Do You Need a Personal Trainer or Coach? What Five Kinds of Paid Help Actually Do

A block of personal-training sessions. A weekly box of meals. A slimming group, or one-to-one appointments. The "add a coach" button in your weight-loss app, or a bet with your own money riding on it.

Five checkout pages, and every one of them asks the same thing: is this help worth paying for?

Here is the straight answer to that, first. Not one of the 17 studies behind this guide put a price on any of it. Nobody can honestly tell you from this research whether a trainer, a meal box or a coach tier is worth the money.

What the studies did measure is something you can actually judge: what each kind of help changed, on which scale, and how many people were still going at the end.

A squat in kilos, a body measured for fat and lean tissue, weight on the bathroom scale. Those turn out to be different questions with different answers.

So the better question isn't "is it worth it?" but "what does it do, and for whom?", and that one has an answer for all five offers.

The evidence behind it: 17 studies, led by four main trials and reviews with 5,368 people enrolled between them, published across three decades, with older trials folded inside the reviews. Every claim was checked against the papers themselves, and the studies that didn't make the cut are named at the end.

Ask around and you get two camps. Gyms and trainers say a trainer is usually worth it. Forums say it's a waste of money. Both are arguing about the same prize, a better body, and the trials answer neither camp's version of it.

Start with what a coach in the room actually did. In the clearest trial, people who had lifted for a year or more followed one 10-week plan delivered three ways. Among those who finished, lifters with a coach beside them at every session put about 7 kg more on their squat than those coached through an app or using the plan as a PDF.

The coached group was small, younger and started with the lightest squats. Still, two other trials of experienced lifters pointed the same way, one of them by a margin that may be close to nothing.

Why the squat? The authors' untested explanation: a squat needs many joints moving in the right order, so feedback during the rep pays off. On the simpler bench press, the three groups gained about the same, and in a trial from 2000 the coached lifters also raised their loads faster.

Then the body itself. A review combining the six trials that measured body composition, mostly in people new to lifting, found almost no difference with or without supervision. Three small single trials found a little more lean tissue (everything that isn't fat) with a coach, but two used a test the newest trial's authors called imprecise. Not settled.

The lifters with no coach's help weren't far behind. In the two trials that printed both sides, those who finished kept about three-quarters of the coached squat gain. The coach's part was the top quarter, not the whole.

That only counts the people who finished, though. With a coach in the room, 15 of 17 lifters reached at least 26 of the 30 sessions; with the plan as a PDF, 25 of 47 did. On that count, the coaching app came in almost level with the coach.

It's one trial, nobody checked whether that gap could be chance, and older trials combined found a much smaller one. But it moves the question. Not "is a trainer worth it?" but "would I get to the end without one?"

One boundary. None of these single trials ran past 12 weeks, included anyone over 58, or cut anyone's calories. If you're lifting while dieting, you sit between this evidence and the next: no study here tested a trainer during a cut.

If what a trainer seemed to buy was getting people to the end, what happens when the help is the food itself, handed over every week?

The measuring stick changes now. The trainer trials tracked strength and body composition; the food, group, app and money studies all count kilos on the bathroom scale, in weight-loss programmes.

The answer first: in these studies, having the food handed over bought a bigger first year. In an 18-month trial from 1993, both sides of the food comparison went to the same group programme. One side was also given five free breakfasts and five free dinners a week, with a written plan to match.

At the one-year weigh-in, the people with the food had lost 9.1 kg against 4.5 kg. Twice as much, at that point. A review of 23 trials pointed the same way at a year for programmes that included meal replacements, mostly shakes, soups and bars. One trial inside it, which cut back its portioned food, had lost its lead by six months.

Why the food worked is less clear than that it did. It came as a bundle (a plan, ready portions and maybe a reason to show up), and nobody pulled the pieces apart. The people with the meals didn't report eating fewer calories than the rest, though less of what they ate was fat.

The same trial tested a second kind of help on the same programme. Some people were paid up to $25 a week for progress toward their weight goal, and the cash made no difference to weight the trial could detect.

Two kinds of help, one programme, one trial. The food moved the scale; the weekly cash didn't. That is one design of money, paid for weight lost over 18 months, and other deals come up later with different results.

You would expect the weight to come back when the deliveries end. In this trial they never ended. Every box went out on schedule, to the last week.

The weight started coming back at month six anyway, with the boxes still arriving. From month 6 to month 18, the food groups put back 3.7 kg and the groups without food 3.6 kg, going by averages nobody tested against each other.

The food groups still finished ahead at 18 months, by about half their one-year lead. So the food bought a lead, and the lead shrank while the food kept coming.

One more gap, and it's ours to declare. A written meal plan without the food, on paper or in an app, was never tested on its own by these studies. FitChef makes written meal plans, so we'd like that answer more than most. The research hasn't given it.

Both sides of that food comparison already sat in a group programme. So is the group itself doing some of the work, and is the one-to-one appointment people pay more attention to the better kind of support?

One-to-one feels like the premium choice: a person's attention, aimed only at you. It seems obvious it should take off more weight than a room full of other people trying.

Notice first that this is a different comparison from the food. Nothing was added inside a group here. The same kind of weight-loss programme was delivered two ways: to a group, or to one person at a time.

Across seven trials, the people in groups had lost 1.9 kg more at a year, and every one of the seven leaned the group's way, though not every trial was clear on its own. An older review, sharing no trial with it, pointed the same way.

For weight at a year, then, one-to-one was not the safer bet. If the lifting results made closer attention sound like the winner, that was a different goal: strength in lifting trials, not weight in weight-loss programmes. No study here put a weight-loss group up against a personal trainer.

Then the hours. Where the trials printed them, group programmes gave 12 to 55 hours of contact and one-to-one programmes 2.5 to 11. The option people pick for more attention came with less time.

So was it the company or the clock? The authors of the larger review offer both explanations and tested neither: other people trying beside you, or simply more hours, because a group lets a programme give everyone more time. They asked for a trial with equal hours on both sides. Neither review contains one.

The next "more attention" test is another different comparison, so it carries a different number. Five trials added a human coach on top of a weight-loss app and compared it with the same kind of app alone. The coached groups lost 1.80 kg more on average, and the review could not separate that from chance.

Why is 1.9 kg a result and 1.80 kg not? The width of the likely range. As the review printed it, every plausible value for the group result still favoured the group; the app-coach range ran from a real benefit to a little worse than nothing. And by its own report, the largest of the five trials added computer-picked messages rather than ongoing human counselling.

The authors' reasoning, untested, is that most tracking apps already do part of a coach's job (the log, the goals, the reminders), so a coach adds only what is left over.

A coach beside the bar added kilos to a squat. A coach on an app wasn't shown to take kilos off the scale. A group beat one-to-one. Which of those numbers is the one for you?

By now you are holding a contradiction. A coach worked, and a coach wasn't shown to work. Help added about 7 kg to a squat, doubled the weight lost, added 1.9 kg, or added nothing anyone could measure.

It dissolves once "coach" is split into its jobs. Beside the bar, a coach corrects a technical lift, judged in squat kilos. Inside an app, a coach answers to the bathroom scale. In a weekly group, the support is a whole weight-loss programme, judged on the scale too.

Pin each number to what it was compared with, what it measured and when, and they stop competing. Each row names the comparison first and the number second, and none can be added to or ranked against another:

• A coach beside the bar. Against a coaching app or a PDF, in people who already lift, asked to keep eating as usual, finishers after 10 weeks: about 7 kg more on the squat. Across six trials, mostly of people new to lifting: almost no difference to body composition.
• A coach on a weight-loss app. Against the same kind of app alone, at the first check after each programme: 1.80 kg, with a range from a real benefit to a little worse. Not shown. Against people given no app at all, app plus coach came out 2.27 kg ahead, which can't split the app from the coach.
• A weekly group instead of one-to-one. The same kind of programme delivered two ways, at a year: 1.9 kg more in the group, with more hours on its side and a margin probably less certain than it looks.
• Food handed over. On top of the same group programme: 9.1 against 4.5 kg at a year, still ahead by about half that at 18 months.
• Money, by design. Up to $25 a week for progress, on a group programme, for 18 months: no difference the trial could detect. A staked deposit or daily prize draw, against monthly weigh-ins: 14.0 and 13.1 lb against 3.9 lb at 16 weeks (about 6 kg against under 2).
• Cash on a free programme. For results or for steps, with monthly check-ins for everyone: about half and four in ten of the people reached a 5% loss by six months, against one in five.

Laid side by side, they show two things no single page can. No lifting trial cut anyone's calories, while every other line comes from a weight-loss programme. If you are dieting while you lift, the trainer line speaks to your strength and the others to the scale: no study here tested a trainer during a cut.

And a pattern runs down the list. Layers added on top of help people already had mostly came out close. That covers a coach on an app, more contact against less (about a third of a kilo apart at a year) and money on a group programme that was already working. Against a thin starting point, the same layers looked bigger.

The food is the exception, the one add-on that clearly moved the scale on top of an active programme. A pattern across differently designed trials, not a tested rule.

If most layers on top came out close on the scale, where did the help show up?

Start with the headcount you have already seen, the coach getting nearly every lifter to the end. The coaching app got almost as many there. But its finishers' squats grew about like the PDF group's. The app showed up in who kept going, not in what the finishers gained.

The free food came with more of the programme, not less. The people given meals turned up to more group meetings and filled in more food-diary days. The trial's own authors wondered whether free food was simply a reason to show up.

Computer-picked messages raised the number of days people hit their calorie target, without more weight lost at a year. Their authors wrote that engagement faded without personal contact.

Then money. Every money deal that helped came with company. People with money riding on their weight phoned it in every day and got a message back on their progress. In the biggest trial, the cash came with feedback, and its authors wondered whether that meant more human contact.

The one trial that added money and nothing else found no difference on the scale.

The belief that someone holding you accountable is what makes it work has one honest number here. People recruited with friends into a team programme, with money riding on the team keeping its weight off, kept their full loss far more often than people recruited alone: 66% against 24% of those assessed. Their average weight change didn't differ.

Stack those side by side and one shape appears. Across a coach, a coaching app, free food, computer-picked messages and money, help showed up first in who kept going and what people did. It showed up only sometimes in the kilos of the people who did it.

Almost all of that rests on shares and patterns nobody tested. And it doesn't mean support changes nothing: the food moved the scale, and the coach moved the squat.

Which leaves what the research never tested: the price, and the months after the help stops.

Back to the question you came in with. Not one of the 17 studies measured what anyone paid.

And the paid versions you would actually buy weren't the ones tested. The slimming-club places and the food were free to the people in the trials. The deposit scheme was run and topped up by researchers, and no trial tested a paid coach tier in a commercial app.

The other gap is time. The single strength trials ran 8 to 12 weeks. Both group reviews stop at a year. The app review used each trial's first check after the programme, and the food trial never stopped the food.

The money trials looked furthest past their paid period, and only by months: three months after the 16-week schemes ended, their lead was too small to be sure it was real.

The one longer look is a pair of single trials at four years, inside the meal-replacement review, comparing the groups rather than what happened once the help ended. What happens after the help stops is the least tested part of all five questions.

None of the five answers reached high certainty. Two are moderate: groups at a year, and provided food in the first year. Three are low: the trainer, the app coach and the money.

And if you are weighing an AI coach against a human one, no study here set the two against each other.