The Grov thesis
Why most programs fail, and what to build instead.
Most programs optimise one dimension (intensity, frequency, novelty) and quietly ignore the others: adherence, completeness, time. A program only works if you actually do it, and only if it covers every muscle group you'll need for the next 40 years. That's a much harder design problem than the one most programs are solving.
Five ways a program fails.
Pick any program off the shelf and it will be world-class at one or two of these, and silently broken on the rest.
Sessions stretch past what you can fit between work and dinner. You skip them.
Novelty wears off in 3 weeks. If the plan depends on you feeling motivated, you're in trouble.
Bro-splits skip the adductors, rotators, tibialis, deep neck flexors. Weaknesses compound.
The optimal program you skip is worse than the okay program you do every week for 20 years.
Without feedback on effort, you're either creeping weight up too slowly or crashing into failure every set.
Three decisions that make the program work.
If you had to rebuild a training program from first principles, optimising not for a lab result but for a real human with a job, a family, and 40 more years of training ahead, you'd make three decisions differently than almost every program on the market.
1. Completeness over optimisation.
Covering all major muscle groups, including the ones most programs skip, matters more than whether your biceps are at 12 or 18 sets a week. You can argue about optimal volume forever, but if your program has no direct work for your adductors, deep neck flexors, or tibialis anterior, those weaknesses are going to compound into injuries the volume debate can't fix.
The cleanest way we've found to cover everything without bloating session length is antagonist pairs: every muscle group has a partner that opposes it. Quads and hamstrings. Chest and back. Biceps and triceps. Tibialis and calves. Train them together and you get balanced movement, efficient supersets (one works while the other rests), and a built-in forcing function for completeness: you physically can't skip one half of a pair without it being obvious.
Grov is 19 antagonist pairs. That's the coverage. Not optional.
2. Autoregulation over rigid prescription.
Fixed 5×5 programs don't know whether you slept 5 hours or 9. They don't know if you're stressed, fighting a cold, or coming off a week of travel. They tell you to add 2.5 kg because Tuesday comes after last Tuesday. That's not a program. It's a calendar.
RPE-based autoregulation (rate of perceived exertion) closes the loop. You rate how hard the last set actually felt, and the next prescription adjusts. Nailed it easily? Go up. Barely got it? Stay. Missed a rep? Come down. This is how serious strength coaches (Helms, Israetel, Ethier) have been writing programs for a decade. It's just that most consumer apps haven't caught up, because "the app decides the weight" is easier to market than "you and the app decide together."
Grov asks one question after each exercise: beat, exactly, or missed. Weight updates automatically.
3. Adherence over intensity.
The best single predictor of strength gains over 5 years is not your 1RM percentage, not your volume, not your split. It's whether you were still training. Everything else is rounding error on top of that one variable. And yet almost every program is designed for the session you'd do on your best day (fully recovered, well-fed, motivated, 75 minutes free), not the session you'd do on your worst.
Design for the worst day, and the best day takes care of itself. That means short sessions (so you can fit one in when the day is against you), minimal friction (no signup, no setup, no thinking), and a weight prescription that adapts down when you're cooked instead of shaming you for missing reps.
Grov is 30 minutes, three times a week, with no signup. That's the adherence design.
How Grov is designed for each factor.
Four design decisions, each mapping directly onto one of the failure modes above.
19 antagonist pairs
Covers every major muscle group, plus the ones most programs skip: tibialis, adductors, deep neck flexors, hip rotators.
30-minute sessions
Two pairs × 2–3 working sets each × 3–5 minutes per pair. Every time. No surprise hour-long days.
RPE feedback
Rate each exercise beat / exactly / missed. Next session's weight adjusts automatically, so you're never guessing.
Zero friction
No signup, no app install, nothing to configure. You open it, you start. The session you do on your worst day.
Where this comes from.
Grov didn't invent any of this. It synthesises decades of work by coaches and researchers who've been careful about the evidence. A few places worth starting:
Jeremy Ethier · The Fastest Way to Gain 20 lbs of Muscle (Naturally). On effective volume, proximity to failure, and why completeness matters.
- Brad Schoenfeld: Strength and Hypertrophy Adaptations Between Low- vs. High-Load Resistance Training
Systematic review and meta-analysis, J Strength Cond Res, 2017. The core evidence that proximity to failure, not absolute load, drives hypertrophy across a wide rep range.
- Eric Helms: Renaissance Periodization
The Hypertrophy Guide and the broader RP library on RPE-based programming. Foundational reading on autoregulation for non-competitive lifters.
- Mike Israetel: Volume Landmarks for Muscle Growth
MV, MEV, MAV, MRV: the vocabulary behind "how many sets is enough for this muscle this week," and why the honest answer is a range, not a number.