trustinsider
English
Try with your brand

Methodology

A transparent method, not a black box

Reputation depends on context. Read the sources, observation period and sample limitations before interpreting a result.

What you get

Define the question first

Choose the brand, product or service, market, language, sources and observation period. Confirm that each observation concerns the correct entity. Competitor comparison requires the same scope and comparable coverage; a familiar name is not enough.

Every indicator answers a different question
PerspectiveObject observedBasis for interpretation
Conversations and sentimentContent attributed to your brand and judgements expressed on topics.Relevant corpus, duplicates, reviews and unclassified share.
SearchResults collected for a defined query set.Engine, context and eligible runs; position distinct from frequency of presence.
AI answersDescriptions, mentions, citations and suggestions in sampled answers.Questions, model and version; incomplete outcomes and refusals remain identifiable.

What you get

Observed mentions

Measurement starts with the counted unit: content, mention or documented experience. The method should record confirmed attributions, exclusions and duplicate groups so a review can reconstruct why one item enters the total and another does not.

Reading path
  1. Content attribution
  2. Duplicate checks
  3. Eligible mentions

What you get

Reviewed sentiment

The distribution uses items classified under the same criteria version as its denominator. The report displays unclassified items separately and retains any mixed outcomes. A classification change needs an identifiable version before two periods are compared.

Reading path
  1. Versioned criteria
  2. Reviewed classification
  3. Classified sample

What you get

Search presence and position

Search coverage uses complete, eligible observations as its denominator; average position uses only those with confirmed presence. The method retains engine, query, market and device to prevent a denominator change from being interpreted as domain movement.

Distinct perspectives
  1. Presence: eligible observations
  2. Position: confirmed appearances
  3. Comparison: matching context

What you get

AI mentions, citations and recommendations

For each AI indicator, the method states the observed event and valid denominator: brand mention, recommendation or source reference. All three may coexist. Review preserves the answer and collection conditions needed to verify each attribution.

Distinct perspectives
  1. Brand name detected
  2. Source cited in the answer
  3. Preference expressed

Origin and meaning

Every data point answers a question.

TrustInsider connects different observations while preserving their origin. An article describes a public narrative; an organic result shows a page encountered in a search; an AI answer shows what a system generated under recorded conditions.

These perspectives may help reconstruct the same topic, but they do not share a measurement unit. The method separates conversations, searches, answers and authorised property data. Connections between evidence must be verifiable: the same topic appearing twice does not prove that one item caused another observation.

Complementary sources. Different units.
OriginWhat it describesWhat it does not prove
Press, web and conversationsRelevant content available through authorised sources.Every customer’s opinion or the entire market conversation.
Google or Bing organic searchResults observed for a query, market, device and time.Visits, conversions or visibility on other engines.
Sampled AI answerMention, recommendation and citation in the collected answer.How many people saw that answer or a universal position.
Customer propertyMeasures exposed by connected systems with the owner’s consent.Unmeasured journeys or certain attribution of visits without an identifiable source.

From collection to measurement

A number with its basis.

Every indicator should be traceable to what was counted. Period, market, language, sources, inclusion criteria and exclusions define the result’s meaning.

In AI sampling, rates include complete, reviewed answers; errors, refusals and unreviewed answers remain separate. In organic search, presence refers to complete results, while average position considers only searches where the brand appears. An empty base does not produce a reputation score of zero. TrustInsider’s calculations are deterministic and versioned; the model can help interpret them without inventing the value to display.

Numerator

The number of observations satisfying the stated condition.

Denominator

The eligible set over which that condition is measured.

Coverage

Planned collections, available outcomes and limitations that qualify the result.

Interpretation and review

From measurement to a reasoned decision.

The result helps select a check or action. Moving from measurement to decision should make evidence, hypotheses and alternative explanations explicit.

A proposal may concern an incomplete page, inaccurate description or recurring experience problem. Your team examines the current version before approving the plan. A source, panel or method change should accompany the comparison; the subsequent outcome remains distinct from proof of work completed. Comparison describes what changed in available sources and keeps hypotheses about causes separate.

Reading path
  1. Verifiable evidence
  2. Hypotheses and alternatives
  3. Reasoned decision

Related paths

Check your brand’s presence, compare results or choose content improvements: continue with the feature you need.

Apply the method to your brand
A transparent method, not a black box · trustinsider