do i actually need a trainer.

Published September 5, 2026·Training science·14 min read

App workouts vs PDF programs vs a personal trainer: the one randomised trial that ran all three side by side, what a person standing next to you actually changes inside a set, and how to get most of that without paying for it.

you have a program. it is in a pdf you bought at 1am, or an app you downloaded in january, or a notes file called gym??? that has three screenshots and a saved reel in it. and every few weeks, usually after a session where you did everything and felt nothing, the question comes back. do i actually need a trainer, or am i just not trying hard enough.

the internet's answer to this is a genre. every "app vs personal trainer" article on page one is written by an app company or a training studio, and every one of them lands on the thing its author sells. i sell both a pdf and an app, so i am not going to pretend to be neutral either. what i can do is read the trials, because there are trials, including one from last year that put a coach, an app and a pdf in the same room for ten weeks and measured what happened.

the short version is that the format is not the variable. what the format supplies is. a trainer supplies three things: the load you pick, the rate at which it climbs, and a reason to be there on a tuesday. a pdf supplies none of those unless you build them yourself. an app sits in between, and where it sits depends on whether there is a human somewhere behind it. the rest of this is the numbers.

what did the trial that compared all three actually find.

in 2025 a German group published the study this whole genre had been missing. they took 112 trained adults, randomised them into three groups, and gave every group the same program: ten weeks, three full-body sessions a week, squat, bench, deadlift and overhead press as the main lifts, every set taken to the point where the next rep would not happen, and one progression rule for everyone. the rule was the 2-for-2 rule: beat your rep target by two, in two consecutive sessions, and the load goes up.

the only thing that differed was who was watching. the coached group had a certified trainer in the room, one coach for every one to four lifters, correcting technique and calling the load increases on the spot. the app group used a coaching app with technique videos, a progress log, and feedback on their form by video, on a delay. the pdf group got the program as a document and nothing else until the end, when their training logs were collected.

now the lifting. every group got stronger, which is worth saying plainly, because the pdf people who stayed did fine: squat one-rep max rose 26.6 kg with the coach, 19.2 kg on the app and 19.4 kg on the pdf, and the coached difference was statistically real against both. bench press rose 9.1, 8.2 and 7.7 kg — no difference between formats. fat-free mass rose 1.4 kg in the coached group and did not move in the other two, measured by bioimpedance, which is a noisy instrument, so hold that one loosely. the authors' own reading of the split is the one i would give: the squat is a multi-joint lift where technique and confidence under the bar decide the load, and a person in the room fixes both in real time. the bench press is simple enough that trained people run it fine off a screen.

the honest lines, since this is the paper the whole article leans on. the coached group was younger [about 26 against 32]. only people who hit the 85 percent attendance bar were analysed, so the strength numbers describe the ones who stayed, not everyone who started. nobody's diet was controlled. and the app arm was not a bare app — it had a coach reviewing form video on a delay. that last one matters for the verdict, because it is the lane most real apps, mine included, actually sit in.

why does a person standing next to you change the result.

the lazy explanation is motivation. the measured explanation is more specific and more useful, because you can copy it. a trainer changes three numbers: the load you choose, the rate at which that load climbs, and the number of sessions you complete. each has its own study.

the load you choose. in 2008 a New Jersey lab asked 46 resistance-trained women to load four exercises with the weight they normally used for a set of ten, then tested their actual one-rep max. the women who trained alone had been picking, on average, 42 percent of their max. the women who trained with a trainer were picking 51 percent. leg press 41 versus 50, chest press 48 versus 57, seated row 42 versus 56. same instruction, same gym, nine percentage points of intensity apart, and the trainer group rated the sets as harder because they were. this was a cross-section, not a randomised trial — women who hire trainers may differ from women who don't — so a Brazilian group repeated it in 21 trained men and women and found the same shape: loads 12 to 27 percent higher with a trainer present, with the biggest gap on the lower-body lifts. and then the line that should be printed on every pdf: even with the trainer, the self-selected loads were, in the authors' words, "considerably lower" than a true ten-rep max. left alone, almost nobody picks a weight that produces the adaptation they came for.

the rate the load climbs. the oldest study in this pile is still the cleanest test of it. twenty moderately trained men, one 12-week periodised program, half of them with a trainer. the program was identical on paper — same sets, same rep ranges, same phases — but the trainer group's kilograms per set were significantly higher from week 3 on the bench and week 7 on the squat, the slope of their load increase was steeper, and their one-rep maxes were higher at the end. the trainer was not running a better program. he was making the program's own progression rule actually happen, week after week, at the moment the lifter would have rounded down.

the sessions you complete. in 42 teenage rugby players on the same 12-week program, the supervised half completed more sessions, and sessions completed correlated with strength gained. that is the boring, decisive channel: the person you booked is a reason to be there. but here is the finding that separates the two mechanisms, and it is the one i want you to hold onto. 124 untrained men, 11 weeks, one coach per 5 lifters versus one per 25. attendance was the same. total work done was the same. and bench press still rose 15.9 percent in the closely watched group against 10.2 percent in the other, with knee-extension strength rising only in the watched group. same attendance, same volume, different result — which means the inside-the-set effect is real on its own and is not just "they showed up more".

so is the trainer effect strength, or size.

mostly strength. when a British group pooled 12 supervised-versus-unsupervised studies — 301 supervised lifters, 276 unsupervised — the effect on performance was small [standardised mean difference 0.28], the effect on strength specifically was moderate [0.40], and the effect on body composition was 0.07, with a confidence interval that crosses zero. in older adults the strength effect is a little larger, 0.51 across 11 trials. so the honest claim is not "a trainer gets you a different body". it is "a trainer gets you to lift more, sooner, and that is the input a different body is made from".

the size data is thinner but it points the same way. in the closest thing to a like-for-like test, 36 resistance-trained men and women ran an identical 8-week program with or without a coach in the room, and the coached group grew more at the triceps and the front thigh by ultrasound and had the bigger squat, while bench press and isometric strength were the same. a health-club trial in 34 men aged 30 to 44 randomised to a trainer or to training on their own found 1.3 kg of lean mass gained with the trainer against none without, and chest press strength up 42 percent against 19. men only, one club, twelve weeks — a range marker, not a law. put it together and the pattern repeats in every study that separates the lifts: where a person changes the outcome is the complex, loaded, slightly scary movement. where the movement is simple, the screen does fine.

what did the app actually do in the trial.

the app group is the interesting one, because it is where most people reading this actually live. it kept 81 percent adherence, lost 9 of 48, and gained the same bench and the same squat as the pdf group that stayed — with the squat behind the coach by about 7 kg. the app in the trial had three things: technique videos, a log that the coach could see, and form feedback by video on a delay. that last one is the lane. it is not a person in the room and it is not nothing.

there is a small literature on exactly how much of the room you can move onto a screen, and it is more encouraging than the genre admits. when 20 trained adults did three weeks of live one-to-one virtual sessions and three weeks in a studio, in a crossover, the strength and body-composition changes were the same, and the clients rated effort, motivation and the quality of supervision as equal between the two. three weeks is too short to grow anything, so read it as an effort study, not a hypertrophy study — but effort is the thing we established a trainer changes. two trials in older adults found the same: app-delivered exercise with remote supervision matched in-person supervision over 8 weeks, and videoconference classes matched a weekly text message over 12. a live human on a screen preserves the effort. a delayed human preserves the attendance. a document preserves neither.

the other thing the app did is the least glamorous and the best evidenced: it made people write it down. when researchers coded 122 physical-activity and diet interventions covering nearly 45,000 people by the techniques they used, self-monitoring explained more of the difference between programs that worked and programs that didn't than any other single technique, and programs that paired it with one more self-regulation tool — a goal, a review, a prompt — averaged an effect of 0.42 against 0.26 for the rest. the log is not admin. the log is the intervention. the progressive-overload article is, underneath, an argument for the same thing.

one honest aside about the phone that carries the app. in 16 trained adults, thirty minutes of scrolling social apps before squats lowered the total volume they lifted, with no change in lactate or in how hard the sets felt — a head effect, not a body one. the app is on the same glass rectangle as the thing that quietly takes reps off your bar. open the log, not the feed.

why did the pdf lose half its people.

because a document has no way to notice you have gone. this is not a fitness finding, it is a digital-health one: the "law of attrition" was named in 2005 for exactly this pattern — in any self-help program delivered without a person attached, a large share of users stop, and the ones who stop are invisible to the program. the pdf in the trial had one control mechanism, which was collecting logs at the end. by then 22 of 47 had left, and 12 of those cited illness and 6 personal reasons, which is to say ordinary life, unopposed.

the psychology behind the number is measurable too. in 588 Brazilian exercisers, the ones training with a personal trainer scored higher on feeling competent, feeling supported and feeling that the choices were theirs, while the ones training unsupervised in a gym scored higher on amotivation and on doing it because they felt they should. a cross-section, not a cause — but it names the thing a pdf cannot give you: someone who has already watched you do the lift, telling you the number for next week. and in 129 people given ten weeks with a student trainer, 73 percent of those who could move up a stage of exercise readiness did. a weekly conversation about barriers and plans moved people who had been stuck.

i will say the self-implicating part once, plainly. the Fighter is a pdf. i wrote it as a pdf on purpose, because a document is the most honest way to hand someone a complete system without a subscription attached, and a lot of people run it exactly as written and get exactly what it promises. but the trial says the document half of my shelf loses half its buyers by week ten unless something else supplies the showing up. that is why the app exists next to it: the log, the demos, the reminders, and a DM inbox where the form checks go to a person, on a delay. that is the trial's app arm, more or less. i did not build it because of the trial. the trial just explained what i had been watching in the inbox for years: the people who stop are the people who stopped writing it down.

how to run a pdf like a supervised block.

the trainer supplies five things, and four of them can be written down. here is the version i give people who buy the document and are not going to buy the person.

and then the end date. the trial ran ten weeks and tested at the end, and every group's lifters who reached that test had gotten stronger. the pdf does not come with a test; put one in. week 10, same lifts as week 1, write down the difference. the number is the thing that turns a document into a program you are actually on. if this is your first block and the words "the squat" made your stomach drop, the first-session article is the on-ramp, and the deadlift one is the honest read on the injury numbers people are actually scared of.

so. do you need one.

here is the scorecard, with the trial's own results in it, and with the caveat that "trainer" in every study above means a qualified person running an evidence-based program — not the floor guy who hands you a bosu ball.

so, the answer to the 1am question. if you are in your first months, or the lifts that scare you are the ones the program is built on, yes — buy the person, or at least buy eight sessions of the person and learn the squat, the hinge and the press from someone who can see you. that is where the effect lives. if you have trained a year or more with a log, an app that makes you log and puts a human on the other end of the form check gets you most of the way there — the trial says within a few points on adherence and a few kilos on the squat. and if all you have is the document, you are not doomed, you are just unsupervised, and the five rows above are what supervision was doing. write the load. move it on a rule. book the slot. fill in the log. film the big lifts. the pdf was never the problem. the empty space next to it was.

Quick answers

is a personal trainer worth it compared to a workout app?

for showing up, an app got close: in the one randomised trial that compared all three formats, 81 percent of app users kept at least 85 percent of their sessions over 10 weeks, against 88 percent with an in-person coach and 52 percent with a PDF. for strength on complex lifts the coach still won — squat one-rep max rose 26.6 kg supervised versus 19.2 kg on the app — while bench press gains were the same in all three groups. so a trainer is worth it most in the first months and on the barbell lifts, and least on simple machine exercises you already know.

why do people get stronger with a trainer even on the same program?

because the trainer changes what happens inside the set, not just whether you turn up. resistance-trained women picked a 10-rep load of 42 percent of their max when training alone and 51 percent with a trainer. in 124 men on an identical 11-week program, the group with one coach per 5 lifters gained 15.9 percent on bench press against 10.2 percent for one coach per 25 — with no difference in attendance or total work. supervision raises the load you choose and the rate at which it climbs, and those two things are most of the strength result.

can an app or online coaching replace in-person supervision?

for effort and adherence, mostly yes; for technique on complex lifts, partly. live one-to-one virtual sessions produced the same perceived effort, motivation and supervision quality as studio sessions in 20 trained adults over 3 weeks, and app-based programs with remote supervision matched in-person supervision in two trials in older adults. where the app arm fell short in the 10-week trial was the squat, where the feedback was delayed video rather than a person in the room. the fix is to send form video on the complex lifts and let the app carry the log and the schedule.

how do i follow a PDF program without quitting?

supply the three things the trainer would have supplied. one, a written load for every set before you walk in, with a rule that moves it — the trial used the 2-for-2 rule: beat the rep target by 2 in 2 consecutive sessions, add weight. two, a log you fill in during the session, because self-monitoring is the single behaviour-change technique with the largest measured effect across 122 interventions. three, a booked slot and an end date: a 10-week block with a test at the end. the PDF arm lost 22 of 47 people over 10 weeks; every one of those three is a reason not to be one of them.

keep reading

• why can't i stay consistent. — program hopping is the pdf problem in a different costume
• what is progressive overload, simply. — the rule the trainer was enforcing, written out
• what do i even do at the gym. — the first session, if the squat is the scary part
• how long does it actually take. — the week-10 test, and what the mirror does after it
• why does the gym make me want to leave. — for when the reason you train alone is not the price of a trainer

---

References

educational content, not medical advice. the numbers above are group results from specific populations and cannot rule on you: the three-format trial analysed only lifters who kept 85 percent of sessions, had a younger coached group and no dietary control, and measured body composition by bioimpedance; the self-selected-load studies are cross-sections of people who chose or did not choose a trainer; the health-club trial is men only; the load-slope study is twenty men; the virtual-training crossover ran three weeks per arm. everything in the tables is sourced.

- MoMo