How MindMeet Signal works

Two layers, doing two different jobs

MindMeet is built around a simple split: research is not the same job as judgment, and we don't let one quietly stand in for the other.

Research finds the evidence. Signal decides what it means.

The Score

Every Signal produces a single number, shown from 0.0 to 10.0. It is not a star rating, and it is not a simple average of Substance, Value and Sentiment. The research framework first produces a structured preliminary assessment; Category Position then determines the appropriate scoring neighborhood for the product; and the evidence determines where inside that neighborhood the product actually belongs. A human editor reviews and approves the final published Score before it goes live.

Category position determines the allowable scoring neighborhood; the research determines where inside that neighborhood the product belongs.

Category Position

The same numerical score can mean very different things if the category itself is defined poorly. MindMeet evaluates products against credible, currently available alternatives within the category that actually reflects how people would shop or compare.

Category Position is a separate, qualitative judgment alongside the Score — it can include labels such as Category-Leading, Top-Tier, Strong, Above Category Average, or Category Average. These labels aren't mechanically derived from the Score alone; they reflect where the product genuinely sits against the alternatives someone would realistically be choosing between.

Three lenses, considered together

Every Signal looks at Substance, Value and Sentiment. They feed a structured scoring framework — not a simple average, and not something invented case by case. A human editor reviews, challenges and approves the result; they don't replace the framework or the weighting behind it.

Confidence — a separate signal

Confidence describes the strength of the evidence base, not how positive or negative the verdict is. A Signal can be confidently strong, confidently mixed, or simply too early to call.

AI expands the research. A human owns the verdict.

AI and research tooling help us examine far more evidence than manual review alone would allow — more sources, more comparisons, faster. That evidence feeds a structured framework that constrains the evaluation; editors don't score products free-form. What editors do is review inconsistencies, edge cases and category context, and approve the result before it publishes. That's what "owns the verdict" means here: accountability and calibration, not discretionary scoring.

Calibration keeps the scale meaningful

Scores only become useful if a 9.0 means roughly the same level of category strength across different reviews. MindMeet maintains calibration references across categories and revisits prior Signals as more products are evaluated, so the scale doesn't quietly drift over time.

Calibration helps avoid three specific failure modes: score inflation, where scores creep upward over time without the underlying evidence improving; category drift, where the same score comes to mean different things in different categories; and one-off editorial standards, where a single reviewer's judgment isn't checked against how similar products have been evaluated elsewhere. We don't publish the internal mechanics of how calibration is maintained — that's part of how MindMeet Signal stays consistent, not something we expose product by product.

How evidence is collected

MindMeet Research draws on a range of source types depending on what's genuinely available and relevant to a given product:

• Official product and company documentation
• Labels, specifications and pricing
• Peer-reviewed research
• Clinical-trial registries
• Regulator and government sources
• Recognized third-party testing and certification
• Credible independent editorial sources
• Customer-review platforms
• Community discussions, when useful for sentiment patterns
• Competitor and product comparison research

We separate what we can verify directly, what a company reports, and what remains uncertain. That distinction carries through into the research record Signal draws on — a claim a brand makes about itself is treated differently from a claim we've independently confirmed.

Signal, depending on where you meet it

Not every context calls for the same amount of detail. Signal adapts to where it appears — a list row needs less than a full review does — without changing the underlying verdict.

Commercial independence

"Brands can fund the research. They can't fund the verdict."

MindMeet may have sponsorships, affiliate relationships or other commercial relationships with brands — those relationships are disclosed on the relevant content. What a brand cannot do is purchase, negotiate or influence the Score, the Category Position, the Confidence level, the lens conclusions, the editorial Take, or any factual conclusion.

If a brand identifies a factual error — an outdated price, a discontinued ingredient, a corrected spec — they can submit it the same way any reader could, and we verify it like any other evidence. The verdict doesn't move because a brand asked; it moves because the evidence changed.

Questions