How Heft measures your training

The session score: four parts, one number

The session score is a 0–100 rating built from four weighted parts: plan (35%), effort (30%), progress (20%) and balance (15%). It measures each workout against your own plan and your own previous sessions, not a universal standard. When a session gives a part nothing to rate, such as balance without a plan, that part drops out and the others share its weight.

Plan · 35%

Heft counts the working sets you completed against the sets your plan asked for. Sets beyond the plan on an exercise do not count extra, and a swapped exercise still does the work. Without a plan, your average session over the previous 28 days is the target; with no history yet, eight working sets count as a full session. Warm-up sets count on neither side.

Effort · 30%

How many of your working sets ended close to failure. Heft rates each set from its weight and reps, your history and your reps in reserve when you log them. High-rep volume sets count half, and four hard sets out of five earn full marks, because a good session still has room for some lighter work.

Progress · 20%

Each exercise is compared with the last time you did it, by estimated one-rep max. Matching or beating last time earns full marks; falling 15% or more earns 20 points. A new best on any exercise adds a 10-point bonus. Exercises you do for the first time are not rated.

Balance · 15%

Every muscle group your plan targets needs three working sets in the session. Each group earns its share as it gets there, so skipping a target group costs points. Without a plan, balance is not rated.

Scores map to plain-language bands: 85 and above is an excellent session, 70–84 strong, 50–69 solid, and below 50 flags something worth a closer look. Tap any score in the app to see how many points each part earned and why.

Strength levels: the Bronze-to-Legend ladder

Every exercise you train is placed on a 24-step ladder: eight tiers — Bronze, Silver, Gold, Platinum, Diamond, Master, Grandmaster and Legend — with three divisions each. Your level on a lift comes from your estimated one-rep max, adjusted for bodyweight and sex, measured against per-exercise strength standards. Each muscle group’s level comes from your three strongest exercises for it, and your overall level averages the muscle groups you train.

Within a tier, your position is interpolated linearly between thresholds, so a level like “Gold 2” moves smoothly as your estimated max grows instead of jumping only at milestones. A new exercise joins your overall level after three training days, so a first cautious session never drags it down.

Strength levels describe what you can do now, not what you once did. Sets from the last 90 days count in full; older sets fade out over the following 60 days instead of dropping off at once.

Estimating your one-rep max

Heft estimates your one-rep max from any hard set using a rep-percentage table: a weight lifted for 10 reps is treated as 75% of your maximum, so the estimate is weight ÷ 0.75. The table runs from 1 rep (100%) down to 30 reps (50%), which keeps estimates conservative for high-rep sets where classic formulas overshoot.

Up to about ten reps the table tracks the familiar Epley estimate closely; beyond that it flattens on purpose, and sets above 30 reps are treated as 30. For dumbbell exercises the rep count is halved before the lookup so their estimates stay comparable with barbell-based standards.

If you log reps-in-reserve (RIR), Heft uses it to classify sets as warm-up, volume work or muscle-building, and to calibrate its rep-endurance model to you. Strength levels and medals deliberately ignore RIR: they only count the reps you actually completed at the weight on the bar, so an optimistic estimate can never inflate a rank.

Where the strength standards come from

Heft ships strength standards for 362 exercises, with separate thresholds for men and women, around 720 threshold sets in total. Most exercises use bodyweight-relative thresholds; bodyweight movements such as pull-ups use repetition thresholds; a small group of fixed-implement movements uses absolute weights.

Three kinds of standards cover the library:

Bodyweight-relative

The default. Tier thresholds are multiples of your bodyweight: at 80 kg, a Gold bench press (1.25×) means an estimated one-rep max of 100 kg.

Repetition-based

Bodyweight movements such as pull-ups, dips and push-ups are ranked by strict reps rather than by estimated 1RM.

Absolute

A small set of movements where load barely scales with bodyweight uses fixed weight thresholds instead.

Two examples from the dataset

One honest caveat: the tier descriptions in the app (“stronger than roughly 50% of lifters” and similar) are orientation bands, not live percentiles. The thresholds are curated from published strength standards and coaching literature and reviewed against how they behave in practice. They are not computed from Heft users’ data, and your personal level is computed on your device.

How the community exercise figures are calculated

The exercise pages use a privacy-floored weekly aggregate of optional synced workouts. An exercise appears only after at least 10 users and 150 valid sets; progression, plateau, retention and strength curves each require larger eligible samples before they are published.

Weighted sets use the Epley estimate, weight × (1 + reps ÷ 30), to estimate one-rep max. Strength curves contain only deciles of each eligible user’s best estimate, never individual sets. Progression uses robust weekly trends, and bodyweight-dominant exercises suppress all weight-derived figures. The aggregate has no sex or bodyweight data, so community markers are unisex and only approximate against the sex-specific strength tables. The website reads a manually reviewed, dated snapshot rather than live training data.

Muscle balance: the 22-muscle map

Every movement in Heft’s 491-exercise library is mapped to its primary movers, assisting muscles and stabilizers across a 22-muscle taxonomy. Weekly set goals, 12 sets for large muscle groups and 6 for smaller ones unless you set your own, drive the muscle heatmap.

The same mapping powers the searchable exercise library on this site, with target muscles for every movement and step-by-step technique guides for the most common lifts. Browse the exercise library →

Private by design

Your session score, strength levels and muscle statistics are computed on your phone from your own log. Sync is optional and your data is exportable at any time. When sync is enabled, eligible sets may contribute only to the privacy-floored exercise aggregates described above; Heft does not publish individual histories or user rankings.

Methodology FAQ