Not every recommendation has the same intensity.
Some Brand Mentions simply mention an option.
Others clearly suggest that another person should consider, choose or prefer a specific brand.
Recommendation Strength is the framework used to describe that difference.
The public framework spans:
RS0 → RS5
The scale is not a measure of purchase probability.
It is not a measure of truth.
It is not a measure of commercial value.
Its purpose is narrower:
to describe how strongly the wording functions as a recommendation.
Recommendation Strength
The complete methodology continues below in its original wording.
Why Recommendation Strength exists
Open section →02Methodology sectionThe Gold Dataset calibration
Open section →03Methodology sectionRecommendation is a conversational role
Open section →04Methodology sectionWhat RS0–RS5 represents
Open section →05Methodology sectionRecommendation Strength is not purchase intent
Open section →06Methodology sectionRecommendation Strength is not purchase
Open section →Why Recommendation Strength exists
A Brand Mention can play many roles.
It can be:
a question,
a neutral reference,
a comparison,
a warning,
or a recommendation.
Even inside the recommendation category, the language can vary in intensity.
A weak suggestion and a direct endorsement should not automatically be treated as identical.
Recommendation Strength gives the research a consistent way to describe that difference.
The Gold Dataset calibration
Within the explicit Brand Mention layer, the current intent calibration includes:
248 question / consideration signals
179 recommendation signals
19 negative / warning signals
8 strong first-hand experience signals
4 comparison signals
Recommendation Strength applies to recommendation-oriented language.
It does not turn the entire Brand Mention layer into a recommendation dataset.
The majority of calibrated roles should remain separate.
Recommendation is a conversational role
Consider:
"Volvo could be worth looking at."
and:
"I strongly recommend Volvo."
Both can function as recommendation-oriented statements.
But they do not carry the same degree of endorsement.
Recommendation Strength exists to capture that difference.
The framework focuses on what the wording does in the conversation.
It does not infer what happened afterwards.
What RS0–RS5 represents
The RS framework provides an ordered way to describe increasing recommendation intensity.
At the lower end, the wording contains little or no recommendation force.
At the higher end, the wording contains a stronger and more direct endorsement.
The current public documentation does not publish operational annotation thresholds for each level.
The framework is presented as a calibrated research model.
It is not presented as a universal industry standard.
Recommendation Strength is not purchase intent
A user can strongly recommend a brand without the reader intending to buy it.
Likewise, someone can consider buying a brand without anyone explicitly recommending it.
That means:
recommendation strength ≠ purchase intent
The framework measures endorsement language.
It does not measure the recipient's later decision.
Recommendation Strength is not purchase
A recommendation can be present even when no transaction occurs.
The research does not have to observe a purchase for recommendation language to exist.
Therefore:
recommendation ≠ purchase
A strong endorsement is still a conversational signal.
It is not a conversion record.
Recommendation Strength is not experience evidence
A person can recommend a brand without describing first-hand use.
For example:
"I've heard good things about Volvo — you should consider it."
This can contain recommendation language.
But it may contain weak or no first-hand evidence.
Conversely:
"I've used Volvo for three years."
can contain stronger experience evidence without necessarily recommending the brand.
Recommendation Strength and Experience Evidence are separate dimensions.
Recommendation Strength is not Brand Context Depth
A short sentence can contain a strong recommendation.
For example:
"Choose Volvo."
That may be recommendation-heavy but context-light.
A longer sentence may contain:
a product,
a feature,
a use case,
and a problem,
while still remaining a question rather than an endorsement.
Recommendation Strength and Brand Context Depth therefore measure different properties.
Stronger recommendation does not mean stronger evidence
A direct endorsement can be confident without being well-supported.
For example:
"Definitely choose Brand X."
is stronger recommendation language than:
"I used Brand X for three years and it worked well for me."
But the second statement may contain more experience evidence.
This is why the methodology keeps recommendation intensity and evidence quality separate.
Recommendation Strength does not determine sentiment
Recommendation-oriented wording is usually favorable in direction.
But the framework itself is not a full sentiment system.
A discussion can contain:
recommendation of one brand,
warning against another,
or mixed evaluation inside the same post.
The RS framework measures recommendation force.
It does not replace broader conversational interpretation.
Why a separate recommendation framework matters
Without a separate framework, all recommendation-oriented statements can collapse into one category.
That removes useful distinctions.
A mild suggestion,
a preference,
a direct endorsement,
and an emphatic recommendation
may all belong to recommendation-oriented language while differing substantially in force.
RS0–RS5 provides a way to preserve that variation.
Why the framework avoids one universal score
Recommendation Strength should not be combined automatically with:
Brand Context Depth,
Experience Evidence,
Brand Mention Complexity,
link structure,
or semantic relationship count
to produce a universal brand score.
A strong recommendation can be shallow.
A detailed mention can be neutral.
A first-hand experience can be negative.
These dimensions are analytically useful because they remain separate.
Why this matters for AI Search research
Users often ask AI systems for recommendations.
That makes human recommendation language a relevant variable for future comparison.
The current framework allows the research to ask:
does the strength of recommendation language in external discussion correspond with how confidently AI systems later recommend or describe a brand?
The current study does not establish that relationship.
It creates a measurable variable for future experiments.
It does not assume that a higher RS level causes stronger AI recommendations.
What Recommendation Strength does not claim
Recommendation Strength does not claim that RS5 is commercially "better" than RS2.
It does not claim that stronger recommendation means a purchase occurred.
It does not claim that stronger recommendation means stronger first-hand evidence.
It does not claim that stronger recommendation means greater contextual depth.
It does not claim that stronger wording is more truthful.
It does not claim that higher RS automatically improves AI visibility.
And it does not claim that RS0–RS5 is a universal industry standard.
Its purpose is narrower:
to describe the intensity of recommendation-oriented language around a Brand Mention.
The practical takeaway
Brand Mention answers:
Was the brand named?
Intent asks:
Was the mention functioning as a recommendation?
Recommendation Strength adds:
How strongly did the wording recommend the brand?
The framework used for that distinction is:
RS0 → RS5