The Review Signal Framework
AI doesn't just read what you say. It checks whether your guests say the same thing.
The Review Alignment Score measures the semantic overlap between your stated identity and the aggregate vocabulary of your reviews. When those two signals agree, AI recommends you with confidence. When they disagree, it quietly removes you from the queries that matter most.
The Corroboration Imperative
AI recommendation systems do not simply index operator claims. They look for independent verification. If a restaurant's website claims a "quiet, intimate date-night atmosphere," but reviews consistently describe it as "loud, fun, and great for big groups," the AI detects a signal conflict.
When signal conflict occurs, recommendation confidence drops — and the entity is excluded from specific occasion-based queries. The goal of this framework is Signal Corroboration: aligning the vocabulary the operator uses with the vocabulary the guest uses.
Measurement Criteria
Three dimensions of alignment
Occasion Match
Do reviews mention the same use cases the operator positions for — dates, business meals, family, celebrations?
Atmosphere Match
Do the descriptive adjectives guests use — quiet, energetic, cozy, loud — align with the intended vibe?
Product Specificity
Do reviewers name the specific signature dishes or techniques the operator highlights, or do they default to generic phrases like 'good food'?
Scoring Tiers
How the machine reads the gap
High Alignment
Corroborated Signal
High recommendation confidence. AI safely matches user intent to the entity.
Mixed Alignment
Fragmented Signal
Moderate confidence. AI may recommend the entity for generic queries but excludes it from highly specific, occasion-based searches.
Low Alignment
Conflicted Signal
Low confidence. AI penalizes the entity due to contradictory data, resulting in exclusion.
Guest-Prompt Strategies
You can't write reviews for guests. You can shape the vocabulary they use.
Signal correction embeds the desired vocabulary into the physical and digital touchpoints of the guest experience. When a server describes a dish, they use the specific, machine-readable terms you want indexed. Menu copy features the target vocabulary prominently. Follow-up emails prompt the guest to mention occasion or atmosphere rather than asking for a generic rating.
Example · Strategic Prompting
Generic
"Leave us a review on Google!"
Occasion-Anchored
"Did you celebrate a special occasion with us tonight? We'd love to hear about your experience on Google."
The second prompt feeds the AI the corroborating signal it needs to confidently recommend you for anniversary and date-night queries.
Corroboration vs. Conflict
The same restaurant. Two different recommendations.
Intended Signal
Intimate date-night spot
Corroborates
"Perfect place for an anniversary dinner. Very quiet and romantic."
Conflicts
"Great spot to grab drinks with a group of friends. Very lively."
Intended Signal
Chef-driven tasting menu
Corroborates
"The 7-course tasting menu was incredible, especially the dry-aged duck."
Conflicts
"The food was pretty good but portions were small for the price."
Intended Signal
Fast-casual lunch
Corroborates
"Quick service, perfect for grabbing a bite during my lunch break."
Conflicts
"Nice place, but we waited a long time for our food."
Next Step
See whether your reviews corroborate or conflict with your identity.
Run a Guest Read and we'll surface the vocabulary gap between what you claim and what your guests actually say.