Subject
What was observed: answer, page, conversation or activity.
Glossary
The language of Reputation Intelligence
Understand what each observation measures before comparing brands or deciding how to act.
What you get
Definitions with a scope
A useful definition clarifies what is counted, which sources are included and what cannot be inferred. Read every metric alongside its period, market, language and sample design.
- Counted object
- Calculation base
- Limit of the conclusion
What you get
Observed mentions
Count eligible items in the selected corpus after attribution and deduplication. This is a count of observations, not the total number of people who saw or discussed your brand.
- Relevant content
- Duplicates excluded
- People reached not measured
What you get
Reviewed sentiment
Show positive, neutral, negative and mixed classifications alongside unclassified items. State the classified denominator: missing classification does not become neutral sentiment.
- Classified tone
- Mixed outcomes retained
- Unclassified is not neutral
What you get
Search presence and position
Search presence is the share of complete, eligible observations in which your brand appears; average best position considers only those where it is present. Engines remain separate. Query, market and device define the comparison, while incomplete observations are excluded rather than converted into brand absence.
- Presence in the sample
- Position when present
- Search engines kept separate
What you get
AI mentions, citations and recommendations
AI measures are calculated from complete, eligible answers, retaining provider, engine, model and prompt version. The denominator is visible. For example, ten mentions in twenty answers describe presence in that sample; they do not estimate how many people encountered your brand in assistant conversations.
- Answer conditions
- Counts for the three signals
- Visible denominator
What you get
Perceived trust
Perceived trust concerns expectations associated with a brand through documented experiences and communication: quality, consistency, support and transparency may contribute differently. It is not a financial index. Observed sources provide signals about these dimensions without representing everyone’s opinion or producing a universal trust percentage.
- Brand expectations
- Documented experiences
- No financial rating
What you get
Citation and recommendation
A citation refers to a source, while a recommendation expresses a preference in relation to the question. They are distinct signals. An answer may name a brand without citing its website, or cite a page without recommending its product; measures therefore retain separate counts and context.
- Mention: the name
- Citation: the source
- Recommendation: the preference
What you get
Comparable scope
Use the same channels, markets and time intervals for your brand and competitors. If scope changes, describe the difference alongside the comparison.
- Matching channels
- Matching markets
- Comparable time windows
What you get
Independent evidence
A story reproduced by several publications or a repeated URL may appear in multiple places. Check duplicates and source independence before treating repeated mentions as separate experiences.
- Original article
- News republications
- Independent experiences to verify
What you get
Record what was asked and where
Use consistent question sets for the brands compared. Separate results from different providers, models and prompt versions. An API answer does not automatically represent the consumer interface, a personalised session or questions asked by real users.
- Panel questions
- API configuration
- Separate consumer experience
Using definitions
The right words change the interpretation.
An indicator becomes useful when you know what it counts, its denominator and exclusions. Definitions help you read measures without giving them a broader meaning.
In TrustInsider, the sample accompanies the measure: valid answers, deduplicated articles and observed queries describe different sets. Comparing periods requires compatible objects and criteria. The following definitions extend glossary terms into an interpretation method.
Denominator
Which items are eligible for calculation and which are excluded.
Context
Period, language, market and method that make the measure interpretable.
Informed comparisons
Similar numbers. Different meanings.
Presence, traffic and reputation may be connected in an analysis, but they are not interchangeable. Each indicator answers a precise question.
A brand may be widely cited because of an issue; a visible page may receive few clicks; an AI answer may mention a product without recommending it. The table clarifies these differences before results are compared.
| Terms | Difference | Interpretation example |
|---|---|---|
| Mention and recommendation | Being named does not mean being recommended. | The model mentions two alternatives but suggests one. |
| Visibility and traffic | Appearing does not mean receiving a visit. | The page appears in results; clicks are a separate measure. |
| Change and effect | A change does not prove a cause. | The sample improves after a change, but checks are needed. |
Related paths
Choose your next step.
Check your brand’s presence, compare results or choose content improvements: continue with the feature you need.
From question to action
From the question to checking the outcome: context, sources, hypotheses and action plans in the TrustInsider analytical workflow.
ReadA transparent method, not a black box
TrustInsider’s criteria for interpreting reputation and trust: provenance, samples, reviews and the limits of conclusions.
ReadFind the right answer
Find guides, FAQs and methodology to clarify indicators, coverage and access, with dedicated routes for data requests.
Read