What 10,000 Form Analyses Say About How People Lift
Read this article in Deutsch , Español , Français , Italiano , Nederlands , Polski , Português , Svenska .
Data from 10,000 Gymscore form analyses reveals the most common squat, deadlift and bench mistakes—and why curls score worse than expected.
The lift people worry about is rarely the part they need to fix.
We analyzed the 10,000 latest workout form analyses completed with Gymscore to find out how people actually lift—not how coaches assume they lift.
The short answer: squat and deadlift performance is most often limited by postural alignment, while bench press is most often limited by foot placement. Range of motion gets plenty of attention, but it is much less likely to be the weakest part of a squat than posture. And supposedly “easy” movements such as push-ups, pull-ups and dumbbell curls are among the lowest-rated exercises in the dataset.
Here are the biggest findings:
- Postural alignment was the limiting factor in 50.3% of squats and 67.1% of deadlifts.
- Foot placement was the limiting factor in 42.6% of bench presses, but only 4.7% of deadlifts.
- Dumbbell bicep curls averaged just 37.7, placing them among the five lowest-scoring exercises. Yes, curls.
- Pull-ups improved fastest, gaining 2.84 points per repeat attempt in the measured cohort.
- More supported variations scored higher: trap bar deadlifts beat conventional deadlifts by 12.9 points, while front squats beat barbell back squats by 11.9.
The data suggests a useful coaching rule: look beyond the most visible part of the rep. The weak link is often somewhere else.
How we analyzed lifting form
The source is the 10,000 most recent analyses in Gymscore at the time of extraction. Each analysis receives an overall score and comparable scores for five technique categories:
- postural alignment;
- foot placement;
- range of motion;
- overall movement quality; and
- bracing.
Rather than asking only which category had the lowest average, we identified the category that was worst within each individual analysis. We call that the limiting factor. This matters because a category can have a respectable average across thousands of lifts while still being the weakest link for a particular person.
For individual faults, the point difference describes an association: analyses where a fault was detected scored that many points below the relevant exercise baseline. It does not prove that the fault alone caused the entire difference.
The most common squat form problem is not depth
Among 2,041 squat analyses from 561 users, the average overall score was 64.6.
| Top squat limiting factors | Share of analyses | Average category score |
|---|---|---|
| Postural alignment | 50.3% | 58.3 |
| Overall movement quality | 19.3% | 58.9 |
| Foot placement | 18.1% | 72.4 |
Postural alignment was the weakest category in half of all squats, and 44% of those postural scores fell below 60.
That result challenges the default squat conversation. Depth was mentioned in 66% of squat analyses, yet range of motion was the limiting factor only 12.2% of the time. When a depth-related issue appeared, the overall score was only 2.5 points below the squat baseline.
The less common faults were more costly. Knee valgus—the knees collapsing inward—appeared in 6.3% of squats, but those analyses averaged 18.8 points below baseline. Heels rising appeared in only 2.2% and was associated with a 10-point drop.
So, should lifters ignore squat depth? No. The finding is narrower and more useful: depth is frequently discussed, but posture is more often the thing holding the complete lift back. If you are checking only whether the hip crease passed the knee, you may be grading the most obvious part of the squat instead of the weakest one.
If depth is your concern, use an objective side-angle check rather than guessing. Our guide to checking squat depth without a coach explains how.
Two-thirds of deadlifts are limited by spinal position
The deadlift produced the clearest result in the dataset. Across 1,339 analyses from 440 users, the average overall score was 57.5.
| Top deadlift limiting factors | Share of analyses | Average category score |
|---|---|---|
| Postural alignment | 67.1% | 42.1 |
| Overall movement quality | 20.3% | 47.5 |
| Range of motion | 7.9% | 58.1 |
Postural alignment was the limiting factor in 67.1% of deadlifts. Its average of 42.1 was the lowest category score anywhere in this analysis, and 64% of deadlift postural scores were below 60.
The two most damaging detected faults were:
- Lumbar rounding: present in 15.1% of deadlifts and associated with a 17-point lower score.
- Hips shooting up before the bar: present in 20.4% and associated with an 11.4-point lower score.
Foot placement, meanwhile, was almost never the weakest link. It limited just 4.7% of deadlifts and averaged 82.4.
In plain English: most people put their feet in a reasonable place, then lose the lift somewhere above the ankles. The setup may look fine from the floor, but the torso and hips fail to hold their relationship as force is applied.
That is also why a deadlift can feel normal while looking very different on video. The lifter is concentrating on moving the bar; the camera can see whether the hips and spine changed first. See the five deadlift form mistakes worth checking for a practical review.
The biggest bench press mistake starts at the feet
Bench press flipped the pattern. Across 541 analyses from 201 users, the average overall score was 51.8.
| Top bench press limiting factors | Share of analyses | Average category score |
|---|---|---|
| Foot placement | 42.6% | 50.9 |
| Range of motion | 20.9% | 55.6 |
| Postural alignment | 18.5% | 52.4 |
Foot placement was the weakest category in 42.6% of bench press analyses. That makes it the single most common bench limiter—and the exact opposite of its role in the deadlift.
Why can the feet matter so much in an upper-body lift? They create the stable base that lets a lifter maintain whole-body tension and use leg drive without shifting on the bench. A loose or changing foot position can make the press less repeatable before bar path or elbow position is even considered.
Excessive arching was uncommon, appearing in 5% of bench analyses, but it had the largest score association of any individual fault in the dataset: 26 points below the bench baseline.
That does not mean every bench arch is an error. A controlled arch is part of many legitimate bench press techniques. The detected fault here was excessive arching within Gymscore's scoring criteria, and the result should be read in that context.
The surprise: curls are among the worst-rated exercises
When exercises were ranked by average overall score, the bottom of the table was full of movements people often treat as simple or beginner-friendly.
Five lowest-scoring exercises
| Exercise | Average score |
|---|---|
| Smith machine bench press | 35.9 |
| Pull-up | 36.1 |
| Chair dips | 36.9 |
| Dumbbell bicep curl | 37.7 |
| Push-up | 40.6 |
The curl result is funny because the movement looks almost impossible to misunderstand: bend the elbow, lift the dumbbell, lower it again. In practice, a curl gives people many ways to make a rep look complete—swinging the torso, moving the elbow, shortening the range or using momentum.
The broader pattern is more interesting than the joke. Pull-ups, chair dips and push-ups also scored poorly. Bodyweight and accessory movements were often executed worse than the major barbell variations. Pull-ups averaged 36.1 and push-ups 40.6, roughly half the score of front squats at 79.8.
One plausible explanation is coaching attention. Lifters rehearse and film their squat or deadlift setup, but may treat push-ups and curls as movements that require no technical feedback. Another is physical scaling: bodyweight exercises do not let a beginner reduce resistance in small increments as easily as a loaded exercise does.
The data cannot determine which explanation is responsible. It can tell us that “basic” does not automatically mean “well performed.”
Why advanced lifts appear near the top
The five highest-scoring exercises were:
| Exercise | Average score |
|---|---|
| Snatch | 81.6 |
| Front squat | 79.8 |
| Goblet squat | 75.5 |
| Weighted pull-ups | 74.6 |
| Hack squat | 74.0 |
No, this does not mean a snatch is easier than a push-up.
This ranking almost certainly contains self-selection. People who record a technically demanding snatch are more likely to have practiced it, received coaching or reached a level where filming it is useful. The pool of people attempting a push-up is much broader.
Weighted pull-ups versus ordinary pull-ups show the same effect. The weighted version averaged 74.6 while pull-ups averaged 36.1. Adding weight did not magically fix technique; the weighted cohort was likely much more experienced.
Treat these as execution scores among the people who submitted each exercise, not as a universal ranking of exercise difficulty.
Pull-ups improve fastest from the lowest starting point
For exercises with enough repeat analyses to estimate change over successive attempts, these improved fastest:
| Exercise | Score change per repeat attempt |
|---|---|
| Pull-up | +2.84 |
| Trap bar deadlift | +2.52 |
| Bench press | +2.46 |
| Conventional deadlift | +1.55 |
Pull-ups started from the lowest measured base in this repeat-attempt cohort—30.9—and improved by 2.84 points per attempt. That combination makes intuitive sense: when the initial execution leaves plenty of room to improve, a small setup change or cue can produce a visible jump.
The conventional deadlift result is particularly well supported within the repeat data: 303 analyses from 56 users, improving by 1.55 points per attempt.
This is an observed trend, not a promise that every new attempt adds the same number of points forever. Progress is rarely linear, and people who submit repeat analyses may be more motivated to act on feedback than one-time users.
Equipment and exercise variation can change the score dramatically
The largest gaps between related movements were bigger than the average improvement seen through repeat practice.
| Comparison | Score difference |
|---|---|
| Trap bar deadlift 69.2 vs conventional deadlift 56.3 | +12.9 |
| Front squat 79.8 vs barbell back squat 67.9 | +11.9 |
| Sumo deadlift 65.9 vs conventional deadlift 56.3 | +9.6 |
| Goblet squat 75.5 vs barbell back squat 67.9 | +7.6 |
| Dumbbell bench press 56.2 vs barbell bench press 49.2 | +7.0 |
Every higher-scoring variation in these comparisons either changes the load position, increases freedom of movement or reduces part of the spinal stability demand. That fits the earlier finding that posture is the dominant limiter in squats and deadlifts.
But this is not a randomized equipment trial. People choose exercises based on experience, goals, mobility, available equipment and prior discomfort. The score gaps show association, not proof that switching equipment will give any individual lifter 7 to 13 points.
The practical takeaway is still valuable: regressions and variations are not lesser versions of a lift. A trap bar, goblet position or dumbbells may give someone a version they can execute more consistently while building the capacity required for another variation.
Note: the headline squat, deadlift and bench sections group exercise families, while this comparison uses exact exercise labels. That is why their baseline averages differ slightly.
What lifters should do with these findings
The dataset points to four useful habits:
- For squats, check posture before obsessing over depth. Depth matters, but it was the limiting category in only 12.2% of analyses.
- For deadlifts, focus on the hips and back. Foot setup was rarely the problem; spinal position and hips rising ahead of the bar were much more likely to limit the lift.
- For bench press, build the lift from the floor up. Set the feet before unracking and check whether they stay stable throughout the set.
- Give “easy” exercises technical respect. Push-ups, pull-ups, dips and curls are not self-correcting just because they are familiar.
A single video is useful, but trends are better. Technique changes with fatigue, load and practice, which is why ongoing form analysis can reveal patterns that one perfect-looking rep misses.
The bottom line
Across 10,000 Gymscore analyses, the most common lifting mistake was often outside the area lifters watch most closely.
Squats and deadlifts were primarily limited by posture. Bench presses were most often limited by foot placement. Range of motion attracted attention without being the dominant squat problem. And curls—perhaps the most apparently straightforward lift in the gym—landed among the five lowest-scoring exercises.
The lesson is not that everyone needs identical technique. It is that intuition alone is a poor diagnostic tool. Record the lift, identify the actual limiting factor, change one thing and measure again.
Gymscore analyzes recorded sets across posture, placement, range of motion, overall movement quality and bracing, so lifters can replace guesswork with repeatable feedback. Read more about how Gymscore approaches its analysis and editorial content.
Frequently asked questions
What is the most common squat form mistake?
Postural alignment was the limiting factor in 50.3% of 2,041 squat analyses, making it far more common as the weakest category than range of motion at 12.2%.
What is the most common deadlift form mistake?
Postural alignment was the limiting factor in 67.1% of 1,339 deadlift analyses. Lumbar rounding and the hips rising ahead of the bar were the two detected faults associated with the largest score drops.
Why does foot placement matter in the bench press?
Stable foot placement helps maintain whole-body tension and creates a consistent base for the press. It was the limiting category in 42.6% of 541 bench press analyses.
Are bicep curls easy to perform correctly?
They look simple, but dumbbell bicep curls averaged only 37.7 in this dataset, placing fourth among the five lowest-scoring exercises. Momentum, torso movement and changing elbow position can make the rep less controlled than it appears.
Does a higher exercise score mean the exercise is easier?
No. Scores describe the submitted analyses, not universal exercise difficulty. Advanced movements may score highly because the people recording them are a more experienced, self-selected group.
What is the fastest-improving exercise?
Pull-ups improved fastest in the repeat-attempt cohort, at 2.84 points per attempt from an initial average of 30.9. This is an observed cohort trend, not a guaranteed rate of progress for every lifter.