Why Your Workout Plan Should Tell You When to Push

Open any popular fitness app and look at a workout. You'll see exercises, sets, reps. Maybe rest periods. Maybe even target weights.

What you almost never see: how hard you should actually train each set.

This is one of the most important variables in muscle growth, and the entire industry has been ignoring it.

What "3 sets of 10" actually means

Here's a simple test. Imagine two lifters doing the exact same prescription: 3 sets of 10 reps on bench press at 80kg.

Lifter A picks a weight where 10 reps feels comfortable. The last rep is fast. They could probably do 5 more.

Lifter B picks a weight where the 10th rep is a grind. They couldn't have done an 11th if their life depended on it.

Same sets. Same reps. Same weight on paper. Wildly different training stimuli.

Lifter B will build muscle. Lifter A will mostly just stay the same. The difference between them isn't volume — it's proximity to failure.

RIR: the missing variable

RIR stands for "Reps In Reserve." It's a measure of how many more reps you could have done at the end of a set before form broke down.

• RIR 0 — you couldn't have done another rep with good form
• RIR 2 — you had two more reps in the tank
• RIR 4 — you stopped four reps short of failure (basically warming up)

The research on this has matured significantly in the last few years. The most comprehensive analysis to date is a 2024 meta-regression in Sports Medicine by Robinson, Pelland, Remmert, Refalo, Jukic, Steele and Zourdos^[1], which pooled data from dozens of studies and modelled the dose–response relationship between proximity to failure and hypertrophy. The conclusion: muscle growth scales as RIR decreases. Training closer to failure produces more growth.

Earlier work by Refalo et al. (2023)^[2] and Grgic, Schoenfeld et al. (2022)^[3] reached a more nuanced conclusion: training to absolute momentary muscular failure isn't required for maximum hypertrophy. The relationship plateaus near failure. Training at 1–2 RIR appears to capture most of the growth benefit. This was confirmed directly in an 8-week training study by Refalo et al. (2024)^[4]: subjects training at 1–2 RIR experienced quadriceps hypertrophy comparable to those training to true failure, with significantly less neuromuscular fatigue.

The independent evidence-based fitness community arrived at the same place. Greg Nuckols at Stronger by Science^[5] summarised the literature with: "Training closer to failure has a reasonably strong positive effect on muscle growth. However, proximity to failure has virtually no effect on strength gains within the published literature."

For hypertrophy, every rep further from failure costs you growth — but you don't need to grind every set into the dirt. The sweet spot is close to failure, not at it.

Why apps don't program this

If RIR is so important, why doesn't every app prescribe it?

A few reasons.

It's hard to explain. Most users don't know what RIR is. Apps optimised for mass adoption avoid concepts that require a glossary. "Do 3 sets of 10" is easy. "Do 3 sets of 10 with the last set at RIR 1" requires teaching.

It's hard to prescribe well. RIR isn't a single number — it should change across the mesocycle (lower as fatigue accumulates), across exercises (higher RIR on heavy compounds, lower on isolation), and across set numbers within an exercise (set 1 should be easier than the last set). Programming this correctly requires real exercise science knowledge. Refalo et al.'s 2022 scoping review^[6] identified more than a dozen variables that interact with proximity to failure.

It exposes lazy programming. If your app generates "3x10 bench press" for everyone, adding RIR forces you to think about whether the user is in week 1 or week 4, whether they're doing this as their first compound or their fifth, whether the exercise is high-skill or stable. Most apps don't bother.

So they leave it out, and the user is left to guess.

The mesocycle progression

Programming without RIR is like prescribing medication without a dose.

A well-designed 4-week hypertrophy mesocycle progresses RIR from conservative to aggressive. The model used by most evidence-based coaches — popularised by Renaissance Periodisation^[7] and validated in recent literature — looks roughly like this:

• Week 1: RIR 3 on compounds, RIR 2 on isolation. You're establishing the loads, learning the movements, and not accumulating much fatigue.
• Week 2: RIR 2 on compounds, RIR 1 on isolation. Pushing harder. The volume is the same, but the stimulus per set is increasing.
• Week 3: RIR 1 on compounds, RIR 0–1 on isolation. This is the hardest week of training. Last sets should grind.
• Week 4: RIR 0–1 on compounds, RIR 0 on isolation. Peak intensity. Some isolation sets can include drop sets.

This works because it accumulates stimulus while managing fatigue, building toward a deload. Without RIR, this progression is invisible. Every week looks identical. The user has no idea whether they're meant to be cruising or pushing.

Within-exercise progression matters too

It's not just week-to-week. Within a single exercise, RIR should also progress. If you're doing 3 sets, the first set shouldn't be at the same effort as the last set — you'd burn out before finishing.

The pattern that actually works: set 1 at target+1 RIR, middle sets at target, final set at target-1.

So in week 2 of a hypertrophy block, with a target of RIR 2, your bench press would look like:

• Set 1: RIR 3 — comfortable, leave reps in the tank
• Set 2: RIR 2 — pushing now
• Set 3: RIR 1 — last set, grind

This isn't intuitive. Most lifters either crush set 1 and have nothing left for set 3, or coast through every set. The structured ramp produces better results than either extreme.

The accuracy problem (and why it's fine)

One genuine concern with RIR: lifters aren't very accurate at estimating it. A 2022 meta-analysis by Halperin and colleagues^[8], building on earlier accuracy work by Helms, Zourdos and Ormsbee^[9], found that lifters typically underestimate their RIR by 1–3 reps on average. When you think you have 2 reps left, you usually have 3, 4, or even 5.

This sounds like a problem for RIR-based programming. It's actually why aggressive RIR targets work.

If a program prescribes RIR 0 on the last set of an isolation exercise, and the lifter thinks they've hit it but actually had 2 reps left, they're still training at a high enough proximity to failure to capture most of the hypertrophy benefit. The systematic underestimation gets baked into the prescription. You don't need lifters to be perfectly calibrated for the system to work — you just need the prescriptions to account for the typical bias.

Accuracy also improves with experience and gets better the closer you are to failure. Estimating "RIR 5" while warming up is hard. Estimating "RIR 1" on a hard last set is much more accurate, because you can feel the bar slowing down.

RIR makes training more sustainable, too

A 2025 study in the European Journal of Sport Science by Refalo and colleagues^[10] looked at the perceptual side of training to failure: how it actually feels. Subjects training to true failure reported significantly more displeasure during and after sessions than subjects training at 1–2 RIR. Similar hypertrophy outcomes, much worse subjective experience.

This matters because adherence is a real variable. A program you'll actually do for two years beats a program you'll abandon after six weeks. RIR-based programming threads the needle — close enough to failure to drive growth, far enough away that training stays enjoyable and sustainable.

How JSON.fit handles this

Every exercise in a JSON.fit program comes with an RIR scheme — per week, per set. Not buried in notes. Displayed alongside the rep scheme, so you can see exactly what intensity to hit at a glance.

The system handles the rules automatically:

• Compound lifts get more conservative RIR than isolation work (heavier loads, more skill demand)
• Heavy barbell movements never go to RIR 0 (injury risk and form breakdown)
• Machine and isolation work goes to RIR 0 in the final week
• Within-exercise progression is calculated for each set count
• Mesocycle progression is calibrated to standard 4-week blocks

This is what evidence-based programming actually looks like. Not just "3 sets of 10" — but "3 sets of 10 at RIR 2, with set 1 easier and set 3 the hardest, this week, in this exercise context."

The honest version of "personalised"

Apps love the word "personalised." Usually it means "we adjusted the rep range based on your goal."

Real personalisation means programming the variables that actually drive results: volume per muscle, frequency, exercise selection, and proximity to failure. Skipping RIR isn't a stylistic choice. It's leaving out one of the most important variables in the equation.

If your current app doesn't tell you when to push, it's not because RIR is too advanced or too niche. It's because including it would expose how shallow the rest of the programming is.