Message Consistency Check
What this does
Takes your customer-facing material together and finds the places where your core message is not telling the same story. It reports where the category you claim, the primary value you lead with, the audience you speak to, and the proof you cite differ from one surface to the next - then ranks those mismatches so you fix the ones that matter first.
No single person usually reads the website, the sales deck, the pricing page, the nurture emails, and the ad copy in one sitting. That is why messaging fractures quietly after a repositioning or a fast launch. This job does the reading no one has time for.
Why this is a Fable-recommended job
Consistency can only be judged by comparison, which means every surface has to be in view at the same time. Claude Fable 5 can hold your full set of live copy in a single run and line it up surface against surface, which is what makes the mismatches visible.
If Fable is not available or the budget is spent: pick one surface as the reference (usually the current homepage or the approved messaging doc), then compare each other surface against it one at a time. Slower, and it can miss mismatches between two non-reference surfaces, but it works.
Evidence to gather first
- Homepage and key website pages
- Sales deck and pitch deck
- One-pagers, solution briefs, battlecards
- Pricing page
- Email sequences and nurture copy
- Paid and social ad copy
- Product UI copy and onboarding, if relevant
Read .agents/product-marketing-context.md first if it exists. If there is an
approved messaging framework, use it as the reference standard everything is
checked against.
Process
- Establish the reference. Either the approved messaging framework, or - if there isn't one - the current homepage as the de facto source of truth. State which you used.
- For each surface, capture how it renders the core message: category, primary value, target audience, top three claims, and proof used.
- Compare across all surfaces. Log every place a surface diverges from the reference or from the others.
- Rate each mismatch by surface visibility (how many buyers see it) and stakes (how central the divergent element is).
- Produce the report.
Output format
- Summary - surfaces analyzed, reference used, count of mismatches by severity.
- Consistency table - message element (category / value / audience / claim / proof) | what the reference says | where it diverges and how | surface | severity.
- Priority fixes - the handful of mismatches to fix first, with the specific edit for each.
- Consistency watchlist - lower-stakes divergences worth cleaning up over time.
The evidence standard
Every mismatch must quote the exact wording from each surface involved and name the surface. Do not report a mismatch you cannot show with the actual copy. When two surfaces differ, present both versions rather than deciding which is "correct" unless the reference framework settles it. Severity ratings should be explained in one line, not asserted.
