Methodology

The evaluator never invents prices or guesses sales numbers. It estimates structural and linguistic qualities that historically correlate with brandability, investor liquidity, and startup fit. Every score is explainable: the detail drawer shows raw metric values, weights, and contributions for every domain.

Confidence is a separate axis from quality. A label with an unusual shape or unknown TLD will be marked low confidence even if its raw score is high — that's the engine telling you to apply your own judgement.

• Pronounceability. Penalises long consonant runs and vowel-less labels. Rewards strong vowel-consonant alternation.
• Spellability (radio test). Can the listener spell it after one hearing?
• Memorability index. Length-aware, with bonuses for premium consonants and balanced syllables.
• Sound symbolism. Premium consonants and crisp endings read as modern startup.
• Audio signature. Resonant consonants (m, n, l, r) plus clean vowels.
• Voice-assistant friendliness. Penalises digits, hyphens, and ambiguous short labels.
• Linguistic smoothness. Penalises consonant clusters of 3+ regardless of language.

• Pattern classification. Detects palindromes, ABAB, ABBA, CVC, VCVC, CVCV and more. Investor-grade patterns get a meaningful bonus.
• Visual symmetry. Mirrored letter positions and palindromes.
• Repetition. Rewards balanced doubles, penalises mindless triples.
• Rhythm. Pure consonant-vowel alternation reads musical.
• Compactness. Length-based monotonic decreasing score.
• Vowel/consonant balance. Closer to 50/50 is more pronounceable.
• Logo potential. Short symmetric labels with iconic letters.
• Cognitive load (inverted). Fewer unique letters → easier mental parse.

• Letter rarity. Premium-letter density (k, q, v, x, z).
• Positional premium. Strong first/last letters.
• AI/Tech fit. Heuristic match to modern AI-startup naming patterns.
• Acronym potential. 2-4 letter labels with consonant-heavy reading.
• IATA match. 3-letter labels matching curated major airport codes.
• Premium prefix/suffix. Detects iconic startup prefixes and endings (get-, hyper-, -ly, -labs, -io).
• Social-handle compatibility. Short, alpha-only labels score higher.
• Trend alignment. Match against AI / SaaS / biotech / Web3 / dev-tool dictionaries.

• Brand score. Weighted average of all label-only metrics. Does NOT include the TLD.
• Extension score. TLD-only score from a curated table. Unknown TLDs fall back to a default.
• Composite. 70% brand × 30% extension by default. Configurable.
• Investor liquidity. Length curve + phonetics + structure + extension + cleanliness + investor-pattern bonus.
• Startup appeal. Modernity + pronounceability + AI fit + trend + scalability + distinctiveness + extension.
• Structural quality. Pattern, symmetry, repetition, rhythm, compactness.
• Human taste. Premiumness + futurism + timelessness + emotional resonance + Silicon-Valley feel.
• Confidence. Separate axis. High confidence requires clean structure, known TLD, sweet length, and clear trend signal.

These cutoffs reflect the current configuration. The admin can override them; what you see below is what the engine is using right now.

Currently active: default. The curve maps label length to a 0..100 score that feeds the liquidity heuristic. Shorter labels score higher; the curve is flat in the 6-10 range and tails off after that.

A domain receives the GEM flag when ANY of these triggers fires (live cutoffs):

• elite phonetic quality (≥ 90)
• elite acronym potential (≥ 88)
• elite investor liquidity (≥ 92)
• elite typing kinematics (≥ 90)
• elite visual symmetry (≥ 92)
• matches a major IATA airport code

• Detect trademark conflicts. That requires USPTO/EUIPO search and human review.
• Predict aftermarket sale prices. The positioning labels (ultra-premium, investor-liquid, etc.) are heuristic categories, not prices.
• Evaluate cultural slurs across languages.
• Replace your domain investor judgement.