Outcome-First Evidence Test
Check human outcomes before trusting an attractive biological mechanism.
- Difficulty
- Advanced
- Time to result
- ~days to results
- Steps
- 5
- Confidence
- 99%
Norton recommends starting with the outcome people care about, then asking whether controlled human research measured it directly. A biological mechanism can explain an outcome once observed, but it cannot establish that the outcome occurs because many pathways operate and compensate simultaneously. Separate proxy measures, such as a transient hormone change or fat oxidation, from outcomes such as body-fat loss or a validated longer-term biomarker. Prefer randomized comparisons because self-selection and bundled lifestyle differences can distort observational groups. Check whether major variables were matched and whether the tested population and duration fit the claim. When direct outcome evidence is absent, retain the mechanism as a hypothesis rather than presenting it as established benefit or harm.
Origin
Norton articulates the distinction while explaining why higher fat burning on a ketogenic diet does not necessarily produce greater fat loss. Extracted from The Diary of a CEO.
Core principles
- 01A plausible mechanism does not guarantee a meaningful outcome.
- 02Biological pathways can compensate for one another.
- 03The preferred evidence measures the outcome people actually care about.
- 04Randomization helps reduce systematic differences between comparison groups.
How to run it
- 1
Define the real outcome
State the result that would change the decision, such as body-fat loss, symptoms, performance, or a validated clinical marker. Avoid substituting an easier proxy without saying so.
Pro tip Phrase the outcome before reviewing the proposed mechanism.
Watch out A surrogate marker may not predict the clinical outcome.
- 2
Separate the mechanism
Write the biological pathway as a proposed explanation rather than a result. Identify what would need to be true for it to affect the chosen outcome materially.
Pro tip Ask what other pathways might offset the effect.
- 3
Find direct human evidence
Look for human trials that measure the actual outcome and compare a relevant intervention with a suitable control. Give greater weight to converging controlled results.
Pro tip Check the population, dose, duration, and outcome definition.
Watch out Animal and mechanistic studies may not translate directly to humans.
- 4
Inspect the comparison
Check randomization, matched variables, adherence, and whether groups differed in other behaviors. Distinguish correlation from intervention evidence.
Pro tip Ask whether self-selection could create healthy-user bias.
Watch out Randomization reduces bias but does not make every trial flawless.
- 5
Calibrate the conclusion
State what the outcome evidence supports and what remains uncertain. Use the mechanism to interpret a result, not to inflate the result beyond the data.
Pro tip Say explicitly when the evidence is observational or speculative.
Watch out Do not turn population evidence into individualized medical advice automatically.
In the wild
Norton agrees that a ketogenic diet increases fat oxidation, then distinguishes that mechanism from net loss of body fat. He says higher-fat intake also increases the amount of dietary fat available for storage, while calorie- and protein-matched trials show little meaningful fat-loss difference between low-carbohydrate and low-fat diets.
→ The intervention is judged by measured fat loss rather than by one pathway that sounds favorable.
Common mistakes
Promoting the proxy
A change in fat oxidation, insulin, cortisol, or another pathway is not automatically the outcome the person cares about.
Ignoring compensation
The body contains interacting pathways, so one directional mechanism can be offset elsewhere.
Treating randomization as magic
Randomization improves group comparability but trial duration, adherence, measurement, and relevance still require scrutiny.
Is it for you?
Best for
Readers, practitioners, and communicators assessing nutrition, exercise, or supplement claims.
Not ideal for
Making an individual medical decision without a qualified clinician or dismissing all mechanistic research as useless.
From the transcript
“fat loss and fat burning or fat oxidation are not the same thing”
“just because there's a mechanism doesn't mean there's an outcome”
From the episode
The Calories Expert: "Health Experts Are Wrong About Calories!" "The Surprising Benefits Of Diet Coke!" & "The Link Between Obesity & Past Abuse!"