, or finishing in a similar time with more weight.
Both of those depend on things a stopwatch doesn't record. "6:12" next to "5:40" looks like progress. But if the first attempt used a barbell and strict pull-ups and the second used dumbbells and banded pull-ups, you didn't retest anything. You did a different workout. The same source notes that Fran is easily scaled by reducing the load, swapping to dumbbells for the thrusters, lowering the reps, or replacing pull-ups with another bodyweight pulling movement. Every one of those options is a variable you have to log.
The conditions worth recording
Treat each attempt as a small experiment and write down whatever could change the result. For mixed benchmarks, these are the conditions that matter:
| Condition | What to write | Why it matters on a retest |
|---|---|---|
| Version | "Standard" or "Scaled," plus every modification | Stops you from comparing different workouts |
| Rep scheme | Exactly as performed (e.g., 21-15-9 or 15-12-9) | Lower reps make a different test |
| Load and implement | Weight per implement, barbell vs. dumbbells, load per side if it's uneven | Load is half of the improvement definition |
| Movement variation | Strict, kipping, banded, ring row, jumping, etc. | Variations differ a lot in difficulty |
| Movement standard | Range-of-motion standard you held yourself to | A looser standard inflates speed |
| Setup and transitions | Distance between stations, equipment reset, who started the clock | Home setups change transition time |
| Pacing and breaks | How you broke up sets (e.g., 21 as 11+10) | Shows where time was gained or lost |
| Result | Finish time, or reps completed at a time cap | The outcome itself |
| Context | Time of day, what you did earlier that day or week, how you felt | Helps explain an outlier |
You don't need to write an essay for each row. One line per condition is enough. The rule is simple: if changing it could move the time, record it.
Label standard and scaled attempts separately
The most common logging mistake is a single "Fran" column with every attempt in it. Split each benchmark into named versions, and only compare attempts within the same version.
A practical naming pattern:
- Fran – Standard: the workout exactly as published, with the load and movements in the official description.
- Fran – Scaled A (DB thrusters, banded pull-ups): a fixed, repeatable modification.
- Fran – Scaled B (15-12-9, ring rows): a different modification, kept as its own line.
Once a scaled version has a name, keep it fixed. If you change the band, the dumbbell weight, or the rep scheme, that's either a new version or a deliberate progression you note on the attempt. Don't quietly blend it into the old line.
Scale to keep the stimulus, then log the stimulus
CrossFit's coaching guidance describes scaling to preserve the intended stimulus by adjusting load, volume, and movements so the challenge fits the athlete. One of its example workouts states the intended stimulus as 5–9 minutes, with each movement done in 1–2 sets. For Fran, CrossFit suggests a thruster load that allows at least 15 unbroken reps and a pull-up option that lets you finish each round with minimal rest.
That gives you something to check after each scaled attempt. Write down whether you hit the intended feel: roughly the target time range, and sets broken about as planned. A scaled attempt that lands in the intended window is a usable data point. One that turned into a 20-minute grind may tell you the scale was wrong, which is useful too, but it shouldn't be treated as your baseline for that version.
A worked example (illustrative values)
The numbers below are made up to show the logging pattern. They aren't targets or prescriptions.
| Field | Attempt 1 | Attempt 2 |
|---|---|---|
| Version | Fran – Scaled A | Fran – Scaled A |
| Reps | 21-15-9 | 21-15-9 |
| Thrusters | 2 × 30 lb dumbbells | 2 × 30 lb dumbbells |
| Pull-ups | Banded, medium band | Banded, medium band |
| Standard | Full depth, arms locked out overhead; chin over bar | Same |
| Setup | Bar 3 steps from dumbbells | Same |
| Breaks | Thrusters 11+10, 8+7, 9; pull-ups 7+7+7, 8+7, 5+4 | Thrusters 21, 8+7, 9; pull-ups 11+10, 8+7, 9 |
| Result | 8:45 | 7:20 |
| Stimulus check | Over the planned window; pull-ups broke heavily | Closer to the planned window |
| Notes | Morning, after a leg day the evening before | Rested day before |
Both attempts share the same version, load, band, standard, and setup, so the 1:25 improvement is a fair comparison. The break pattern shows where it came from: fewer pull-up breaks. The context row also flags that attempt 1 followed leg training, which reasonably accounts for part of the gap. That's worth remembering before you credit all of it to fitness.
Now suppose a third attempt uses a lighter band and finishes in 7:50. That isn't a regression on Scaled A. It's either the first attempt of a new version or a progression step toward the standard, and the log should say which.
Keep home loads repeatable
At home, "the same weight" is often where comparisons quietly break down. Mixed sets of dumbbells, unlabeled plates, and improvised substitutions make it easy to drift a few pounds between attempts without noticing.
The Speediance Gym Monster 3 is an all-in-one home gym with digital cable resistance, resistance adjustments in 1 lb increments, and independent load on each side.
If you load cable-based movements for a benchmark, write the exact number per side in your load row. Then set the same values next time, or raise them by a small, known amount when you want to test "similar time, more weight." A cable movement is still a different implement from a barbell or dumbbells, though, so a benchmark done that way belongs under its own named version.
Fields that are easy to forget
A few conditions rarely make it into a log, but they explain a lot of strange results:
- Clock method. Running clock, phone timer, or app, and whether the clock started on "go" or on your first rep.
- No-reps. If you repeated reps that didn't meet your standard, note it. Otherwise a sloppy fast day can look like a breakthrough.
- Time cap result. If you capped, record the reps completed, not just "DNF," so the attempt can still be compared within its version.
- Equipment changes. A new bar, different rings, or a swapped band all count. Name them.
- Warm-up. Keep it roughly the same between attempts, and note it when it isn't.
If you're new to logging and want the smallest possible set of fields first, start with the basics and add these benchmark conditions only when you begin retesting.
Retest checklist
Before you compare two benchmark results, confirm:
- Same named version: standard vs. standard, or the same named scaled version.
- Same rep scheme and movement variations, including band or substitution.
- Same load per implement, or a recorded, deliberate increase.
- Same movement standard, with no-reps noted.
- Comparable setup and clock method.
- Context noted, so you can explain an outlier rather than trust or dismiss it.
If all six hold, the time difference is a real comparison. If any one fails, log the attempt as a new version or an annotated exception, and pick a later retest date when you can reproduce the conditions.
