A brand is recognised through repeated, consistent encounters, and those take years. The final chapter covers two things that only show over time: how to keep a brand from changing on a whim, and how practitioners keep what they say in line with what they can actually do.
Quick read: the core statement, when it comes up, and the test to take home. Full read also opens the reasoning, the limits and the sources. Each principle comes before its practices, so the numbers do not run in sequence.
Keep the constants, renew the variables: creativity makes customers notice, consistency makes them remember, and creativity adds to the brand only when it carries the constants with it.
When it comes upWhen planning a new campaign, seasonal packaging, a new website design; when the team says the brand has grown stale; when a new creative agency proposes a radically different direction.
Resolution
Divide the brand into constants and variables. The constants, the brand assets (1.02), are the identity assets customers use to recognise the business: the core colour, the mark, the typeface, a character, a form of address. The variables are each campaign's story, imagery and ideas. Creativity goes entirely into the variables, and a creative idea is accepted only when it carries the constants with it. A hypothetical example: a fish sauce bottle keeps the same label and cap colour for ten years and tells a different story around that label each season; customers never get lost, and each new story has something to hold on to.
Full reasoning
Two goods pulling against each other
Creativity is genuinely needed: buyers only notice what gives them a reason to, and a bland campaign goes unseen however often it is repeated. Consistency is also genuinely needed: memory builds only on meeting the same signal again.
Why it holds
Buyers pay very little attention and forget very quickly (0.2), so every time a brand changes a constant, their memory must start again from scratch instead of adding to what is there (0.4). As Mark Ritson keeps reminding us, consumers do not get tired of ads, only marketers do: people inside the business meet their identity assets every day, customers meet them a few times a year. The first to feel bored are therefore the makers, not the buyers. The practical consequence: proposals to change usually arise from that boredom rather than from customers, so ask where a proposal comes from before discussing whether it is good. On the other hand, a brilliant campaign that replaces every constant may be remembered without anyone remembering who made it; the creative effort then falls into a void, sometimes into a competitor's name. Byron Sharp and Jenni Romaniuk make this mechanism the foundation of building brand assets (1.02): an asset becomes an asset only when used long and consistently enough for customers to link it to exactly one brand.
When it fails
When a constant itself is broken, keeping it is keeping a burden: a constant tied to an incident, confused with others in the category, or unable to live in new channels. Then the question of whether to change belongs to 1.03. A new business with nothing to keep starts its consistency today, and the early stage may experiment more. Consistency does not mean repeating one advertisement forever; it is about the constants, not the content.
What it costs
The creative team must work within a narrower frame and turn down good ideas that cannot carry the constants. The brand rarely surprises in form. Practitioners seldom get to make showy pieces for their portfolio, because the most valuable work is keeping everything looking familiar.
Sources
Byron Sharp, How Brands Grow, Oxford University Press, 2010
Jenni Romaniuk, Building Distinctive Brand Assets, Oxford University Press, 2018
Mark Ritson, Consumers don't get tired of ads, only marketers do, Marketing Week (column, around 2023)
Judge every new proposal by one question first: does it keep, transform by the rules, or lose any of the brand's constants?
When it comes upWhen approving a campaign, seasonal packaging, a new website design, materials from an outside agency. The name-covering test in 1.04 is for finding assets; this test uses the assets found to guard each proposal.
A test to take home
(1) Have the list of constants ready, the identity assets customers use to recognise the business (12.01), taken from the keep group of the inventory (1.05): core colour, mark, typeface, form of address, images or phrases customers already recognise. (2) Set the proposal beside the list and mark each constant: kept as is, transformed under existing rules, or lost. (3) Cover the name on the proposal and show it briefly to someone outside the business: can they guess whose it is? (4) Each lost constant must have a written reason and go through the change rules (12.02); without a reason the proposal goes back for revision, however good it is. Pass: the proposal is new in content yet outsiders still recognise the brand. Fail: the proposal is praised as brilliant but a glancing viewer guesses another name, or the only reason for dropping a constant is to look new.
Rules for change must be written when nobody wants to change anything, because when the need to change arrives, it comes with a deadline and someone who already loves the new option.
When it comes upWhen handing over an identity or guidelines; when the business is about to have new leadership, new products, new branches, franchise partners; when setting governance rules for an asset library.
Resolution
Rules for change do not forbid change, they decide by which route change happens. A truly new situation opens the rules through exactly the procedure the rules record, not by going around them.
Full reasoning
Two goods pulling against each other
Flexibility is good: markets change, products change, the business must react quickly without being bound by any book. Constraint in advance is also good: it shields the accumulated assets from hasty decisions.
Why it holds
Decisions to change a brand are rarely made under normal conditions. They arrive with a new director wanting to leave a mark, a competitor that has just refreshed, a product that must launch in three weeks, and usually someone who loved the new option before anyone asked what it destroys. In those moments the gains of changing are immediately visible, while the losses (recognition built over years) are invisible. Jon Elster describes how people bind themselves in advance, like Ulysses tied to the mast before hearing the sirens' song, knowing they will not be clear-headed then. A brand's rules for change are that rope. They need three layers: constants (12.01) change only by the owner's decision with a written reason (11.08); rule-bound variables may add variants following existing templates; free variables, such as each campaign's content. Along with: who may propose a change, what evidence they must bring, who decides, from what date the old version stops being used once changed, and who informs workshops, printers and outside agencies.
When it fails
A real legal or safety incident, such as a trademark dispute or a new regulation banning a word on labels, means change at once and record it afterwards. A business too early to know what its constants are should keep the rules short and clearly marked as provisional. Rules for change longer than one page are not read when needed.
What it costs
Good ideas must also go through the procedure and move more slowly. Whoever wants a change must write down a reason, which many find tiresome. Sometimes a real opportunity is missed because of the procedure; that is the price of keeping changes made on a whim from slipping through.
Sources
Jon Elster, Ulysses and the Sirens: Studies in Rationality and Irrationality, Cambridge University Press, 1979
Review the brand standards on a fixed schedule, and call a part outdated only when it no longer does its job, not when people inside are tired of looking at it.
When it comes upA brand that has been running for a year or more. This test reviews the standards themselves; whether use follows the standards is checked under 11.05, and which step of adjustment to take follows 1.08.
A test to take home
(1) Schedule a review once a year in a fixed month, in the calendars of the owner and the standards keeper, not tied to the arrival of new leadership or a competitor's change. The review month falls at least three months before the artwork deadline for the year's biggest season. The milestone is the artwork deadline, not the holiday: Tết or Mid-Autumn products usually lock artwork many months before the holiday. A review close to the season reaches conclusions too late to change templates, and the old version gets printed for another year. (2) For each part of the standards, ask whether it still does its job: does it display in new channels and new display sizes; does it meet new labelling or advertising regulations; do new products have a place in the system; do new materials and workshops produce it correctly. (3) Each proposed adjustment must state the job that part no longer does. (4) Items that qualify go into the change rules (12.02) and are adjusted at the lowest sufficient step (1.08). Pass: a short adjustment list, each line with a specific job that has broken. Fail: the reasons are looked at it too long, not on trend, a competitor just refreshed, or the same thing is adjusted every year.
A practitioner speaks with certainty about what the craft can verify, with conditions about what depends on the user, and promises nothing the market decides.
When it comes upWhen pitching, writing proposals, answering the question will doing this sell more; and when commenting on other people's work.
Resolution
Divide what you say by what you can verify. Speak with certainty about the craft, so certainly that clients need no extra promise about the market.
Full reasoning
Two goods pulling against each other
Clients need someone who dares to speak with certainty so they can decide to spend, and a hesitant practitioner is hard to hire. On the other hand, cautious words keep speech from running ahead of work.
Why it holds
What a brand practitioner says falls into three kinds. The first is what the craft can verify: diacritics that do not collide with the line (9.02), colour within its tolerance range (8.03), templates that hold the longest product name (10.07). These can be measured, so they are said with certainty and the speaker is accountable. The second depends on how the business uses the work: whether the brand is still recognised after a few years depends on keeping the standard (11.02); it can be said, but with conditions. The third is what the market decides: sales, market share, growth rate. Nobody controls it, not even the best experts. Philip Tetlock followed experts' forecasts on politics and economics over many years and found those confident forecasts often did little better than simple extrapolation models; that research did not measure sales, but there is no reason to think a brand's market is easier to predict. The pressure of a pitch pushes practitioners to speak with certainty about the third kind too, and that certainty borrows credibility from the first (0.7). When the market promise fails, clients lose trust even in what the practitioner said correctly.
When it fails
When a practitioner has measured the same thing many times across many real projects and has their own data, part of the second kind can be said more confidently, along with that data. Speaking with conditions does not mean speaking vaguely: the conditions must be specific enough for the client to check. This also applies when commenting on others' work: speak only as much as you know of the brief and the conditions they had (12.07).
What it costs
Losing some work in pitches where the listener wants a big promise. Proposals must be longer to state clearly which parts are certain and which are conditional. And real accountability for whatever was stated with certainty.
Sources
Philip E. Tetlock, Expert Political Judgment: How Good Is It? How Can We Know?, Princeton University Press, 2005
Every promise in a capability profile must point to work already done; any sentence that cannot point is rewritten or dropped.
When it comes upWhen a business writes its capability profile, about page or website welcome; and when reading the capability profile of a supplier it plans to hire.
A test to take home
(1) Print the profile and highlight every claim about yourself: quality, experience, process, commitment. (2) Beside each, note the evidence it points to: a project done, a photo, a measurement, a person who can confirm. (3) For a sentence without evidence, choose one of two: rewrite it as what you are doing or plan to do, saying exactly that; or drop it. (4) Replace each self-praising adjective such as reputable, leading, professional with something specific the reader can check. As for words like most, only, best, number one: when the document is used as advertising material, Vietnam's Advertising Law forbids them without lawful documents proving them; whether your profile falls under that case and which documents count as proof, ask a legal specialist before keeping them. Pass: every remaining sentence has a line of evidence, and the profile gets shorter. Fail: most sentences are dropped and none is rewritten as specific work, meaning the profile is selling words, not work. When reading the profile of a supplier you plan to hire, do the reverse: ask them for evidence for the three sentences you care about most.
Sources
Sources
Vietnam Advertising Law 16/2012/QH13 (amended by Law 75/2025/QH15), Article 8, clause 11
When taking over a brand someone else made before, judge their work by the brief and conditions they had, and talk about what is broken, not about the person.
When it comes upWhen a business changes brand agency, when inventorying an old identity (1.05), when the client asks whether what the previous agency did is acceptable.
A test to take home
(1) Before commenting, ask what the previous maker had: the brief, budget, timeline, decider, and the materials and channels of the time. (2) Sort each part of the old set into three groups: living assets to keep; choices right at the time that have since aged because conditions changed (12.05); genuine craft mistakes. Most of what looks wrong usually falls in the second group. (3) Speak to the client about the fault and its consequences, not about the maker: diacritics colliding with the line at sign size so the words read wrong, instead of the previous agency was sloppy. Pass: after hearing it the client knows what to fix and what to keep, and no sentence ties a fault to a person's or a firm's name. Fail: the comments contain adjectives about people, or everything in the old set is sorted into the mistakes group. Today's practitioner will also be someone's predecessor one day. The AIGA Standards of Professional Practice (the American professional association for graphic design) record the same spirit: comments on others' work must be objective, balanced, and must not disparage their work or reputation.
Sources
Sources
AIGA, Standards of Professional Practice, section 2.4
