A good name is memorable, ownable, and doesn't fight the product.
Naming decisions are made hurriedly (usually the week before launch), by whoever's in the room (rarely with linguistic or trademark expertise), and are then extremely expensive to reverse (rebranding a shipped product typically costs 12-18 months of marketing momentum plus real legal and infrastructure work). Treating naming as a small task produces a compounding tax: every future integration, every SEO effort, every trademark filing, every localization is more expensive than it needed to be. A modest amount of process — 2-4 weeks, a structured brief, real legal screening — dramatically improves the outcome.
(1) Descriptive — literally describes what it does (Bank of America, PayPal, Salesforce). Easy to understand, hard to trademark strongly, becomes limiting as the product expands. (2) Suggestive — evokes the benefit without describing it (Uber, Slack, Notion). Sweet spot for most startups. (3) Abstract/coined — invented word (Kodak, Xerox, Zynga, Häagen-Dazs). Strong trademark, hard to remember, needs marketing spend to become meaningful. (4) Founder or place — Adobe (creek), Amazon (river), Ford. Feels timeless, works only if the reference has resonance. (5) Compound or portmanteau — Facebook, Instagram, Pinterest. Middle ground, often the pragmatic winner for products.
Write a one-page naming brief before generating names. It covers: (a) what the product does in one sentence, (b) who it's for, (c) 3-5 adjectives it should feel like (and 3-5 it should not), (d) categorical constraints — must it work in French? Must it be pronounceable by a support agent in Manila? (e) practical constraints — do we need a .com? Must it fit in a 30-char app store name? (f) competitive landscape — what names are your closest competitors using, and how do you want to differentiate? The brief separates 'names we like' from 'names that fit the brief,' which is the single biggest quality-lifting move in naming.
Generate 100-300 candidates in structured batches: 30 descriptive, 30 suggestive, 30 abstract, 30 compound, 30 metaphors from a specific domain (nature, science, mythology). Do not filter during generation. Then filter in stages: (1) obvious rejections (unpronounceable, unfortunate meanings in target languages, near-conflicts with major brands). (2) name-brief fit — does it match the adjectives? (3) domain and social handle availability — .com is preferred but a good .ai/.co/.io is often acceptable now. (4) linguistic check across your top 5 target languages. Shortlist 5-10, live with them for a week, cross-reference before final selection.
Before falling in love with a name, run a preliminary trademark screen: USPTO TESS search (US), EUIPO (Europe), your specific classes (usually 9 for software, 42 for SaaS services). This is a first-pass screen — a clean search doesn't mean you're safe, but a hit means you're probably not. If the shortlist survives, hire a trademark attorney for a proper clearance search ($1,500-4,000) before committing. Common mistake: choosing a name based on domain availability alone. Cheap domain + unregisterable trademark is a much worse outcome than harder-to-find domain + defensible trademark.
Sub-naming decisions compound. Do you name features (Slack Huddles, Notion AI) or number them (v3, 2.0)? Named features feel premium and are marketable independently; numbered features feel like technical iteration. Rule of thumb: name features that customers will talk about by name; leave iterations unnamed. Avoid parallel naming systems (some features named, some coded internally with never-shipped codenames leaking into UI). Establish a small naming committee (product marketing, design, PM lead) that reviews new feature names before launch — a 20-minute review prevents most naming embarrassments.
Investor directory · Fundraising library · Articles A–Z · Company funding database