Skip to content

Brand, Product, Account Structure

How the pipeline organizes information about who you're making content for. Three layers — brand, product, account — each in its own file in the reference library.

The three layers

flowchart TB
    B[Brand<br/>e.g. XYZ Wellness] --> P1[Product 1<br/>magnesium glycinate]
    B --> P2[Product 2<br/>rhodiola]
    B --> P3[Product 3<br/>ashwagandha]
    P1 --> A1[Account A]
    P1 --> A2[Account B]
    P1 --> A3[Account C]
    P2 --> A4[Account D]
    P3 --> A5[Account E]

A brand has voice and positioning. A product has ingredients and claims. An account has an avatar and a persona. A workflow ties them together: this product + this account + this script.

Brand structure

A brand is the identity layer. It defines:

  • Voice and tone — formal vs. casual, clinical vs. conversational, urgent vs. warm
  • Approved angle categories — emotional / story angles the brand can use
  • Banned words and angle categories — what's off-limits (compliance + brand-safety)
  • CTA style — how the brand asks viewers to act
  • Platform strategy — TikTok / Reels / Facebook positioning

A brand file lives at reference/brands/{brand-slug}.md. One file per brand.

What's NOT in a brand file

Things that belong elsewhere:

  • Product specifics → product file
  • Account personas → account file
  • Workflow-specific tactics → the workflow's storyboard / brief

Brand files describe how we talk about the brand. Product files describe what the product is. Account files describe who's delivering the message.

Product structure

A product is the specifics layer. It defines:

  • Ingredients and dosing
  • Mechanism — how the product works (the science)
  • Specific claims — what we can say about results
  • Competitor comparisons — how this product differs from alternatives
  • Production notes — manufacturing / origin facts that matter

A product file lives at reference/products/{product-slug}.md. One file per product.

Why claims matter

The product file is where compliance lives. The Script Writer reads this file when writing a script and only uses claims that are explicitly approved here. Banned claims (from the brand file) AND unapproved claims (anything not in the product file) get filtered out.

If a script suggests a claim you can't substantiate, that's a compliance violation. The product file is what prevents this.

Account structure

An account is the persona layer. It defines:

  • Avatar reference sheet (R2 URL to the 3-panel face-lock photo)
  • Avatar archetype (see Avatar Archetypes)
  • Default wardrobe / styling
  • Default setting (kitchen / bathroom / bedroom / car / etc.)
  • Voice characteristics (accent, pitch, energy)
  • Platform (which platforms this account posts on)
  • Sales channel (TikTok Shop / Amazon / Meta Shop)
  • Persona references — specific phrasing this avatar uses ("my mom" vs. "my patient")

An account file lives at reference/accounts/{account-code}.md. One file per account.

How they combine in a workflow

When a workflow is built:

  1. The brief specifies brand + product + account + workflow type (sales / growth)
  2. The Manager loads the brand file to get voice / banned words / CTA style
  3. The Manager loads the product file to get claims / ingredients / mechanism
  4. The Manager loads the account file to get avatar reference / archetype / wardrobe default

All three feed into the Script Writer and Visual Planner downstream.

What's shared vs. account-specific

Element Source Per-account variation?
Brand voice Brand file No — same for all accounts
Banned words Brand file No
Approved claims Product file No
Product photo Product file No
Avatar reference Account file Yes — per account
Wardrobe default Account file Yes
Setting default Account file Yes
Sales channel Account file Yes (different accounts may use different channels)

This is the same STANDARD vs CUSTOMIZED logic from Chapter 7, applied at the brand/product/account level instead of the workflow level.

File naming conventions

Layer Path Naming
Brand reference/brands/{brand-slug}.md Lowercase kebab-case
Product reference/products/{product-slug}.md Lowercase kebab-case
Account reference/accounts/{account-code}.md Account code (e.g. account-a.md)
Avatar reference photo reference/avatar-sheets/{account-code}.{jpg,png} Account code

The {account-code} is the short code (Account A, Account B, etc.) used in workflow filenames, tab names, and the version registry. Consistent across all files.

Adding a new layer

The Chapter 9 — Add a Product or Account chapter walks through how to add new entries. The mechanics are straightforward once you know the structure:

  • Create the file at the right path with the right shape
  • Register it (where applicable) in the version registry
  • Upload any associated assets (avatar reference photo) to R2 and record the URL

When you're ready

→ Next: Avatar Archetypes and Reference Sheets — the persona patterns we use for accounts and how the reference photo works.