Brand Mentions Before the Decision

Recommendation is only one stage of the journey

A recommendation is easy to recognise because it usually sounds decisive.

"Choose this brand."

"I recommend them."

"I used this company and it worked well."

Those statements matter.

But they describe a relatively late stage of the conversation.

Earlier stages can look much less dramatic.

"Has anyone used this company?"

"Is this brand worth considering?"

"Would this product work for my situation?"

"Which of these two options is better?"

A brand that appears in those questions is already inside the decision process.

It has entered the user's field of consideration.

That is a different kind of visibility from recommendation, but it is still visibility.

The 248 question / consideration signals

The largest calibrated intent category in the Gold Dataset is question / consideration, with 248 signals.

This does not mean that 248 people were ready to buy.

It does not mean that every question represents strong purchase intent.

It means that the brand was present while the user was still evaluating, checking or trying to understand an option.

That is useful because it captures a stage that recommendation-only monitoring can miss.

If a brand repeatedly appears in questions, it may already be part of the market's mental shortlist before direct endorsement occurs.

The current research does not claim that this automatically leads to conversion.

It only shows that the brand has entered the conversation earlier.

The 179 recommendation signals

The calibration also identified 179 recommendation signals.

Recommendation is a different role from consideration.

A recommendation does not merely ask whether the brand is worth considering.

It suggests that the brand should be chosen, tried or trusted.

That makes Recommendation Strength an important later-stage dimension.

But even within recommendations, the evidence can differ.

Some are brief suggestions.

Others include a claimed first-hand experience.

Some describe a result.

The research therefore avoids treating every positive mention as the same kind of recommendation.

Warning signals show that visibility is not always positive

The Gold Dataset calibration contains 19 negative / warning signals.

That number is much smaller than the question and recommendation categories, but it is still important conceptually.

A brand can become visible because somebody warns another user about it.

That is still a Brand Mention.

It still contributes to the public context around the company.

But the conversational role is very different from recommendation.

This is another reason why total Brand Mention volume cannot be treated as a quality score.

Visibility and endorsement are not the same thing.

Strong first-hand experience is rarer than simple recommendation

The calibration identified 8 strong first-hand experience signals.

These are cases where the text contains a stronger structure of claimed experience rather than a simple suggestion.

That distinction matters because:

"I recommend this brand"

and

"I used this product in this situation and achieved this result"

are not the same kind of evidence.

The second contains more contextual information about the claimed experience.

The current research does not verify whether the claim is true.

It measures how much experience evidence is present in the text.

Comparison is a separate conversational role

The calibration includes 4 comparison signals.

This is a small count and should not be overinterpreted.

But comparison deserves to remain separate because it places the brand inside a choice between alternatives.

A comparison can reveal which other brands or products appear in the same decision set.

That becomes useful later for semantic analysis and co-mention networks.

The current calibration count is too small to make broad claims about comparison prevalence.

Its value is methodological:

comparison exists as a distinct role and should not be silently merged into recommendation.

A simple decision path

The findings suggest a useful way to think about Brand Mentions:

Problem → Question → Consideration → Comparison → Recommendation → Experience → Result

Not every discussion follows that sequence.

Some begin with a recommendation.

Some stop at a question.

Some contain a warning.

Some never reach an experience or result.

The path is not a funnel model being imposed on every thread.

It is a way to show that Brand Mentions can appear at different stages of the same decision process.

That is why the research keeps intent categories separate.

Why early-stage Brand Mentions matter

A brand does not need to be recommended in order to become part of a user's decision environment.

If people repeatedly ask about the company, the brand is already being evaluated.

That can matter for brand strategy because it changes the question from:

"How many recommendations did we receive?"

to:

"At what stage do people begin talking about us?"

Those are different measurements.

The second one can reveal visibility before final endorsement.

Why this matters for AI Search research

AI systems often answer questions that resemble consideration-stage conversations.

Users ask:

"Which company should I consider?"

"Is this brand trustworthy?"

"What are the alternatives?"

"Which product is best for this problem?"

The current study does not claim that the question / consideration signals in the human discussion corpus directly determine AI answers.

It does show that brands accumulate public context before direct recommendation.

That makes early-stage conversational context a legitimate object for future AI Search research.

Before testing whether AI systems use that context, the context itself has to be measured.

What this study does not claim

The study does not claim that a question equals purchase intent.

It does not claim that a recommendation equals a completed purchase.

It does not claim that a warning means the brand is objectively bad.

It does not claim that a strong first-hand experience statement has been independently verified.

It does not claim that the calibration percentages describe all 1,779,805 discussion URLs.

The finding is narrower:

Brand Mentions appear at multiple stages of a decision process, and recommendation is only one of those stages.

The practical takeaway

If a company measures only recommendations, it may be measuring the conversation too late.

A fuller Brand Mention analysis should ask:

When does the brand first enter the discussion?

Is it being questioned?

Is it being considered?

Is it being compared?

Is it recommended?

Is an experience described?

Is a result claimed?

Those roles create a more useful picture of brand visibility than a single positive-mention count.

Related research

Not All Brand Mentions Are Equal
Why conversational role is one of several dimensions hidden inside raw mention volume.

The Repetition Problem
Why intent should be interpreted separately from repeated wording.

The Anatomy of a High-Context Brand Mention
How products, problems, features and results add semantic depth.

Gold Dataset v1
The deep-analysis dataset behind the intent calibration.

Research Limitations
The boundaries that apply to calibration findings.