Picking a Brand Voice You Can Keep for Years

A startup I advised three years ago built its entire brand voice around one perfect tweet. It was genuinely funny, it went modestly viral, and the founder decided on the spot that this was who they were now: sharp, irreverent, always ready with a clever line. Eighteen months later, during a difficult quarter that included layoffs, that same voice was writing internal memos and customer apology emails, and it did not survive the transition. Irreverence that reads as confidence in good times reads as callousness in bad ones.

Most brand voice guides get written during a company's best week, when morale is high and the story is easy to tell with a smile. That timing is the actual problem, because a voice chosen at a peak rarely gets tested against the trough until the trough arrives, and by then it is load bearing across a hundred templates, a support team's habits, and a year of published content that all has to change at once if the voice turns out not to fit.

Why voices chosen at the peak tend to break

A brand voice is not really a style choice, it is a promise about how the company will sound in every situation, including the ones nobody wants to imagine while writing the guide. Clever and playful is easy to sustain when announcing a product launch. It is much harder to sustain, without sounding tone deaf, when writing about a service outage, a price increase, or a mistake the company made. If the voice cannot flex into those situations, the company ends up with two voices, the branded one and the real one that shows up under pressure, and readers notice the seam.

The fix is not to pick a duller voice out of caution. It is to test a candidate voice against the worst plausible week before locking it in, not just the best one. Write a mock apology email, a mock outage notice, a mock layoff announcement, in the proposed voice, before you commit to it across every channel. If the voice sounds wrong in any of those drafts, it is not durable, however well it performed in the launch tweet that inspired it.

Where most style guides fail in practice

The standard brand voice document lists adjectives, friendly, confident, knowledgeable, and pairs them with do and don't examples. This is not wrong, but it is close to useless on its own, because adjectives are interpreted differently by every writer who reads them. Two copywriters both aiming for "confident" will produce noticeably different sentences, and neither is wrong according to the document, which means the voice drifts slightly with every new hire until it is not really one voice anymore.

What actually holds a voice together over years is not the adjective list, it is a small set of concrete rules that are checkable rather than interpretable. Sentence length ceilings. A list of words the brand never uses, and why. A rule about whether the company refers to itself as we or by name. These are boring compared to adjectives, and they are the part that actually survives staff turnover, because a new writer can check their draft against a rule in a way they cannot check it against a mood.

An adjective tells a writer how to feel. A rule tells a writer what to do. Only one of those transfers reliably to someone who was not in the room when the voice was chosen.

A voice that has actually lasted

The most durable brand voice I have worked on belongs to a regional insurance company, which is not a category anyone associates with personality. Their voice guide is four pages, almost entirely rules rather than adjectives: sentences under twenty two words wherever possible, no exclamation points outside of confirmed good news like a claim approval, and every piece of copy read aloud before publishing to catch anything that would sound wrong spoken to a person who just had a car accident. Six years later, across three rounds of staff turnover, the voice is still recognizably itself, not because anyone remembers the original adjectives, but because the rules are still checkable by whoever is writing this week.

The startup with the viral tweet eventually rebuilt its voice guide from the outage notice outward instead of from the tweet outward, choosing a tone that could apologize sincerely first and be playful second, rather than the reverse. It is a less flashy voice than the original tweet promised, and it has now survived two difficult quarters without anyone having to write around it.

If you are choosing or revisiting a brand voice, write the hardest email first: an apology, an outage notice, a price increase. Whatever voice survives that draft without sounding wrong is the one to build the rest of the guide around, not the one from your best week. The opposite failure, a voice so locked down it stops sounding like anyone, is covered in when consistent messaging starts to sound robotic, and both live inside the wider work of brand messaging that has to hold up in a bad quarter, not just a good launch.