User manual

Read results and their limitations

Before deciding what to do, check which brand, period and sample a result describes. Results describe what was observed; they do not automatically represent every search or every possible answer.

Section 1

Check the context before the value

Check the brand, period, source and market for the reading you are viewing. Context tells you which audience and answers the value refers to; without it, a comparison can mix different things.

After changing the period or context, wait for the new reading to load before comparing results. Compare the same scope: a change of source or audience can change what the comparison means.

The context of a reading in AI Visibility: the period observed, the comparison with the previous period and the topic chosen among the panel’s, before looking at any value. (fictional brand, demo data)
Demo data · Ilmarde Audio

Section 2

Read the result alongside coverage

Look for the observed base and any exclusions or limitations beside the value. Coverage shows how broad the sample behind the result is and where it stops applying.

The number of collected answers describes the available sample, not the volume of all your audience’s conversations. Before interpreting a share, check which questions and answers support it.

The overview’s “Analysis context” panel: workspace, brand and period, coverage by market and language, the status of the collections and the latest observation. (fictional brand, demo data)
Demo data · Ilmarde Audio

Section 3

Distinguish missing data, zero and errors

An unavailable reading does not mean a zero result. Loading, errors and missing data are different states: telling them apart keeps you from reading a technical problem as a conclusion about the brand.

Wait while a reading is loading. If it fails, use the retry command when available. If a collection or connection is missing, follow the message shown; do not read an empty card as absence of the brand in the market.

Four states of a card without a value

An empty card can mean different things: loading, error, missing collection or zero value, each with its own action.

StatusWhat it meansWhat to do
LoadingThe reading is arriving from the server.Wait before comparing or deciding anything.
Reading errorThe request failed for a technical reason.Use the retry command if it appears.
Missing collectionThere is no collection of answers for this scope.Follow the message shown to start one.
Zero valueThe collection exists and the brand does not appear in the answers.Open the answers and see which alternatives appear.

Section 4

Keep the four questions separate

SEO, AEO, GEO and Trust describe different aspects of the brand’s presence. Each has its own method and base, so a strong result in one lens does not offset a weakness in another.

Read each lens with its own method and base. Search, direct answers, assistant answers and reputation are not four values to add together into one result.

The overview’s four lenses, each with its own basis and method, and the rule that keeps them apart: they are not added up and do not enter the AI Trust Score. (fictional brand, demo data)
Demo data · Ilmarde Audio

Section 5

Open the evidence for a specific case

When a result provides sample details or a source link, open them before using it as evidence. The individual case shows what was actually observed.

Check the wording of the question, the observed text and the available origin. A name mention, a recommendation and a website link are different signals: check which one is actually present.

The evidence of a single case: an assistant’s answer with the brand, the competitor and the cited source marked on the sentences that name them. (fictional brand, demo data)
Demo data · Ilmarde Audio

Section 6

Interpret the interval and trend

In the score detail, read the interval, sample and period alongside the value. The interval shows how stable the measure is and helps you avoid mistaking a small fluctuation for a real change.

Assess a change together with the comparability of the observations. If the score cannot be calculated, read the stated reason: do not replace it with an estimate or interpret missing data as deterioration.

The AI Trust Score trend table by closed cycle: for each period the judgement, the score and the interval that expresses its uncertainty. (fictional brand, demo data)
Demo data · Ilmarde Audio

Section 7

Connect actions to evidence

Before starting an action, check which observed problem it addresses. An action tied to specific evidence has a clear goal and a starting point from which to assess what follows.

After an action, compare relevant new observations while retaining their context and limitations. A later change alone does not prove that the action caused it and does not guarantee a commercial outcome.

Why an open action exists: the problem observed in the answers, the cited page behind it and the dated sources that document it. (fictional brand, demo data)
Demo data · Ilmarde Audio

Related paths

Definitions, FAQs, guides and methodology: from here you reach the material you need to read the results and explain them to your team in the same words.

From question to action

Learn how to use TrustInsider: choose the brand and the questions, read the collected answers, update your content and compare the results.

Read

Find the right answer

Help with access, credits, collections, integrations and data: guides, FAQs, the site assistant and support requests from your workspace.

Read