Weighted average of the three sub-scores. No confidence multiplier. No hidden penalty. The score popover shows the exact math.

When one sub-score is missing, the missing input weight redistributes proportionally across the remaining two. Two of three inputs means medium confidence; all three means high confidence. With fewer than two scored inputs, no Mindber Score is shown and the page renders Score pending instead.

mindberScore = round(innovationIndex x 0.40 + functionalityScore x 0.50 + activityScore x 0.10, 1)

Fallback weights when one input is missing

• No Activity: Innovation 44% · Functionality 56%
• No Innovation: Functionality 83% · Activity 17%
• No Functionality: Innovation 80% · Activity 20%

Functionality is weighted highest because buyers first need to know whether a product appears capable of solving the job it claims. Novelty matters, but a distinctive concept without enough usable surface area should not outrank a more complete product that covers the core workflow.

Activity receives a smaller share because public momentum is a freshness signal, not a quality verdict. The composite rewards products that combine clear differentiation, practical capability, and recent evidence without letting any single signal family dominate.

• High: all three sub-scores present. Default 40/50/10 weights apply.
• Medium: two of three sub-scores present. Fallback weights apply.
• Insufficient: fewer than two sub-scores present. No Mindber Score is shown.

Bands are absolute; they do not shift with the category. The category percentile label separately shows how the product ranks against peers.

• Not investment advice
• Not a guarantee of business performance
• Not a measure of company financial health
• Not a recommendation to buy or avoid
• Not a substitute for due diligence

• All score inputs are derived from public sources.
• Disputes start from the correction widget on the product page.
• Score changes within 7 days surface in the product page audit log.
• Click any score chip to see per-input contributions.

Average of three sub-scores: Novelty, Differentiation, and Concept clarity. Each uses a 0-10 scale.

innovationIndex = round((novelty + differentiation + conceptClarity) / 3, 1)

• Novelty: how distinct the core concept is from the category baseline.
• Differentiation: features present in this product but missing from at least 50% of category peers.
• Concept clarity: how clearly the tagline and description communicate the value proposition.

Confidence-weighted average of four sub-scores. Feature breadth and performance reliability are deterministic; feature depth and integration surface are inferred from public evidence.

functionalityScore = sum(score x baseWeight x confidence) / sum(baseWeight x confidence)

• Feature breadth: count of features from collected source records.
• Feature depth: quality of the top features versus the category leader.
• Integration surface: public API, webhook, and native integration evidence.
• Performance reliability: website liveness score plus homepage response time.

A score in isolation is not enough. Mindber computes per-category distributions (median, p25, p50, p75, p90) and buckets each final score into a peer label. A baseline is only published once a category has at least 10 scored products; below that minimum, percentile is shown as unranked.

• p90 or higher: top 10%
• p75 or higher: top 25%
• p50 or higher: above median
• p25 or higher: below median
• below p25: bottom 25%

• Full baselines (median, p25, p75, p90) refresh weekly. A drift alert fires when a category median moves by more than 1.0 point.
• Percentile labels refresh daily. Raw product scores stay fixed until the product is regenerated; only the bucket label updates.
• Product scores regenerate whenever source inputs change or the scoring methodology is upgraded.

• Every LLM-written sentence must cite a specific feature, use case, or baseline data point. Marketing adjectives are banned.
• A banned-word safety filter runs on every rationale; offending text is blanked out rather than retried.
• Any sub-score below 4/10, composite confidence below 0.6, or a safety-filter replacement flags the record for admin review before publication.
• Scores compare within category, never absolute. Below median is factual; bad is not a word we use.

Mindber assigns a liveness status to each AI product based on automated public-signal checks. This explains which signals we use, how they are weighted, and what each status label means.

• DNS (weight 10): whether the domain resolves and has valid A, MX, and NS records.
• Homepage (weight 15): HTTP status code and response time of the product homepage.
• Pricing page (weight 10): whether a pricing page exists and contains price signals.
• Social activity (weight 20): days since last post on X/Twitter.
• Hiring (weight 15): whether an ATS is detected and open roles are listed.
• Traffic (weight 30): monthly visitor estimates from third-party data providers and 30-day change percentage.

• Alive: composite score at least 70/100 with multiple active signals.
• Slowing: composite score 40-69 with reduced but detectable activity.
• Dying: composite score 20-39 with few active signals and elevated inactivity indicators.
• Dead: composite score below 20 across 3 consecutive checks.
• Unknown: insufficient signal data to determine status.

Status verdicts are based entirely on observable public signals. A Dead label does not mean Mindber has verified that the company has shut down, filed for insolvency, or ceased operations. It reflects publicly observable signal absence, not ground truth.

Signals are checked on a rolling schedule. Most products are rechecked every 48-72 hours. High-traffic products may be rechecked more frequently. Data shown on any product page may be up to 7 days old.

If you believe a product has been incorrectly labeled, use the Report correction widget on the product page and include links to recent public evidence. We review correction requests within 48 hours.