Startup Naming Scorer

Startup Naming Scorer

Research-backed naming evaluation skill. Built from analysis of 10,029 AI company names (DomainBFF Feb 2026), 300+ YC companies (W24-W26), and portfolios from a16z, Sequoia, First Round, Founders Fund, Techstars, and 500 Global.

When to Use This Skill

• Evaluating candidate names for a new company or product
• Generating name ideas that avoid common traps
• Scoring existing names against proven patterns
• Checking if a name falls into "AI slop" territory
• Naming a B2B SaaS, infrastructure, or revenue-focused company

The Scoring Rubric

Score every candidate name against these 6 criteria:

| Criterion | Weight | Slop (0-1) | Good (2-3) | Winner (4-5) | |-----------|--------|-----------|-----------|--------------| | No AI/tech slop words | 30% | Contains AI/ML/Deep/Neural/Smart/Data/Labs | Minor tech-adjacent word | Zero technology words | | Syllable count | 20% | 4+ syllables | 3 syllables | 1-2 syllables | | Radio test | 15% | Needs spelling out, ambiguous | Minor ambiguity | Heard once, found instantly | | No category words | 15% | Describes the product category | Hints at category | Zero category words in name | | Domain availability | 10% | Only obscure TLDs | .io/.so/.co available | .com available or strong branded alt | | Metaphorical resonance | 10% | Describes the technology | Neutral/abstract | Evokes the feeling/outcome |

Score ranges: 0-1.5 = Slop, 1.5-3.0 = Weak, 3.0-4.0 = Competitive, 4.0-5.0 = Winner

The AI Slop Index

Tier 1: Nuclear Slop (instant disqualify)

| Pattern | Occurrences (of 10K) | Rule | |---------|---------------------|------| | "Labs" suffix | 477 (1 in 20) | Never use unless you're literally a research lab | | "AI" in name or TLD | 30% of recent YC batches | When everyone is "AI," nobody is | | "Intelligence" suffix | 105 | Tells people what you are instead of what you do | | "Deep" prefix | 75 | Called "now passe" by multiple agencies | | "Systems" suffix | 93 | Government contractor energy from 1987 |

Tier 2: Heavy Slop (strong negative signal)

• "Open" prefix: OpenAI won. You can't.
• Neural / Neuro: Peaked 2018-2022. Dated.
• Cognitive / Cogni-: IBM Watson cosplay.
• "Spark" prefix: 95.5% prefix rate. Overcrowded.
• Nova / Nexus / Apex / Pulse: Sounds like Transformers.
• Smart- prefix: Implies competitors are dumb.
• Auto- prefix: 2014 thinking.
• Quantum: Used by companies with zero quantum involvement.
• Cyber-: 1995 sci-fi movie.

Tier 3: Suffix Graveyard

| Suffix | Status | Note | |--------|--------|------| | -ly | Dead | Peaked ~2015. Absent from YC 2024-2026. | | -ify | Dead | Shopify won. Everything else is derivative. | | -io (TLD) | Declining | Acceptable for devtools only. | | .ai (TLD) | Oversaturated | 28% of YC+Techstars already. | | -Hub | Dead | GitHub owns this forever. | | -Stack | Niche | Sounds infra even when you're not. | | -X | Risky | Elon-adjacent connotation. | | -ware | Legacy | 2005 energy. |

Tier 4: Compound Slop

• Data[Thing]: Saturated.
• Cloud[Thing]: Signals 2014.
• Smart[Thing]: See prefix slop.
• [Thing]Mind: DeepMind took it.
• [Thing]Sense: 6sense took it.

The 4 Naming Tests

Run every candidate through all 4:

1. Radio Test

Someone hears the name on a podcast. Can they spell it and Google it?

• Pass: Stripe, Gong, Zoom, Slack
• Fail: Xyleme, Qwilr, unusual spellings

2. Crowded Bar Test

Mentioned once in a loud room. Remembered the next morning?

• Pass: Short real words dominate
• Fail: Multi-syllable invented words

3. Building Test

On the side of a building, does it look like a real company?

• Pass: Anthropic, Stripe, Linear
• Fail: Names that are too cute or too clever for enterprise

4. Tell-a-Friend Test

Can a customer recommend you without spelling it out?

• Pass: "Check out Gong." That is the whole recommendation.
• Fail: "Check out Proximitty, that is P-R-O-X..."

What Winners Actually Do

The Data

• 82% of unicorn SaaS companies use single-word names
• 47% action/verb-based (Stripe, Slack, Zoom)
• 31% abstract/invented (Figma, Notion, Vercel)
• 55-60% of all YC names are real English words or compounds
• ~90% of top YC performers by valuation use real words
• 2 syllables is the statistical sweet spot

The 7 Naming Patterns (ranked by success rate)

1. Recontextualized Real Words: Single dictionary word, new meaning. THE winner pattern. (Stripe, Ramp, Notion, Slack, Zoom, Maven, Clay, Cursor)
2. Transparent Compounds: Two real words combined. (Dropbox, Coinbase, DoorDash, Instacart)
3. Invented CVCV Words: Made-up but phonetically natural. (Twilio, Figma, Vercel)
4. Metaphorical / Evocative: Suggests attributes through metaphor. (Gong, Clari, 6sense, Chorus)
5. Human First Names: Rising trend. (Harvey, Jasper, Oscar, Ada)
6. Literary/Mythological (Palantir, Anduril, Asimov)
7. Name + Descriptor: Declining. (Scale AI, Norm AI)

Revenue Intelligence Winners Specifically

• Gong: sound of a closed deal
• Clari: clarity into your pipeline
• 6sense: sixth sense for buyer intent
• Chorus: sales team in harmony

Pattern: Every winner uses a metaphor or sensory word, not a description.

Category-Specific Guidance

B2B SaaS / Revenue / Infrastructure

• Short, sharp, competent: Ramp, Vanta, Clay, Stedi
• Avoid whimsy: credibility and recall efficiency matter
• Real words that sound sturdy and institutional

B2C / Consumer

• Whimsical, emotional, memorable: Spotify, TikTok, Robinhood
• Human first names gaining ground
• Two-word story names: Warby Parker, Blue Apron

Developer Tools

• Descriptive compounds: Databricks, LangChain, Retool
• .io acceptable here
• Action verbs: Factory, Crunched

How to Use This Skill

Evaluate a Single Name

Score the name "Ballast" for a B2B revenue intelligence platform

Generate Candidates

Generate 20 two-syllable real-word candidates for a 
compliance-focused revenue engine

Compare Finalists

Compare and score: Timber vs Mandate vs Vertic 
for a governed revenue intelligence platform

Workflow

1. Gather context: What does the company do? Who buys it? What vertical?
2. Generate candidates: Use Pattern 1 (real words) and Pattern 4 (metaphorical) primarily
3. Score each name: Apply the 6-criterion rubric
4. Run the 4 tests: Radio, Crowded Bar, Building, Tell-a-Friend
5. Check the Slop Index: Flag any Tier 1-4 violations
6. Domain/trademark screen: Check .com, .so, .io availability + USPTO classes 35, 36, 38, 42
7. Present top 3-5: With scores, test results, and domain status

Constraints to Respect

When the user has established naming constraints, honor them. Common constraints:

• Syllable count limits (e.g., 2-3 max)
• No "AI" in the name
• No specific letters (e.g., no X)
• No specific suffixes (-ly, -ify, -io, -ai, -hub)
• No compound words
• No fake Latin/Greek assembly
• Must pass coffee shop test
• Must look good "on the side of a building"
• Trademark-clean in relevant classes

Sources

• DomainBFF: 10,000 AI Company Names Analysis (Feb 2026)
• NameStormers: "Name Slop" analysis
• River + Wolf: AI naming trends
• Crunchbase: Multi-year startup naming trends
• SmartBranding: YC batch domain analyses (W24, F25, W26)
• namemy.app: SaaS Naming Guide (unicorn analysis)
• YC company directory and unicorn list
• a16z, Sequoia, First Round, Founders Fund portfolio analysis

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: NulightJens
• Source: NulightJens/jensai-skills
• License: MIT

Install and usage instructions live in the source repository linked above.