How to Spot Bad Fitness Advice
Most bad fitness advice shares a handful of tells. Learn them once and you can filter the internet, including this wiki.
Last updated: 30 September 2026
Overview
You are going to be given far more fitness advice than you can evaluate. The useful skill is not knowing every answer; it is having a filter fast enough to use on a phone.
This article is that filter. It also applies to this wiki, which is the point of the sources listed at the bottom of every article here.
Quick Answer
The most reliable warning signs of bad fitness advice are a claim of secret or suppressed knowledge, a single mechanism used to explain everything, an unnamed study, a product attached to the claim, and certainty about something the research is genuinely unsure of.
Key Points
- Anyone who is certain about a contested question is not reading the literature.
- "Studies show" without a named study means nothing.
- A meta-analysis beats one trial. One trial beats an anecdote.
- Check who is selling something, including attention.
- Mechanism is not outcome. Plausible is not proven.
The tells
None of these prove a claim is false. All of them mean slow down.
| Tell | Why it is a tell |
|---|---|
| "They don't want you to know" | Nutrition research is public and free to read |
| One weird trick | Real effects are small and cumulative |
| A single mechanism explains everything | Insulin, cortisol, inflammation, "toxins" |
| "Studies show" with no study named | The study, if it exists, usually says less |
| A product attached to the claim | Not automatic disqualification. Always relevant |
| Before and after photos | Lighting, water, pump, time of day, and selection |
| Certainty on a contested question | The literature is not certain. They are guessing |
| "Natural" as an argument | Not a property that predicts anything |
| Elite athlete does it, therefore | Different genetics, drugs, time and job |
| Detox, cleanse, reset, flush | Your liver and kidneys are already doing this |
| Muscle confusion | Muscles have no opinions |
The most reliable single one is the seventh. On most interesting questions — optimal protein distribution, ideal frequency, how much cardio interferes — the honest answer contains "probably" and a range. Somebody who gives you a number to two decimal places is performing confidence.
What a study is worth
Not all evidence weighs the same, and the difference is large enough to matter.
| Strength | Type |
|---|---|
| Strongest | Meta-analysis or systematic review of randomised trials |
| Strong | Large, long randomised controlled trial |
| Moderate | Small randomised trial |
| Weak | Observational or cohort study |
| Weaker | Animal or cell study |
| Weakest | Anecdote, testimonial, expert opinion alone |
Most viral fitness claims are built on the bottom three rows presented as the top one.
Five questions to ask about any study
When someone cites research, these five answer most of it:
1. Who was studied? Untrained undergraduates, elite athletes and mice respond differently. A supplement that works in untrained people often does nothing in trained ones.
2. How many, for how long? Twelve people for four weeks is a pilot, not an answer.
3. Was it randomised and controlled? Without a control group you cannot separate the intervention from time, enthusiasm and regression to the mean.
4. Who paid for it? Industry funding does not make a study wrong, and it is worth knowing.
5. Does the abstract say what the headline says? Frequently not. This is the single highest-yield check available, and the abstract is usually free.
Mechanism is not outcome
The most sophisticated-sounding bad advice works by describing a real mechanism and then assuming the outcome.
"Insulin stores fat, therefore avoiding carbohydrates at night prevents fat gain." Every clause about insulin is true. The conclusion does not follow, and when it was tested, meal timing did not do what the mechanism predicted.
The pattern to watch for: a true statement about physiology, a plausible leap, and no outcome data. If someone explains *why* something should work at length and never mentions whether it *did*, that is the tell.
Where to check
All free, all quick.
| For | Use |
|---|---|
| Any research claim | PubMed — search the topic, filter to meta-analysis |
| Systematic reviews | Cochrane Library |
| Health and medical claims | NHS Health A to Z |
| Supplements | Position stands from the ISSN, ACSM or NSCA |
| Nutrition in the UK | The British Dietetic Association |
A practical trick: search PubMed for your topic plus "meta-analysis". If a meta-analysis exists and contradicts the claim, you are done in about ninety seconds.
Credentials, and their limits
A relevant qualification is weak evidence in favour. It is not strong evidence, and its absence is not disqualifying.
| More reliable | Less reliable |
|---|---|
| Cites sources you can check | "Trust me, I've been doing this 20 years" |
| Says "I was wrong about that" | Never changed a position |
| Gives ranges and conditions | Gives one number for everyone |
| Explains who advice does not suit | One protocol for all |
| Relevant qualification, and cites work | Qualification used as the whole argument |
The third and fourth rows do more work than the credential does. Someone who tells you the limits of their own advice is telling you they understand it.
How to check this wiki
Fair is fair.
Every article here lists its sources, and they are links rather than names, so you can open them. Where the evidence is contested, the articles say which way each study went — training frequency reports both meta-analyses because they disagree, returning after a break says the mechanism everyone cites is not established in humans, and diet breaks states the population its strongest trial actually studied.
If you find an article here that is confident where the literature is not, that is a fault in the article. The edit and suggestion system exists for exactly that.
Common Mistakes
| Mistake | Why it causes problems | Better approach |
|---|---|---|
| Trusting confidence | Certainty is the cheapest thing to fake | Prefer sources that hedge honestly |
| Accepting "studies show" | Usually one small study, often misread | Ask which study |
| Following what a pro does | Different drugs, genetics and job | Follow what works for amateurs |
| Believing mechanism over outcome | Plausible is not proven | Ask whether it was measured |
| Judging by physique | Physique reflects genetics and time | Judge by reasoning |
| Dismissing everything as bro science | Overcorrection, equally unhelpful | Weigh the evidence |
Frequently Asked Questions
How do I know if fitness advice is good?
Check whether it names sources you can open, gives ranges rather than single numbers, admits what is uncertain, and has nothing attached to sell.
Is a single study enough to change what I do?
Rarely. Look for a meta-analysis or systematic review. A single small trial is a reason to be interested, not to act.
Does industry funding make a study wrong?
No, but it is worth knowing about, and it is worth more scepticism when a funded study is the only evidence for a claim.
How can I check a claim quickly?
Search PubMed for the topic plus "meta-analysis" and read the abstract's conclusion. It takes a minute or two and settles most arguments.
Should I trust someone because they look fit?
No. A physique reflects genetics, years of work and sometimes drugs. It says nothing about whether their reasoning is sound.
Sources
PubMed, the primary index of biomedical literature - https://pubmed.ncbi.nlm.nih.gov/
Cochrane Library systematic reviews - https://www.cochranelibrary.com/
NHS Health A to Z - https://www.nhs.uk/conditions/