Audits every brand surface - product, marketing site, social, email, decks, ads - against the brand's locked fixed core and computes a drift score with a per-surface conformance table and a fix routing plan. Use when someone asks "audit our brand consistency across every surface", "our marketing site and product do not look like the same company", "how far has our brand drifted from the guidelines", "score how on-brand our decks and emails are", or after a rebrand ships and templates start mutating. Do NOT use for creating the brand system itself - use brand-guidelines, color-palette-builder, or typography-system instead; for checking one asset before it ships, use brand-application-review; for coherence across one campaign's asset set, use campaign-coherence-review; for verifying a built UI against its design spec, use design-qa-checklist.
---
name: brand-consistency-audit
description: Audits every brand surface - product, marketing site, social, email, decks, ads - against the brand's locked fixed core and computes a drift score with a per-surface conformance table and a fix routing plan. Use when someone asks "audit our brand consistency across every surface", "our marketing site and product do not look like the same company", "how far has our brand drifted from the guidelines", "score how on-brand our decks and emails are", or after a rebrand ships and templates start mutating. Do NOT use for creating the brand system itself - use brand-guidelines, color-palette-builder, or typography-system instead; for checking one asset before it ships, use brand-application-review; for coherence across one campaign's asset set, use campaign-coherence-review; for verifying a built UI against its design spec, use design-qa-checklist.
---
# brand-consistency-audit
Brand drift is a supply problem, not a discipline problem: people ship from whatever file is nearest, and every unaudited quarter multiplies the templates that quietly replace the brand. Identities are lived overwhelmingly on mundane surfaces - email footers, sales decks, ad templates - so drift concentrates exactly where nobody senior is looking. Without a numbered audit, "stay on brand" is an opinion war; with one, it is a **worklist**. This skill is the pack's first move: it names which surfaces are bleeding and hands every other skill its jobs.
## Operating procedure
Run the steps in order, because the order is the argument. Constants must be written down before anything can be graded against them; surfaces must be inventoried before grading so the embarrassing ones cannot be quietly skipped; grading precedes scoring precedes routing, because the score is only the case for the routing. Skip to the score and you get a number nobody trusts and a fix nobody owns.
### Step 1: Elicit the client's charter and the surface inventory
Gather these inputs before grading anything. If any input is a guess, label it a guess and move on.
1. **The charter** - the fixed-core contract of 3-5 locked constants and 3-5 designed-to-flex variables (a system that locks everything dies, one that locks nothing dissolves). Paste the charter block if one exists. If the client has a charter but the foundations pack is not installed, elicit the needed values directly and label them provisional. If no charter exists at all, halt and route to premium-design-foundations - never substitute worked-example values. Auditing against inferred rules produces inferred verdicts. **Precedence:** the charter is the single ledger of constants - guidelines, compiled tokens, and campaign briefs cite it and may narrow but never relax a value; where two documents disagree, the charter wins and the relaxing document is itself the violation.
2. **The surface inventory** - default seven: product web, product iOS/Android, marketing site, social templates, email templates, sales decks, paid ads. Tag each surface **revenue**, **launch**, or **internal**, because the tag drives the verdict cap in Step 3.
3. **Real current artifacts per surface** - screenshots, exported decks, live URLs - plus the measurements the artifacts cannot yield by inspection: accent coverage measured per premium-color-discipline, motion durations and damping ratios pulled from source code or frame capture. Never audit from memory; never eyeball a measurement.
4. **The misuse gallery** if one exists (misuse-gallery-builder), used as the grading rubric - the same per-class do/don't sheet reviewers police, at its convention of at least 6 wrongs per element class.
### Step 2: Grade every surface against every constant, declaring each check's evidence mode