2026-09-15
Nural Choudhury
Brand architecture is a naming decision, not a diagram to produce.
Most of what gets blamed on naming confusion is an unresolved question about who buys each product, dressed up as a taxonomy problem. Settle the buyer first, and a model falls out of that answer; settle the model first, and the buyer question resurfaces at every naming argument that follows.
the naming argument that returns every quarter because nobody settled who a sub-brand is for, and the acquisition that sits half-integrated because no one decided whether it keeps its own name.
brief a designer against a model instead of a preference, defend a naming decision against the next executive who wants their product to stand alone, and give an acquired brand a documented path instead of an open question.
the buyer written down for every brand in the portfolio, a model assigned to each relationship rather than to the whole company, and a decision record naming who approves an exception and what triggers one.
The vocabulary most brand strategists reach for, branded house, house of brands, endorsed, and sub-brand, comes from David Aaker’s Brand Portfolio Strategy (2004), which named and organised categories practitioners had already been using informally. Aaker wrote as an academic and consultant specialising in brand equity, and his framing treats the portfolio itself, not any single brand, as the asset under management. His observation that few real portfolios are a pure example of any one model; most run a hybrid, one model for the core business and another for what came in through acquisition, is also his.
For every brand or product in the portfolio, you must write down who buys it and why, in one sentence a stranger could follow without context. Two products that share a buyer and a reason to buy are candidates for a shared brand. Two that do not share either belong in a different conversation, whatever the org chart says.
I will not let a naming discussion start before this sentence exists for every brand in scope. A workshop that skips it produces an opinion about logos, not a decision about the portfolio.
You must list every brand, sub-brand, and product name currently in market, including the ones nobody remembers approving. Note which ones came from acquisition, which grew organically, and which exist because a team named something without asking. This map is the starting position, not the target state, and it must be honest about the mess before anyone proposes a fix.
A portfolio rarely needs one model applied everywhere. Choose a model for each parent-to-brand relationship, using the buyer answer from the first step to decide.
| Model | What the customer sees | Choose it when | Watch for |
|---|---|---|---|
| Branded house | One name covers everything; products are named descriptively under it | The parent brand can credibly stretch to the new offering and trust should transfer automatically | One product’s failure becomes every product’s failure |
| House of brands | Each brand stands alone; the parent is invisible to the customer | Buyers for different products hold conflicting expectations, or an acquisition needs to keep its own equity | No shared efficiency across marketing spend, and internal complexity multiplies with every brand |
| Endorsed | A sub-brand leads, with the parent visibly backing it | The sub-brand needs room to position itself but still benefits from the parent’s credibility | The endorsement hierarchy needs active management, or it drifts into looking monolithic or looking independent by accident |
| Sub-brand extension | Parent name and product name function as one unit | The product needs distinction but family membership matters more than independence | Naming constraints tighten as the parent brand’s meaning gets stretched across more products |
Most portfolios, no matter the size, run more than one row of this table at once. A company can be a branded house for its core product line and a house of brands for what it has acquired, and that is a hybrid, not an inconsistency, provided each relationship was chosen on purpose.
You must write down what is fixed and what can flex for each relationship: naming pattern, logo lockup, colour ownership, and who signs off an exception. A rule nobody can find is not a rule. Put it where the people who name things next quarter will look.
Assign a single person or role to own the architecture decision in the future, separate from whoever owns the visual identity guidelines. Naming decisions and identity guidelines fail for different reasons, and conflating the two owners means neither problem gets caught early.
I hand over the decision record and the model assigned to each relationship. I keep the exception approval, because that is the one decision point where architecture erodes.
What I check, and when: not the finished naming proposal, but the buyer sentence, when someone first asks for a new name rather than after the launch deck exists. A proposal that arrives without a buyer sentence attached goes back before I look at the name itself.
The conversation that goes wrong is always the same shape. A product lead asks for a distinct sub-brand because their team worked hard on it and feels it deserves its own identity. The question that fixes it is not about the work: I ask who buys this, and whether that buyer needs to know it is different from what they already buy.
Most of the time the honest answer is no, and the request quietly becomes a feature launched under the existing name. I know a team has got it when they bring me the buyer question before they bring me the name, without being asked. That is the point I stop reviewing every naming request personally and start reviewing only the exceptions that reach the process I wrote down.
If exceptions stop reaching me and start happening anyway, the architecture is no longer being coached; it is being ignored. That is a different problem with a different fix.
Take a professional-services firm that has grown by acquiring three regional practices, each still trading under its own name. The founder wants one firm, one name, by next year. Before the rebrand brief goes anywhere, the buyer question comes first: do the same people buy from all three practices, or do they serve different clients who chose the smaller practice because it wasn’t the big firm?
The answer splits two ways. Two of the acquired practices share a buyer with the core business, corporate clients who want scale and consistency, so they fold into a branded house with descriptive naming. The third serves a buyer that explicitly avoids large firms, so it stays endorsed, with its own name leading and the parent named as backing rather than the front door.
The founder gets one firm in the two markets where one name helps, and keeps the buyer relationship intact in the market where it does not.
It fails first when the model gets chosen before the buyer question is answered. A workshop that opens with “should we be a branded house or a house of brands” is already asking the wrong question, because the honest answer depends on who buys what, and that has not been established yet.
It fails second when the architecture serves the org chart rather than the customer. A business unit gets its own sub-brand because it has its own budget and its own leader, not because its buyer needs to see a different name. I treat “which team owns this” as irrelevant to the naming decision; it decides who executes the decision, never what the decision is.
It fails third as a favour. A sub-brand launched to recognise an internal team’s effort, or to avoid an awkward conversation about folding a pet project into the main brand, costs nothing on launch day and costs real governance attention for years afterwards. I do not grant a name as a reward, because a name is a promise to a buyer, and a team’s effort is not the buyer’s problem.
an endorsed brand leads with its own name and the parent appears as backing; a sub-brand extension puts the parent and product name together as one unit, with less room for the product to position itself apart from the parent.
Can one portfolio run more than one model at once? Yes, and most portfolios do. A company can be a branded house for its core line and a house of brands for acquisitions that need to keep their own equity, provided each relationship was chosen rather than inherited by accident.
one named person or role, separate from whoever owns the visual identity guidelines. The naming decision and the visual system fail in different ways, and one owner catching both means neither gets caught early.
at a trigger event, not on a schedule: an acquisition, a major rebrand, or a pattern of naming exceptions that suggests the current model no longer matches how the business buys and sells.
| Fact | Detail |
|---|---|
| Term | Brand architecture |
| Core models | Branded house, house of brands, endorsed, sub-brand extension, and hybrids combining them |
| Framework vocabulary attributed to | David Aaker, in Brand Portfolio Strategy (2004) |
| Commonly cited branded house example | |
| Commonly cited house of brands example | Procter & Gamble |
| Commonly cited endorsed example | Marriott’s Courtyard by Marriott |
A brand sprint is a time-boxed workshop that compresses brand strategy, verbal identity, and visual identity development into one structured engagement with a fixed team and a fixed calendar.
Read it
A brand personality vocabulary does not settle an argument about taste; it exposes whose opinion is doing the work.
Read it
Design systems architecture is the set of structural decisions that let a design system keep working as an organisation grows: how tokens layer, how components are organised and configured, and how the code that implements them is housed, built, and shipped.
Read it
A team isn’t misaligned because it lacks a vision; it is misaligned because the cadence that surfaces disagreement before it hardens doesn’t exist.
Read it