Evidence-Based Profile

A content model for brandbook packages, grounded in evidence-based marketing science: Byron Sharp (How Brands Grow: penetration, light buyers, mental and physical availability), Jenni Romaniuk (Building Distinctive Brand Assets, Category Entry Points), and Mark Ritson (diagnosis → strategy → tactics; category defined in buyer terms). It is the content model the authors of brandbook.md use and recommend — and, like every profile (base spec §1.1), entirely optional.

A package opts in through root frontmatter (F-14):

profile: evidence-based

The declaration is per package root and is never inherited: in a multi-brand estate (base spec §7), each child that wants this profile’s checks declares it itself. Every rule in this profile is advisory — validators surface violations as warnings, and no profile rule affects base conformance (§9).

1. Required knowledge#

• EB-03. A package SHOULD carry a type: dba file (its Distinctive Brand Assets with usage rules, per dba-inventory-spec.md) and a type: cep file (its prioritised Category Entry Points, with stable CEP-xx ids per C-03). Rationale: Romaniuk — distinctive assets and mental availability are the two growth levers this profile exists to govern; a package that records neither is a style guide, not a growth instrument.

2. Root sections#

• EB-01. The root body SHOULD contain sections 1 to 7 below, in this order, as ## headings; sections 8 to 11 are further options. Validators warn on missing or out-of-order recommended sections. Child packages that declare inherits: are exempt even from the warning: their omitted sections resolve from the parent (H-01). Rationale: a predictable shape keeps human navigation and agent prompts portable across brands, while a missing section is an honest gap to fill, not a reason to reject the package.

3. Narrative genres#

The profile draws a hard line between operational and narrative content, sharper than the base standard’s §5.2. These genres do not pass the actionability test, so under this profile they live in context/, never in core files:

• Mission, vision, and purpose statements. Often genuinely important to the organisation, and rarely an instruction anyone can execute. In context/ they stay available to every human reader without spending the token budget of every agent call.

• Persona archetypes (“The Explorer”, “The Sage”). Sharp and Romaniuk show that a brand’s buyers are largely the category’s buyers; under this profile the operational buyer definition is a category-buyer definition in audience/, and archetype work belongs in context/. Organisations that work with personas operationally simply use the base standard without this profile — it accommodates them fully.

• Values lists. Words like honest, innovative, and human carry no operational instruction on their own. The actionable translation is a voice rule or an anchor rule; the list itself can live in context/.

• Manifestos. Often the best writing in a legacy brand book, and still not something a generator or approver can act on. context/ keeps them readable without making them load-bearing.

• EB-02. Validators SHOULD warn on core-file sections whose headings match the narrative genres above (e.g. “Our Mission”, “Brand Values”, “Manifesto”, “Archetype”) and point authors to context/. The check is advisory: conformance is decided by structure, never by content. Rationale: the profile coaches authors towards actionable core files; warning instead of failing keeps adoption low-friction.

Migration tooling SHOULD move such material to context/ and link it from there, not delete it.