Press · Library
Public Intelligence Library
Every public work of The Observatory — readings, context, method, reference and actors — ready to find, cite and reuse.
Every public work of The Observatory, in one searchable place: the governed readings served today, the contextual readings of each market, the method and reference that govern them, and the organisations they observe. Each entry states what it is, what it covers and when it is from, and cites it in the form its nature allows. The source page remains authoritative — this library describes it, never replaces it.
Public tables and prepared capsules may be reused free of charge, including commercially, under our Terms. Free reuse, including commercial republication, under The Observatory’s Terms. Retain attribution, source link, scope and date. See the citation and reuse standards. Observed by The Observatory — not approved or endorsed by it.
Find, cite and reuse Observatory intelligence.
Search across published evidence, context and reference works. Narrow the library by market or work type, open the original source, cite a work in the format you need, or build a selection to export for research.
Find → Cite → Build a selection → Export / Follow
- AI Visibility
AI visibility concerns whether an actor appears on a named AI answer surface within a stated scope and window. Presence alone does not establish citation, reference treatment or an explicit recommendation.
Last reviewed 3 June 2026
- AI Visibility vs AI Authority
Appearing on an AI surface is not the same as being treated as a reference. How to tell visibility from authority, and what an authority gap indicates.
Last reviewed 3 June 2026
- AI-Shaped Search
The ecosystem of answer-generating surfaces — AI Overviews, large-language-model answers, and source panels — through which audiences increasingly discover and compare brands.
Last reviewed 3 June 2026
- AIO Opening
The AIO Opening is the actor or source identified as opening an AI Overview under the applicable capture rule. It is read separately from citation order, organic rank and any explicit recommendation.
Last reviewed 3 June 2026
- Answer Engine Optimization (AEO)
Answer engine optimization (AEO) is a market label for practices concerned with visibility in answer-generating systems. The label alone does not establish what was measured or whether an intervention changed the observed result.
Last reviewed 3 June 2026
- Answer Layer
The answer layer is the surface where AI assistants, AI overviews and search engines decide which brands and actors get named, cited and framed. The Observatory reads this layer market by market.
Last reviewed 3 June 2026
- Answer Surface
Any surface where a response is assembled and displayed, including AI Overviews, LLM answers, organic results, and source panels.
Last reviewed 3 June 2026
- Attestation Standard
Scope, Basis, Window, Source and Limitation tie an Observatory reading to its evidence and bounds. Attestation supports traceability, not independent assurance.
Last reviewed 3 June 2026
- Attestation Standard
The disclosure attached to a reading: scope, basis, window, source, and limitation.
Last reviewed 3 June 2026
- Authority Gap
An Authority Gap is strong observed visibility without corresponding observed reference treatment on the same surface, within the same scope and observation window. It does not compare an actor with its assumed commercial standing.
Last reviewed 3 June 2026
- Battles
Battles is the desk of head-to-head comparisons — platform versus operator, one model versus another, one market versus another — that point back to the relevant ranking rather than creating their own.
Last reviewed 3 June 2026
- Benchmark Governance
Why method stability matters: market-level rules, version context, and the distinction between methodological changes and observed movement.
Last reviewed 3 June 2026
- Benchmark Governance
Benchmark governance is the set of rules that keep The Observatory's method stable and neutral.
Last reviewed 3 June 2026
- Brand Mention vs Citation
A mention names a brand in an answer; a citation explicitly attributes support to a source, usually through a cited URL. A cited third-party page about a brand is not the same source relationship as a citation to the brand's own website, and neither alone establishes a recommendation.
Last reviewed 3 June 2026
- Business Market vs Positioning Surface
The business market is where commercial activity happens: who operates, transacts, and serves customers in a category. The positioning surface is where visibility, citation, and framing are contested in AI-shaped search.
Last reviewed 3 June 2026
- Cadence
Cadence is the recurring rhythm of observation or publication. A single observation and a period aggregate are different reading bases: a period brings comparable observations together, with its actual coverage and limitations. Aggregation alone does not establish stability or a trend.
Last reviewed 3 June 2026
- Can an AI Know Why It Ranks a Brand?
A language model can describe possible ranking factors, but cannot reliably state current cross-surface behavior for a live market without governed observation.
Last reviewed 3 June 2026
- Canonical Glossary
Stable definitions for AI-search market reading: answer surfaces, semantic authority, signal styles, citation surface, Market Authority Score, and more.
Last reviewed 3 June 2026
- Capture
A Capture is the recorded source output and provenance from which a governed Observation is established. A Capture is not automatically a public raw-data payload.
Last reviewed 3 June 2026
- Citation Surface
The citation surface is where an AI answer attributes its sources. Being cited here is distinct from ranking organically and from holding the AIO opening — three different forms of presence.
Last reviewed 3 June 2026
- Comparability level (Model Watch)
A level — high, moderate, low or degraded — assigned by the AAI Kernel delta synthesis to a pair of readings. It is lowered when the two readings answer different subjects or countries, or when a live-web side returned no citations. It qualifies how far the two sides can be set against each other, not the market.
Last reviewed 3 June 2026
- Confidence level (Model Watch)
A level — high, moderate or low — assigned by the AAI Kernel delta synthesis to its own reconciliation of two readings: how clearly the recorded statements support the convergences and divergences it reports. It is not a measure of the market or of either model's accuracy.
Last reviewed 3 June 2026
- Contested Band
The contested band is the range of positions on a surface where no actor holds a stable lead and ranks shift run to run — distinct from the settled top and the periphery.
Last reviewed 3 June 2026
- Deep Narrative Capture (DNC)
Deep Narrative Capture records the response of a named AI system or answer product about a market, with its mode or surface identified verbatim. Its governed public reading may select, contextualize and interpret that response without silently changing it; narrative framing remains separate from measured position and scores.
Last reviewed 3 June 2026
How each kind of work is cited
- Measured readings — rankings and Query Spotlight
- Governed measured reading. Cite it with its market, surface and observed window. The published metric table is available as data.csv and the complete record as citation.txt.
- Narrative readings — DNC Lens and Model Watch
- Governed narrative reading. Model statements are recorded verbatim and fact-checked; no metric table exists, so no CSV is offered. The complete record is available as citation.txt.
- Context, method and reference
- Editorial and contextual work. Cite it as context; for a factual or ranking claim, cite the dated governed readings it links to.
- Actor records
- Living record. Cite it with the date you retrieved it. For a ranking or factual claim, cite the dated evidence the record links to — never the record in its place.
Dates mean what they say: Observed is the window the surface was read; Published is when a reading went public; Text revision is when the exact accepted text first existed; Last reviewed is the review date a page declares. No digital object identifier is assigned; cite the canonical address.
Follow what matters to you
Email follow covers a market and The Observatory Brief. To follow a work type, a market type or an organisation, subscribe to its RSS or JSON feed — every library view offers the feed that matches it.