The marketing artefact agencies and in-house teams use to keep voice consistent across writers and channels. Honest 2026 definition, standard structure, practical limits, and the comparison to voice prompts for AI-driven content production.
A brand voice document codifies how a brand communicates in writing — voice attributes, do/don't lists, audience descriptions, sample content. Designed for human writers to read before drafting; not designed for AI tools. Standard length 6-15 pages. Voice prompts have overtaken brand voice documents as the primary production asset for AI-driven content but the two complement rather than replace each other when human and AI production run alongside.
A brand voice document is a written artefact that captures and codifies how a brand or person communicates in writing. The purpose is keeping voice consistent across writers, channels, and time. Brand voice documents typically include voice attribute descriptors, do/don't lists for word choice, audience descriptions, brand value alignment statements, and sample content showing the voice in use.
The artefact originated in agency and in-house marketing teams managing many writers. As brands scaled their content production from one or two principals to teams of 5-20 writers across blog, social, email, and sales channels, the need for a centralised voice reference document emerged. Pre-2020s, the document was sometimes called a "tone-of-voice guide" or "editorial style guide"; the broader "brand voice document" terminology became standard around 2020-2022 as content marketing matured.
Most brand voice documents in 2026 share a six-section structure:
1. Voice attributes (1-2 pages). 4-8 descriptors like "direct", "warm", "analytical", "playful", with explanations of what each one means in practice. Often paired with negative attributes ("not corporate", "not breathless") to triangulate the actual register. Voice attributes are the reference layer most writers consult when in doubt.
2. Audience description (1-2 pages). Who the voice speaks to. Demographics, psychographics, professional context, awareness level, sophistication. Some documents include 1-3 audience personas with specific names and details.
3. Do/don't lists (2-4 pages). Specific vocabulary and phrasing the voice uses or avoids. "Do say 'customers'; don't say 'users'." "Do use contractions; don't use 'i.e.' or 'e.g.'." This section gets heavily consulted by writers in active drafting.
4. Brand values alignment (1-2 pages). Why the voice exists. The link between the brand's positioning, mission, or values and the voice choices. Useful for stakeholder communication and onboarding more than for active drafting.
5. Sample content (2-4 pages). Examples of the voice in use across formats — email, social, blog, sales page, customer service. Most useful section for writers learning the voice for the first time.
6. Common scenarios and tone shifts (1-2 pages). How the voice shifts across context. The same brand might be "playful but professional" on social, "warm and clear" in customer service, "direct and analytical" in B2B sales. The shifts are part of the voice; capturing only one register produces wrong-toned output elsewhere.
Total length: 6-15 pages for most useful brand voice documents. Some agency-produced documents balloon to 30+ pages; the additional length rarely produces additional consistency in practice and often becomes shelf-ware.
Three patterns of use observed in marketing operations:
1. Onboarding new writers. The document is required reading for any new contractor or hire. The new writer reads the document, drafts a sample piece, gets editorial feedback referencing the document. Usually 2-4 cycles before the writer's output reads on-brand.
2. Active drafting reference. Writers consult the document when in doubt about specific choices. Most-consulted sections are do/don't lists and sample content. The voice attributes are read once at onboarding and rarely re-consulted.
3. Stakeholder communication. The document is shared with executives or clients who want visibility into how their brand is being represented in writing. The brand values alignment and audience description sections do most of the work in this use case.
Typical lifecycle: drafted at brand launch or rebrand, refreshed every 18-36 months as brand or audience evolves, retired when the brand undergoes major repositioning. Documents that go more than 36 months without refresh tend to drift out of alignment with how the brand is actually being used.
Both artefacts capture how a brand communicates. The difference matters because they are designed for different production paths.
| Dimension | Brand Voice Document | Voice Prompt |
|---|---|---|
| Designed for | Human writers reading before drafting | AI tools following mechanical rules |
| Length | 6-15 pages | 500-800 words |
| Structure | Descriptive — attributes, examples, scenarios | Structural — sentence length range, banned-word counts, signature move examples with numerical constraints |
| Usage | Read once, consulted occasionally | Loaded into AI as instructions; runs every conversation |
| Production | Agency-led or internal-led brand work | Pattern extraction from existing samples |
| Cost | £4-15k agency or 12-28 hours internal | £497-997 DFY or 4-6 hours DIY |
| Output | Reference document for human use | Working asset that drives AI output |
| Best for | Teams of 3+ human writers, stakeholder communication | Solo operators, AI-driven production, voice consistency across tools |
The structural difference matters. Brand voice documents say "be direct"; voice prompts say "sentence length range 4-22 words, average 11, with at least 3 sentences under 8 words per 200-word draft, banned words: very, really, just, actually, basically." The first is descriptive guidance a human writer can interpret; the second is mechanical rules an AI tool can follow exactly.
Detail in voice prompt vs Custom Instructions for the layered AI-tool perspective and how to build a voice prompt for the construction methodology.
Solo operators using AI tools: voice prompt only. The voice prompt covers the same ground more efficiently for AI-driven production. A brand voice document at solo scale is over-engineered.
Teams with 3+ human writers: brand voice document. Required for onboarding and consistency across writers. AI is supplementary at this scale.
Hybrid teams (human writers plus AI-driven production): both. The brand voice document onboards human writers; the voice prompt drives AI tools. Same underlying voice; two delivery mechanisms.
Agencies producing for multiple clients: brand voice documents per client (for human delivery) plus voice prompts per client (for AI-augmented production). The agency runs both because client teams expect the document and modern production economics expect the prompt.
Modern marketing operations between £1m and £10m revenue: typically benefit from both. The brand voice document handles stakeholder visibility and writer onboarding; the voice prompt handles content production at scale.
Two reasons:
1. AI is now drafting most volume content production. For marketing operations producing more than 20 pieces per month, AI handles the bulk of first-draft generation. Brand voice documents written for human writers do not load directly into AI tools and need translation. Voice prompts skip the translation step.
2. Structural rigour produces more consistent output. The shift from descriptive ("be direct") to structural ("sentence length 4-22, average 11, with specific variation requirements") produces more reproducible voice match across both human and AI output. Writers consulting brand voice documents interpret descriptors differently; AI tools running voice prompts apply rules exactly.
Brand voice documents are not being replaced entirely. They remain useful for onboarding human writers, stakeholder communication, and brand-level voice education. The shift is in primary production: voice prompts handle the day-to-day output; brand voice documents handle the strategic and onboarding work.
For brand voice documents: agency-produced typically £4-15k over 2-4 weeks. Self-paced internal: 12-28 hours of focused work plus stakeholder review. Common methodology: voice extraction from existing samples + audience research + values alignment + sample content production. Several agencies specialise in brand voice development; rates vary by agency tier.
For voice prompts: 4-6 hours DIY using the discovery methodology and the construction guide, or 2-3 working days at £497-997 via the Syxo DFY Voice System.
For both (the hybrid approach): build the voice prompt first because it forces structural rigour. Use the voice prompt as the foundation for the brand voice document, expanding from 500-800 words to 6-15 pages by adding audience description, brand values alignment, and sample content. The two artefacts share underlying voice analysis; the document adds context the prompt doesn't need.
Three patterns of failure observed in agency-produced and internally-produced brand voice documents:
1. Vague attribute descriptors. "Confident but humble", "professional yet warm", "authoritative but accessible". The contradictory pairings sound balanced but provide no usable guidance. Writers cannot operationalise "confident but humble" without examples and constraints.
2. Missing tone-by-context shifts. Most brand voice documents capture one register and assume it applies everywhere. The honest brand voice has 4-7 contextual shifts (email vs social vs sales page vs customer service). Documents missing this section produce wrong-toned output in non-default contexts.
3. Sample content written by the agency, not extracted from the brand. Sample sections often contain content the agency wrote to demonstrate the voice rather than content the brand has actually produced. The result is a document that describes a voice the brand does not actually use yet — aspirational rather than descriptive.
DFY Voice System ships the voice prompt, Custom GPT, Claude Project, hook library, and profile rewrite in 2-3 working days at £497 founder pricing. The voice analysis underneath maps directly into a brand voice document if your team needs one.
See The Voice BuildA written artefact that codifies how a brand communicates in writing. Standard structure: voice attributes, audience, do/don't lists, values alignment, sample content, scenarios. 6-15 pages typical.
Six sections: voice attributes, audience description, do/don't lists, brand values alignment, sample content, common scenarios with tone shifts.
Brand voice documents are descriptive guidance for human writers (6-15 pages). Voice prompts are structural rules for AI tools (500-800 words). Different production paths.
Solo operators: voice prompt only. Teams with 3+ human writers: both. The two complement rather than replace each other.
Agency: 30-60 hours over 2-4 weeks at £4-15k. Self-paced internal: 12-28 hours plus stakeholder review.
AI now drafts most volume content; voice prompts load directly into AI tools while brand voice documents need translation. Structural rigour also produces more consistent output.