How Marketing Can Own the Company Voice Without Writing Every Post
Frankly Speaking Team
July 27, 2026 · 7 min read
Marketing teams have spent years building control over how the company sounds. The website, the deck, the campaign copy, every word passes through someone whose job is to keep the language consistent.
Then the company starts an employee LinkedIn programme, and the reach moves to a channel marketing doesn't own: personal profiles, written in personal voices, published under other people's names.
At that point most marketing teams pick one of two bad options. Either they take over, rewriting every post until the whole leadership team sounds like the brand page, the sameness problem we described in Voice Bleed. Or they stay hands-off and hope, which lasts until a director invents a product name, misquotes a customer number, or describes the company in a way legal has to unwind.
There's a third option, and it's the actual job: own the company's language, not the individuals' voices. The distinction sounds subtle. In practice it's the difference between a programme that scales and one that dies in review.
Separate what's shared from what's personal
Every post a team member publishes contains two kinds of language, and they need opposite treatment.
Company language is shared and non-negotiable. Product names and how they're capitalised. How you describe what the company does, in one sentence, the same sentence. Which customer names and numbers are approved for public use, and which have to stay anonymous. The claims you can make and the ones you can't. The words you never use, every company has them, whether it's a competitor's trademark or the word "cheap".
Personal voice is individual and protected. Sentence rhythm, formality, humour, favourite phrases, how much story versus how much data. This is the part that makes employee content outperform the brand page in the first place, and as we argued in Why AI LinkedIn Content Sounds Generic, it's precisely what generic AI content lacks. It's also not something marketing writes. It belongs to the person whose name is on the post.
Write the first category down. One page is enough: the one-liner, the naming rules, the approved numbers, the banned words. Most companies have this scattered across a messaging doc, a brand deck and three people's heads. The exercise of putting it on one page is where half the future review arguments get settled in advance.
Leave the voice with the person
Here's the part most marketing teams get backwards. They treat each person's voice as another thing to author and police, a house style to enforce per byline. It isn't. It's the one layer marketing should keep its hands off.
The way to make that safe is to define each voice explicitly, but from the person, not from marketing. A usable voice definition covers four things: formality (does this person write "we shipped it" or "we are pleased to announce"), style (storyteller, data-driven, tactical, contrarian), rhythm (short and punchy versus long and structured), and vocabulary, the register they use and the handful of phrases that are authentically theirs. Your CFO probably says "runway" a lot. Your Head of Product might never use the word "leverage". That definition comes out of how the person actually talks, captured from a short interview, not from a marketer's guess about how they should sound.
Two things happen once it's written down. First, drafting gets faster, because "make it sound like her" becomes a checklist instead of a feeling. Second, disagreements get resolvable. When someone says "this doesn't sound like me", you look at the definition together and fix the definition, instead of relitigating adjectives post by post. And the person, not marketing, signs it off. It's their name.
Voices also move. People change how they talk as their role changes, and a profile written in January is quietly wrong by June. The fix isn't a quarterly audit by the brand team. It's a feedback loop: as posts go out and gather engagement, watch which of a person's posts actually land with their audience, and let that suggest one small tuning to their voice at a time. The person accepts or rejects each suggestion. The voice evolves toward what works for them, and never without their sign-off. Marketing's whole contribution to this layer is discipline: protect it, don't standardise it.
Build one library, not five folders
The operational failure mode of team content is not bad writing. It's scattered inputs: the case study PDF in someone's inbox, the approved product screenshots in a design folder nobody can find, three versions of the messaging doc, and no one sure which is current.
The fix is a single shared library with a deliberate structure, and the structure matters more than the tool:
Team reference material is the drafting input: the messaging doc, positioning, the language rules, case studies, approved proof. This folder is what makes every draft start from the same facts. When the ghostwriter, or the AI, drafts a post that touches pricing, the current pricing narrative is in the reference folder, not in a Slack thread from March.
Team assets are the publishing output: approved images, charts, product visuals, event photos. Posts with visuals need them on-brand and current, and a shared assets folder means nobody screenshots an outdated slide because it was the one they had. Reference feeds the writing; assets go on the finished post. Keep them separate.
Personal folders hold each member's own material: their photos, their talk recordings, the podcast clip only they would post. Shared governance for shared things, personal space for personal things, the same principle as the voice split.
One person, usually the marketing lead, owns the library's hygiene: old versions removed, new proof added, one source of truth. It's an hour a month, and it's the cheapest brand-safety investment in the whole programme.
Run it through one seat
With the language written, the proof approved and the library stocked, the last piece is the operating model, and the strongest one for most teams is the internal ghostwriter: one operator drafting across the whole team from a single seat.
The reason this works for language management specifically is concentration. When five execs each write their own posts, the naming rules and the approved numbers live in five imperfect memories. When one marketer runs every pipeline, that shared language is applied consistently by the person who wrote it, and each individual's voice definition stops that consistency from collapsing into uniformity.
The interviews stay personal, short talking sessions with each exec, because spoken answers are where the individual voice actually lives. The drafting is central. And nothing publishes without the named person's sign-off, the approval step that keeps the byline honest: the marketer guards the company's language, the exec guards their own voice and hits approve.
That's the full division of labour, and it's worth stating plainly, because it's the answer to the question every leadership team asks at kickoff: "who controls what we say?" Marketing controls the shared language, the proof, the assets and the process. Each person owns their voice and their final approval. Nobody controls both.
Where the tooling fits
Everything above can run on documents and discipline: a one-page language sheet, a tidy shared drive, a diligent marketer. Teams do it. What Frankly Speaking changes is that the system enforces itself.
Every member has a voice profile, formality, style, rhythm, vocabulary register and their own natural phrases, drawn from their own interview and owned by them. Every draft is written through it, so the CEO and the Head of Sales stay distinct even when one marketer drafts for both.
The company language layer has three enforced parts. The glossary holds your product names and preferred terms, and it doesn't just prompt the model, it swaps the wrong variants for the right ones on every draft, so a banned spelling can't ship even if the writer misses it. Proof Points holds the customer results you're cleared to cite, each one marked as nameable or anonymous, optionally linked to the case study behind it, so a draft uses your exact numbers and never invents a customer metric. The brand brief carries the voice-of-the-company, the audience, the topics to lean into and the phrases to never use.
The Library gives the workspace exactly the structure above: a team reference folder the drafting engine actually reads, a team assets folder so posts ship with approved visuals, and personal folders for each member. The ghostwriter seat lets one operator run every member's pipeline, and the approval queue routes each draft to its named author, publishing under that person's own account only once they approve. And as posts perform, the system watches what lands for each author and proposes small, optional tunings to their voice, which a human accepts or rejects.
The point isn't the features. It's that the marketing team's real value in an EGC programme was never typing posts for other people. It's owning the language layer, the vocabulary, the claims, the proof, the assets, so that ten people can publish under their own names and every post is unmistakably theirs and unmistakably on-brand at the same time. Get that layer right and marketing stops being the bottleneck of the programme. It becomes the reason it works.
One seat. Every voice. Nothing off-brand.
Frankly Speaking gives your marketing team per-person voice profiles, a shared library, an approved glossary and customer proof, a central ghostwriting seat and an approval loop, so the whole team publishes without the brand losing the plot.