The AGD evidence ladder separates platform documentation, methodical proprietary data, controlled experiments, repeated observations, hypotheses and folklore. Every claim should retain provenance.
01STORY BRIEFThe idea in three screens.3 SCREENS
A hypothesis needs a label, not a disguise.
The AGD evidence ladder separates platform documentation, methodical proprietary data, controlled experiments, repeated observations, hypotheses and folklore.
Documentation may confirm a feature exists without proving its effect on reach. Internal observation may reveal a pattern without establishing a cause. Write the exact proposition supported by each source. Evidence level belongs to the claim, not author prestige or the number of links below it.
During review, underline verbs asserting a cause: increases, reduces, causes. Ask what evidence supports each one. A descriptive rewrite may be more accurate: a difference is observed and its cause remains unknown. This sentence-level work is more useful than a global “data-driven” badge above the article. Readers need to know which proposition a source actually supports, especially when one paragraph combines documented features with untested performance hypotheses.
Make the claim fit the material underneath it
A platform document can establish that a feature exists. A public post can show how a creator used a format. An internal log can establish that an operation completed. A controlled experiment may support a narrower causal statement about an outcome. These are different kinds of evidence, not interchangeable decorations for the same headline. Before publishing a claim, ask what the cited material actually contains and whether a reader could follow the link and reach the same limited conclusion. Remove the part that depends on invisible assumptions.
Match the sentence to the evidence
- 01Capability
Official documentation describes what is supported.
- 02Observation
A dated record shows what was visible.
- 03Causality
A suitable design tests competing explanations.
Editorial model: a way to reason about the process, not measured platform results.
Use explicit verbs. Documentation describes a capability; an observation shows what was visible at a stated time; a comparison suggests an association; a well-designed test can support an effect within its scope. Keep hypotheses interesting without laundering them into facts. For social examples, separate the observable hook from source-reported context and from commercial outcomes nobody has supplied. This discipline does not make an article timid. It lets the writer take a strong editorial position while giving the reader a clean boundary between judgement and evidence.
“A hypothesis needs a label, not a disguise.”
02FIELD KITFrom concept to practice.2 SCREENS
Attach a source or label to each claim.
Adapt this to your audience, budget and test scope.
Put it to work.
- 01Attach a source or label to each claim.
- 02Keep unmeasured effects unknown.
Notes & sources.
The limit. A promotional source does not become independent because another publication cites it.
Further reading is linked throughout this article. Worked examples explain a process; they are not campaign results.
THE ORGANIC CLUB is an editorial project initiated by TokPortal. That relationship is separate from the evidence behind claims. Editorial signatures and image credits are explained in our policy. Our editorial policy →
