Brand Intelligence
TheObservatory

The Observatory / Method

Publication Record

Method Updates, Corrections & Editorial Changes

This record lists, by date, the changes that affect how The Observatory's published readings should be read: updates to the measurement method, material corrections to published readings and significant editorial changes.

A method update applies from a stated observation window. Readings produced before it keep the method under which they were produced and are not recalculated. A material correction changes a published figure, scope, attribution, identity or claim. Routine production, maintenance and technical operations are not listed here.

Types of entry

Method Update
A measurement rule changes from a stated observation window onward. Earlier readings are unchanged.
Material Correction
A published reading contained an error in a figure, scope, attribution, identity or claim, and was corrected.
Editorial Update
A significant change to the explanatory or editorial content accompanying the readings.

November 2026

Method Update

AI Opening measurement and canonical actor identity

Effective
Observation window of November 9–15, 2026, and subsequent windows governed by OPENING_V2.
First readings published
Entry published

Two method changes adopted on November 13, 2026 apply from the observation window of November 9–15, 2026: a revised AI Overview opening measurement, and the application of governed canonical actor identity to the weekly AI Opening, AI Citation and Query Spotlight readings.

AI Opening measurement update

For observation windows through November 8, 2026, an actor's presence in the opening of a Google AI Overview was identified within the first 500 characters of the answer. From the window of November 9–15, 2026, it is identified within the first substantive prose block of the observed AI Overview. The measurement recognizes governed actor names and aliases, identifiable domains, and links or references attributable to that opening block; references that appear only in the general source panel receive no opening credit. Each actor receives at most one presence credit per capture, however many times it is mentioned in the same opening, and credits are accumulated across all eligible captures of the observation period. Readings for earlier windows keep the method under which they were produced and are not recalculated, and readings produced under different methods are not compared directly. The measured population, the share denominator, Presence and the tie-breaking rules are set out on the Method page.

See Method — AI Overview Ranking — Top 10

Canonical actor identity update

From the same window, the weekly AI Opening, AI Citation and Query Spotlight readings identify actors under the governed canonical identity rules that the Organic ranking already applied: an organisation that directly operates several listed domains is counted and named as one actor, while franchisees and independent operators remain distinct. This update concerns how actors are identified and named; it does not change the ranking formula of any family. Readings for earlier windows keep the identity rules under which they were produced.

See Method — Observed actors

No material corrections are currently listed in this publication record.

Corrections and right of reply

Readers and the organisations named in a reading can report a factual error or send a response through Right of Reply. The corrections policy is set out in Method — Corrections and right of reply.