THE INDEX OF EVIDENCE

What we believe. How strongly we believe it. Why.

Search has always attracted certainty faster than evidence.

AI has accelerated the problem.

The Index of Evidence is a living record of claims that affect search strategy, our current position on them and the evidence strong enough to change our mind.

Confidence Scale Search Claims Audit Living Research Record

Method

A confidence scale, not a truth machine.

Every claim in the Index is evaluated against four explicit confidence tiers:

CONFIDENCE LEVEL: STRONG

STRONG

Supported by repeatable evidence, credible research or extensive observation across relevant environments.

CONFIDENCE LEVEL: DEVELOPING

DEVELOPING

There is meaningful evidence, but causality, scale or generalisability remains uncertain.

CONFIDENCE LEVEL: LIMITED

LIMITED

Plausible or frequently repeated, but current evidence is insufficient for us to treat it as a strategic assumption.

CONFIDENCE LEVEL: CHALLENGED

CHALLENGED

Available evidence currently points against the claim, or the claim is materially overstated.

Structure

Evidence entry format

Each record in the Index of Evidence contains the following mandatory schema fields:

Field 01

Claim

The specific search statement or hypothesis evaluated.

Field 02

Current Position

Strong / Developing / Limited / Challenged.

Field 03

Last Reviewed

Timestamp of the most recent evidence check.

Field 04

Applies To

Engine, platform, market, or architecture scope.

Field 05

Our Read

Detailed strategic synthesis and practical implications.

Field 06

What Would Change Our Mind

Explicit criteria or test data required to shift confidence.

Field 07

Sources

Primary documentation, patents, experiments, or observational logs.

Research Queue

Claims currently queued for review

These are framed explicitly as hypotheses to research before publication, not conclusions:

Claim 01

Traditional organic rankings influence visibility in generative search.

Claim 02

Brand mentions can matter independently of followed backlinks.

Claim 03

llms.txt materially improves brand visibility in major AI assistants.

Claim 04

Structured data directly increases the likelihood of being cited by LLMs.

Claim 05

Reddit is an important discovery and evidence surface for conversational search.

Claim 06

Content freshness is a universal ranking advantage.

Claim 07

Publishing more content increases topical authority.

Claim 08

Zero-click search makes SEO traffic an inadequate measure of search value.

Claim 09

Entity consistency across third-party sources matters to AI retrieval.

Claim 10

AI-generated content is inherently disadvantaged in organic search.

Start Here

Test your search assumptions against market evidence.

Start with an Index Review to evaluate your technical, structural, and AI search performance.