---
name: brand_architecture
description: Decide how brands, sub-brands, and product lines relate: choose between branded house, house of brands, endorsed, and hybrid models with a scored decision worksheet, build the naming system that holds the model together, and plan the migration when the model changes. Triggers when someone asks how to structure a brand portfolio, whether a new product gets its own brand or the master brand name, how to handle an acquired brand, or how to fix a confusing set of product names.
---

# Brand Architecture

## Purpose
Give a portfolio a deliberate structure instead of a pile of one-off naming decisions. The output is a model choice with the reasoning on paper, a relationship map for every offering, a naming system with rules a new hire can apply without asking, and, when the model is changing, a phased migration plan that protects existing equity.

The framework is the Brand Relationship Spectrum from Aaker and Joachimsthaler (California Management Review, 2000). It treats the models as points on one continuum, ordered by how separated the brands are in execution and in the customer mind. The position on that continuum is set by the driver role: which brand actually drives the purchase decision.

## Workflow
1. Inventory the portfolio first. List every offering, its audience, its price tier, the promise it makes, and the name customers actually use for it. Ask a few customers what brand they bought. The answer names the current driver, which is often different from what the org chart says.
2. Score the model choice. Run the decision worksheet in `scripts/architecture_score.py` (eight tests, each scored with evidence, not opinion). Read `references/models-and-spectrum.md` for what each model costs and risks. A split result means hybrid, and hybrid needs a written rule for which lines get which treatment, or it decays into no model at all.
3. Assign a role to every offering. Each one gets exactly one role: master brand, sub-brand (master as driver, or co-driver), endorsed brand, independent brand, or plain descriptor. Descriptors are functional labels, not brands, and get no logo, no campaign, and no trademark filing of their own.
4. Build the naming system. Fill in `templates/naming-system.md`. Set the pattern per role, the name type allowed at each level (coined, suggestive, descriptive), and the tests a proposed name must pass before anyone designs anything around it.
5. Draw the relationship map. One page, parent at the top, every offering placed under its role, with the driver named on each line. If a placement needs a paragraph to justify it, the placement is wrong or the model is wrong.
6. Stress test with the next three offerings. Name the next product, the next market, and a plausible acquisition using only the rules just written. If any of the three breaks the system, fix the system now, while it is still paper.
7. If the model is changing, plan the migration. Follow `references/migration-playbook.md`. Sequence by equity at risk, keep the old and new names linked during transition, and set a date when the old name stops appearing anywhere.
8. Write the governance rules. Who can create a new brand, who approves an exception, and when the portfolio gets reviewed. Without this step the architecture lasts until the next launch meeting.

## Output Contract
Deliver these five pieces, in this order:
1. Model decision: the chosen model (or hybrid split), the worksheet scores, and the two strongest arguments against the choice, answered honestly.
2. Relationship map: every current offering placed by role, driver named for each.
3. Naming system: the completed template from `templates/naming-system.md`, with at least three worked names showing the rules applied, including one name the rules reject and why.
4. Worked architecture at the right scale: use the single product pattern or the multi-line pattern from `references/worked-architectures.md`, adapted to the real portfolio, not copied from it.
5. Migration plan when the model changes: phases, order of brands, transition naming, customer communication points, and the equity check that proves the move did not lose people.

## Operating Rules
1. Default to the branded house when audiences and promises match. Building a second brand costs a second awareness budget. A new brand must earn its separation with evidence: a conflicting audience, a conflicting price tier, a conflicting promise, or risk that must be firewalled.
2. Never let two brands chase the same audience with the same promise. That is cannibalisation with extra marketing spend. Merge them, differentiate them for real, or retire one.
3. Protect equity you already paid for. A strong acquired brand stays independent or gets endorsed. Absorbing it on day one destroys part of what was bought. A weak acquired brand gets absorbed, on a schedule, not by neglect.
4. Match separation to risk on purpose. In a branded house a reputation problem in one line hits every line, because there is no firewall. If one line carries real regulatory, safety, or scandal risk, that fact alone can justify endorsement or independence for that line.
5. Keep the master brand name off products it cannot credibly promise. Every extension spends master equity. An extension that confuses people about what the master stands for costs more than the launch earns.
6. Names follow the trademark strength order. Generic names cannot be protected. Descriptive names are weak and hard to register until they build secondary meaning. Suggestive names protect better. Coined or arbitrary names protect best but need more marketing to gain meaning. Pick the level deliberately per role, and get a trademark search before design work starts.
7. One name, one job. A name either identifies the source (a brand), modifies the master (a sub-brand), or describes the function (a descriptor). Names that try to do two jobs end up doing neither and clutter every list, store page, and support script they touch.
8. Hybrid is a decision, not a drift. It is valid when different lines genuinely need different treatments. It is invalid as a label for a portfolio nobody has decided about. Every hybrid needs the split rule in writing.
9. Decide architecture before identity. Logos, colours, and campaigns come after roles and names. Redesigning a confusing portfolio without fixing its structure produces a prettier confusing portfolio.
10. Review the portfolio on a fixed cadence, yearly at minimum and after every acquisition or major launch. Retire brands that no longer carry their own audience or promise. A brand nobody would miss is a cost line, not an asset.