Below is the minimum observation standard from which one can begin building a cumulative database across brands, categories, and answer systems. The card is intentionally simple: it should be reproducible and suitable both for desk research and for regular monitoring within a product or marketing team.

The card's core principle is to record not only the fact that a brand appears, but the entire contour of the answer: the wording of the question, the user's intent, the language, the system, the type of sources, the brand's role, the presence of citations, the nature of category drift, the update lag, and the final practical interpretation. Otherwise, the database will quickly turn into a repository of attractive screenshots with no analytical depth.

Card fields

System and mode

For example: Google AI Overviews, AI Mode, ChatGPT Search, Copilot Search, Perplexity.

Date and geography of the research run

Record the date, locale, interface language, and country if it affects the results.

Original question

The exact wording of the user's question, without editorial changes.

Intent type

Informational, comparative, commercial, local, navigational, research-oriented.

Brand's role in the answer

Absent; mentioned; cited; shapes the frame of the answer; appears in the shortlist.

Source type and quality

Official, editorial, user-generated, institutional, catalog; strong or weak.

Signs of category drift

Whether there is task drift, a shift in the language of comparison, or the insertion of unrelated alternatives.

Signs of update lag

Whether the answer matches the latest known facts and where the delay likely came from.

Editorial conclusion

A brief interpretation: where the problem lies, where the strong area is, and what to check next.

Filled example

System #001

System and mode
ChatGPT (GPT-5.4, web-search mode)
Run date and geography
12 March 2026, Russia, Russian language
Original prompt
'Which analytics service is best for a mid-size e-commerce business with annual revenue of 50–200 M RUB?'
Intent type
Comparative, commercial
Brand role in the answer
Mentioned in a list of 5 options at position 4. Described in one sentence with no specific advantages. Not cited, does not influence the comparison frame.
Source type and quality
The model cited 3 external reviews (editorial, medium strength) and 1 catalogue source. The brand's official site was not cited. Reviews from 2024 — possible lag.
Category drift signs
Yes. The brand positions itself as an 'intelligent analytics environment for commerce', but the model reframed the task as 'reporting tools for online stores'. As a result, two services from an adjacent category (BI dashboards) appeared in the list — ones the brand does not consider direct competitors.
Update lag signs
The brand launched a new pricing plan in February 2026. The answer referenced the old pricing structure. Probable cause: the site update has not been picked up by the search index; external reviews have not yet reflected the change.
Editorial conclusion
The brand is functionally invisible in a key commercial scenario. Three problems: (1) category substitution due to a mismatch between brand language and demand language, (2) absence of external endorsements for the new positioning, (3) update lag on pricing. Recommended: rerun via Google AI Mode and Perplexity for comparison; check for comparison pages on the site; assess which external sources have already reflected the new pricing plan.

Related materials

Research article 7 min

The “answer bubble”: why the same brand looks different in ChatGPT, Google, Copilot, and other systems

Why there is no single AI visibility: the same brand can look noticeably different across ChatGPT, Google AI Overviews, Copilot, and Perplexity.

Open the material →
Research article 7 min

Update lag: how quickly AI systems change their view of a company after news, a product launch, or a price change

Why there is a time gap between a fact changing about a brand and its stable appearance in machine answers — and how to observe this lag in practice.

Open the material →
Next step

What a full report looks like, not just one card

An observation card captures a single run. A full AI100 report combines dozens of such observations into a comparative picture: group ranking, growth zones by storyline, and personalised action cases.

Open the sample report →