How to build a naming system for multiple products
8 min read
The first product gets a name because somebody cared. The second gets one because a launch date was approaching. By the fourth, the pricing page is a museum of whoever happened to be in the room, and every new feature triggers the same argument from the beginning.
That is what a naming system prevents. It is not a style guide about tone of voice — it is a set of decisions about how names relate to each other, written down once so that the next launch is a lookup rather than a debate.
This guide covers the four structures worth choosing between, the tests that only appear when names sit beside their siblings, and what to actually write down.
Decide what the parent brand is for
Everything else follows from this, so make it explicit before naming anything. There are four structures, and most confusion comes from a company drifting between two of them without noticing.
| Structure | How it reads | Suits | Costs |
|---|---|---|---|
| One master brand | Acme Invoices, Acme Payroll, Acme Reports | Products sold to the same customer, bought together | No product can build its own reputation |
| Master plus sub-brands | Acme Ledger, Acme Pulse | A portfolio where products are recognisable but related | Two names to maintain per product |
| Independent brands | Ledger, Pulse, Sift — Acme barely mentioned | Different audiences, different markets, or an acquisition strategy | Every product starts its marketing from zero |
| One brand, descriptive parts | Acme, with Invoices and Reports as features | A single product with capabilities, not a portfolio | Nothing is separately memorable — usually correct |
That last row is where more companies belong than choose it. The instinct once a product grows is to brand its major features, and most of the time those features are not things a customer needs to recognise independently. A capability with a name is a thing to learn; a capability described plainly is a thing to use.
Choose a consistent level of description
Within a family, the names should sit at roughly the same distance from what they do. One literal name beside one abstract invented one reads as an accident, because it is.
So pick a register and stay in it. Either everything is plainly descriptive — Invoices, Payroll, Reports — or everything is a distinct word at a similar remove — Ledger, Pulse, Beacon. What you cannot do is mix Invoices with Zephyr and expect a customer to work out how the two relate.
Consistency does not mean the names look alike. It means somebody encountering the third one can tell it belongs to the same family as the first two.
Create a pattern loose enough to survive
Patterns are what make a portfolio legible: a shared prefix, a shared ending, a theme the names are drawn from, a consistent grammatical shape.
The trap is a pattern too tight to extend. A theme with a small vocabulary runs out — a company that named three products after birds will name the eighth one after something that is technically a bird — and every forced name afterwards advertises that the system broke. Before adopting a pattern, generate the next five names inside it and see whether they are still good. If the fifth one is a stretch, the pattern is too narrow.
A shared grammatical shape ages better than a shared vocabulary. 'Two syllables, ends in a vowel' will never run out; 'named after Greek gods' has a hard limit and a tone you may not want in four years.
Test names together, never alone
This is the test that only exists for portfolios, and it is the one that catches most problems. A name that works in isolation can be unusable beside its siblings.
Put the whole set into the places it will actually appear:
- A pricing page, in a row — where similar-length names with similar endings turn into a blur.
- A navigation menu, where they compete for a glance.
- A sales slide, read aloud in order.
- A support conversation: 'that is in Pulse, not Pace' is a sentence somebody will say a thousand times.
- A dropdown or a settings screen, alphabetised, where accidental adjacency appears.
Read the full set out loud in one breath. Two products whose names share a first syllable, or rhyme, or differ by one vowel, will be confused by customers and by your own staff — and that cost is paid on every support call rather than once.
Protect distinguishability by sound
Customers discuss several of your products in one meeting. Each one has to be identifiable by ear, which is a stricter requirement than being individually pronounceable.
Two practical rules. Vary the stressed syllable and the vowel shape across the family rather than only the consonants, because consonants are what get lost on a phone call. And avoid names that differ by a single sound in the same position — those are not two names, they are one name with a coin flip.
The individual tests still apply to each name: a stranger should be able to say it from the spelling and spell it from the sound. What makes an invented name easy to say and spell covers those, and in a portfolio the failures compound rather than adding up.
Decide the domain architecture before you need it
Products can live on paths, on subdomains, or on their own domains, and the choice should follow the brand structure rather than whatever was available on the afternoon of the second launch.
| Approach | Fits | Watch for |
|---|---|---|
| Paths on one domain | One master brand, or descriptive features | Nothing — this is the default and usually right |
| Subdomains | Products needing separate apps or teams | A subdomain per feature, which is architecture as branding |
| Separate domains | Genuinely independent brands or acquisitions | Renewals, certificates, and diluted effort across sites |
Two things worth deciding early rather than discovering. Whether a product needs its own domain is a question about audience, not about whether the domain happens to be free. And every separate domain is a permanent operating cost plus a place where your brand can drift — which is why the default should be paths, and each exception should have a reason somebody wrote down.
If a product does earn its own domain, it earns the whole naming process too, including the availability constraint. Reserve the obvious ones early, while the pattern still has room: it is much cheaper than discovering at launch four that your naming system produces names whose domains are all gone.
Do not brand every feature
Every branded term is a word customers have to learn, and the budget for that is much smaller than it feels from inside the company. A portfolio of nine proper nouns is a vocabulary test attached to a product.
The test for whether something deserves a name: does a customer need to refer to this independently — in a purchase decision, a support request, or a conversation with a colleague? If it is only ever encountered inside the product, describe it and move on.
Reversing this later is possible and unpleasant, because retiring a name you taught people is a small rebrand each time. Adding names later is easy; removing them is not.
Research conflicts at both levels
A clear parent brand does not make every sub-brand clear. Each product name that customers will encounter independently needs its own search: existing companies and products, app stores, and the trademark registers for the classes and markets it will trade in.
This is where a naming pattern can create a specific hazard. A system that reliably generates one kind of word will reliably generate collisions with everybody else using the same convention, and the fifth product in a themed family is the most likely to land on something taken. How to check whether a name is already being used covers the searches.
Write it down
A system that lives in one person's head is not a system. Two pages is enough, and it should be specific enough that somebody who was not in these conversations can name the next product without asking.
What the naming guide should contain
- Which of the four structures you have chosen, stated plainly
- The register: descriptive, distinctive, or a defined mix
- The allowed patterns, with the next five example names generated inside them
- Length and syllable limits, and any sound to avoid because a sibling already uses it
- Capitalisation and how the parent brand is written alongside a product name
- The bar a capability must clear before it gets a name at all
- The domain rule: paths by default, and what earns an exception
- Where product names live in the codebase and the marketing, so a change is findable
- Who decides, and what research is required before a name ships
The last two matter more than they look. A documented decision-maker is what stops the argument recurring, and a required research step is what stops a launch discovering a conflict in its final week.
Expect to revise it once
A naming system written before the second product is a hypothesis. Around the fourth or fifth, you will find that the pattern is tighter than it needed to be, or that two names are being confused, or that a branded feature should never have been branded.
Revise the rules then, deliberately, rather than quietly making exceptions to them. A system with three documented amendments is still a system; a system with three undocumented exceptions is back to improvising, and nobody will be able to tell which of the two you have.
In short
Pick one of the four structures explicitly, keep every name in the family at the same level of description, and choose a pattern loose enough to name five more products. Then test the set together — beside its siblings, out loud — because that is where portfolio names fail.
Frequently asked questions
Should every product have its own brand name?
- No, and this is the most common portfolio mistake. Reserve names for things a customer needs to refer to independently; describe everything else. Each branded term is a word your customers have to learn, and the budget for that is small.
Should each product get its own domain?
- Usually not. Paths on one domain is the sensible default, and separate domains suit genuinely independent brands or acquisitions. A domain per feature is architecture masquerading as branding, and every extra domain is a permanent cost.
How do I stop two product names being confused?
- Vary the stressed syllable and the vowel shape across the family rather than only the consonants, and never ship two names that differ by one sound in the same position. Test by reading the whole set aloud in one breath.
What if we outgrow the pattern?
- Amend the documented rules rather than making a quiet exception. A system with recorded amendments still works; a system with unrecorded exceptions is improvisation that nobody can distinguish from a system.
Sources
These support factual background only. The guide itself is original writing, and nothing here should be read as legal advice.
Related guides
Brand Naming
What makes an invented name easy to say and spell
Why some made-up names feel like words and others feel like passwords — vowel rhythm, consonant runs, the joins between parts, and the spelling decisions you make for the listener.
10 min read
Brand Naming
50 ways to turn an ordinary word into a brand name
Fifty edits — openings, endings, vowels, consonants, expansions, compressions, semantic and phonetic moves — and how to keep an invented name from sounding random.
9 min read
Business Naming
How to check whether a business name is already taken
Nine places to look before you build a brand around a name, why two companies can sometimes share one, and where a founder's own research stops being enough.
8 min read
Creators & Digital Products
How to name a digital product so it feels like a real thing
Courses, templates, tools and communities live or die on feeling intentional rather than uploaded. The checkout sentence, the category-word trap, and how to name a second product.
7 min read
Domain Names
How to rebrand when your domain no longer fits
Changing a live domain is a migration, not a swap. How to work out what actually needs to change, build the redirect map, keep email working, and come out the other side.
11 min read