November 17, 2026

How to Keep a Brand Coherent as It Evolves and Prevent Identity Drift

Why coherence gets built in editorial decisions, one brief at a time

A style guide can't stop a brand from drifting. Most brands that lose their identity in public pass every compliance check on the way there. The logo sat in its clear space, the colours matched the swatches, the tone was friendly and confident as instructed. Nobody broke a rule, and the brand still ended up saying things it had never said, to people it had never addressed, in a voice nobody chose.

The usual definition of drift is a compliance problem: off-shade colours, stretched logos, templates edited by helpful colleagues. Those are real and easy to fix. The costlier version is a loss of meaning, where an asset follows every guideline and still contradicts what the brand stands for. Northwest Creative draws the line well. Whether a change counts as evolution or creep depends on whether anyone decided it. Evolution has an author, and creep is what builds up when many people make one-off choices at different times, and none of them agreed to the result.

Brands have to evolve, because markets move, products multiply, and audiences change their minds. The work is to separate what must stay recognisable from what should stay free to change, then build the editorial machinery that keeps that line visible to every contributor. In a content-led organisation, that machinery decides the outcome, since every article, email and post adds a small deposit to what the audience believes the brand is.

How sensible decisions add up to a brand nobody chose

Picture a specialist risk advisory firm (a composite). It built its reputation on rigorous, evidence-led views about one narrow problem. Over the next few years, it opens in new regions and adds service lines, a couple of them through acquisition. Each region writes its own thought leadership, and each service line needs leads. The search team finds that broad, trend-driven topics bring traffic the specialist ones never did. Some campaigns work, and some don't, and none of it looks alarming. Eventually the homepage promises "innovative solutions from a trusted partner," and prospects describe the firm in the same words they use for its competitors.

No team made a bad call. Growth adds contributors, approval layers, agencies and channels faster than it adds shared understanding. Short-term performance pressure adds to it, since content that attracts attention gets rewarded sooner than content that reinforces position, the tension Binet and Field describe between activation and brand-building. Assets get reviewed one at a time. Audiences meet them all at once.

The commercial case is hard to pin down. Lucidpress (now Marq) surveyed marketers and linked consistent brand presentation to revenue gains of 10 to 33 per cent, depending on the edition. The data is self-reported, and well-run companies tend to be both consistent and growing, so I'd read it as a signal and leave causation open.

Clearer evidence comes from a brand that lost its cues overnight. PepsiCo replaced the Tropicana Pure Premium carton in January 2009, dropping the orange with the straw. Sales reportedly fell about 20 per cent, roughly $30 million, and the old design returned within two months. That was a redesign, which makes it a useful contrast with drift: a sudden change shows what lost recognition costs in eight weeks. Drift charges the same price in instalments.

Separate the anchors from the expressions

Working definitions: consistency means recognisable patterns in communication. Coherence asks whether those patterns express one identity and one meaning. Drift is the accumulation of changes that erode that alignment without anyone deciding.

Sort the brand into anchors and expressions. Anchors are positioning, the promise to the audience, the distinctive perspective and the beliefs behind the expertise. Expressions include campaign themes, formats, channel language, examples and the topics of the moment. Anchors can change, but only through a repositioning someone chooses and can explain. Expressions should change constantly.

• Positioning. Anchor: clear market role and differentiation. Expression: which competitors to frame against this year.

• Brand promise. Anchor: consistent expectation of value. Expression: supporting benefits and proof points.

• Voice. Anchor: recognisable personality and principles. Expression: tone by channel and situation.

• Content perspective. Anchor: consistent expertise and beliefs. Expression: topics, formats, examples.

• Creative identity. Anchor: distinctive characteristics. Expression: campaign concepts and treatments.

For ambiguous elements, ask three questions. Would changing it alter what customers expect? Would a long-standing client notice? Could you explain the change to that client as a deliberate choice? A yes to any one makes it an anchor.

Attach a reason to every rule. "No hype language" wears away quickly, but "no hype language, because our authority rests on evidence" gives a contributor something to reason from when a situation comes up that the rule never covered.

Apple has added product categories for decades while holding to a perspective on simplicity and design-led experience. The 1997 Think Different campaign celebrated creative rebels without showing a single product, at a time when the company was close to failing. Later campaigns changed the expression and kept the logic, and even the logo went from rainbow apple to monochrome silhouette without losing recognition. Nobody could preserve every historical expression, so Apple preserved the logic that ties them together.

Make content strategy the translator

Guidelines that call a brand "bold but approachable" describe a personality and leave a writer with no way to decide what to publish on Tuesday. Content strategy turns positioning into decisions about topics, angles, evidence standards and which arguments the brand will take on. Pillars should follow the brand's areas of authority, and the industry's topic of the quarter shouldn't dictate them. Douglas Holt's work on cultural branding points the same way: brands gain power by owning an idea or tension in culture, and content is where that ownership is built or slowly rewritten.

Before production starts, every major brief should answer these:

1. Which part of the brand's positioning does this reinforce?

2. What perspective should the audience associate with the brand afterwards?

3. Which real audience need does it address?

4. What would make this recognisable as ours with the logo removed?

The fourth question carries the most weight, and Jenni Romaniuk's distinctive-asset research at Ehrenberg-Bass gives a way to apply it. She scores assets such as shapes, colours, characters and sounds on fame (how many category buyers link the asset to the brand) and uniqueness (whether they link it only to that brand).

Her research covers assets like logos, colours, and sounds. Running the same two tests on editorial habits is an extension, not something she tested, but the logic carries over. Say the risk firm always publishes its methodology. Do buyers in the category associate that habit with the firm, and with nobody else? If so, the firm owns something no competitor's template can copy. If they don't, the habit is decoration, and it can be dropped without anyone noticing.

A messaging hierarchy backs up the brief. One primary narrative sits at the top, audience-specific messages beneath it, proof points beneath those. Without it, every team promotes its own message to primary, and the main narrative drowns.

Fragmentation is a reasoning problem

A LinkedIn essay, a sales email and a landing page should differ in language and share a point of view. Uniform execution is impossible across formats and wouldn't be worth having anyway.

Most fragmentation traces back to rules handed over without reasons. A freelancer who understands why the firm avoids superlatives can write a persuasive case study in that spirit. One who knows only the rule writes something flat, or guesses wrong. The number of contributors matters far less.

Heavy review doesn't close the gap. Adobe, citing Lucidpress data, reports that 71 per cent of brand professionals say seven or more people must approve a single on-brand asset, and 59 per cent say content often goes live before approval finishes. A queue that long gets bypassed.

Localisation is more than translation. Keep a global core of argument, methodology and beliefs, and let each market choose its own examples, references and proof points. The risk is that the firm's European and North American teams can use different case material and still make the same argument.

Channel specialists need a seat at the same table. An SEO team chasing high-volume keywords outside the firm's expertise is hitting its own target, and the brand pays for it. A shared brief puts both goals on one page.

The cheapest internal test I know costs nothing. Ask five people from different functions to describe the brand's positioning in a sentence. The spread between their answers measures the drift already there, and it shows up faster than any audit.

Read the pattern, not the asset

Drift stays invisible while assets are reviewed one by one, and it appears once they're read together. Gather the last 20 to 30 pieces across channels, strip the logos, and hand them to someone unfamiliar with the brand. Ask what the company is about and what it seems to believe, then compare the answer with the intended positioning. Add the language customers use in sales calls, interviews and reviews. The gap between intended identity and perceived image is where drift lives.

In the review meeting, I'd put five questions on the table. Does the combined content reinforce a recognisable market position? Do different assets sound like the same personality? Is authority piling up in the areas the brand wants to own? Does the audience's language match the intended identity? Were recent changes decided, or only noticed afterwards? These call for judgment. A numerical coherence score would invite teams to optimise the score.

Healthy experimentation has a time limit, a hypothesis and a review date. If a change has none of those, call it drift.

Governance that permits movement

Rules written to prevent every deviation slow production and produce formulaic work. A better structure has three tiers: anchors that stay fixed, matters that need a conversation, and decisions that belong to the team. Escalation triggers define the middle tier: a new claim, a new audience, a new topic area, a change in tone or personality, a comparison with a named competitor. Everything outside those triggers stays with the people closest to the work. Demanding senior sign-off on wording only produces the bypass behaviour described above.

AI-assisted production raises the stakes. Generative tools drift toward the generic middle unless they're given specific context, and they amplify whatever signals they receive, clear or muddled. Prompt drift is how that happens in practice. Picture a team that tweaks a prompt to sound punchier, saves the best outputs as examples, and finds a few months later that the tone differs across sectors with nobody having decided anything. The fix is to give the tools positioning, editorial principles and approved examples, and to keep a human editor wherever meaning is at stake. A tone description alone gives a model little to go on.

Coherence needs maintenance

A rebrand, a messaging workshop or a polished PDF can't secure an identity permanently. Coherence needs a place in the calendar: a review during quarterly planning and another after each major campaign.

Keep a decision log. One line per change is enough: what changed, which anchor or expression it touches, why now, the expected outcome, a review date. Within a year, it answers the question of why the brand sounds the way it does.

A test for any leader: can anyone name the last time the brand decided to change its message, as opposed to noticing it had changed? If that takes more than a few seconds, drift has probably started.

A log entry costs minutes. A repositioning nobody chose costs years.

‍