The research behind the plugin’s recommendations. Each row: a claim, the strongest study on it, and what organic-os actually does about it. Tiers are strong, moderate, and weak – popular tactics that fail controlled tests are labeled, not hidden. Canonical source: docs/evidence.md. This site’s own live results are on the proof page.
Strong evidence
| Claim | Evidence | What organic-os does |
|---|---|---|
| Server-rendered HTML is required for AI crawlers to see your content | Across a billion-plus requests, no major AI crawler executed JavaScript (Vercel and MERJ) | The onsite audit flags client-side-only rendering as a blocking issue, not a suggestion |
| Named statistics, quotes, and cited sources measurably increase AI citation | 10,000-query controlled experiment (GEO, KDD 2024): named quotes +40.9%, sourced statistics +30.6%, inline citations +27.5% citation probability | The content pipeline requires a named statistic or quote with a real source before a draft passes QA |
| Bing indexing determines most ChatGPT search citations | Seer Interactive found 87% of SearchGPT citations matched Bing’s top organic results | The onsite audit checks Bing indexation status, not only Google’s |
| Fresh content is cited disproportionately across AI engines | AI-cited content is 25.7% fresher on average; 76.4% of ChatGPT’s top citations were updated in the prior 30 days (Semrush) | The monthly audit flags content older than 12 months without an update as a freshness gap |
| Extractable structure helps passages get lifted into AI answers | Consistent across the GEO paper’s ablations and industry crawler studies | Answer-first paragraphing and a TL;DR capsule are the default draft shape |
Moderate evidence
| Claim | Evidence | What organic-os does |
|---|---|---|
| Community discussion (Reddit and similar) contributes to AI answer sourcing | Widely observed in citation mixes, but without controlled measurement | The citation tracker records it; the plugin does not recommend seeding community posts |
| Covering the fan-out of related sub-queries improves overall visibility | Consistent with how retrieval-augmented answer engines assemble context; reported across vendor studies | Keyword gap analysis includes adjacent queries alongside the primary seed |
| Third-party listings help entity-level queries get answered correctly | Observed in entity-query citation mixes; not rigorously isolated as causal | Missing listings are noted as a finding, not a mandated fix |
Weak or contradicted
| Claim | Evidence | What organic-os does |
|---|---|---|
| Adding schema markup increases AI citations | 1,885 pages that added JSON-LD vs 4,000 controls: no platform showed a meaningful citation increase (Ahrefs) | Ships schema anyway – it holds up for Google rich results; the AI-citation case is stated as unproven |
| Publishing an llms.txt file improves AI visibility | Of 137,000 sites with an llms.txt file, 97% got zero bot requests to it; no AI provider has committed to parsing it (Ahrefs) | Available as an optional, low-cost hedge; never presented as a visibility tactic and not recommended by default |
How this ranking gets used
The content pipeline recommends only strong-tier tactics as load-bearing; the audits use the strong-tier checks as criteria. The table is reviewed, not fixed: when a new controlled study changes a tactic’s tier, the row moves.
Found a study that changes a row? Evidence updates with primary sources are a welcome contribution.