Field NotesAnchor 08 · The method

The eight Discovery Signals.

The AI Discovery Score is computed across eight signals. Each measures a distinct property of how AI systems treat a subject, and each is weighted in the Index methodology. This anchor is the institution's plain reading of the eight — what each measures, why each matters, and what good looks like.

Field
The method
Reading time
11 minutes
First published
10 May 2026
Reviewed
Methodology Council
The institution's position

The eight signals are not a checklist. They are an instrument. A subject that scores well across all eight is a subject the AI will reliably surface; a subject that scores well on a few and poorly on the others is a subject whose visibility is fragile under any change to the underlying systems.

01 · Mention Frequency

How often the subject appears in the source set the AI systems consult. Mention Frequency is the most visible signal and the easiest to game; the institution weights it lower than its visibility would suggest. A high mention count from low-quality sources counts for less than a small number of mentions from Tier I sources.

02 · Authority

The standing of the subject's own properties — the website, the author profiles, the credentialing pages — inside the broader authority graph. Authority composites several inputs: domain age and continuity, structured data accuracy, named credential visibility, and editorial voice consistency.

03 · Citation Source Quality

Where Mention Frequency counts mentions, Citation Source Quality grades them. A mention in the Australian Dental Association editorial counts more than a mention in a generic comparison aggregator. Source Quality is graded against the published Source Register, and weighted accordingly.

04 · Entity Resolution

How cleanly AI systems can identify the subject as a single entity. Entity Resolution measures the consistency of the subject's name, address, identifiers, and trading variations across the sources the systems consult. A subject that appears under three different identities is, to the system, three different subjects — each weakly authoritative.

An observation from engagement

Across the institution's engagements, Entity Resolution is the signal whose improvement produces the fastest observable lift on AI surfaces. It is also the signal whose deficit teams most consistently underestimate before audit.

05 · Schema and Structure

How accurately the subject's properties expose machine-readable structure to the systems that consult them. Schema and Structure measures the fit between the schema applied and the page's actual content, the depth of structured data across the property, and the cross-link consistency between schema-described entities.

06 · Authority Graph Distance

How many hops the subject is from a small set of authority anchors — recognised professional bodies, accredited registers, university and government domains, established editorial publications. The Index measures distance directly through the published graph, with weighting calibrated per system.

07 · Behavioural Reinforcement

The recency-weighted volume of behavioural signal around the subject — reviews, commentary, ongoing editorial mention, and (where available) AI system citation density over time. Behavioural Reinforcement is the most volatile of the signals and the one that responds most quickly to engagement effort.

08 · Recency Stability

How stably the subject has held its standing across recent quarters. A subject that scored well last year and scores well now is weighted more heavily than a subject that has just begun to appear. Recency Stability is the institution's discipline against rewarding flash-in-the-pan visibility.

A closing observation

The signals are not a sequence; they compound. The subject that improves three signals while neglecting five is brittle. The subject that improves all eight is durable, and the durability is the point.

The institution's engagements are designed to lift all eight signals, in order, over a twelve-month course of work. The Score is read quarterly, and the work is sequenced against the signal whose deficit is largest at the read.

Read the Index

The Index applies the signals at scale.

This anchor defines the eight signals. The Index measures them across thousands of subjects each year.

Read the Index →