# A brand audit checklist: what to measure, what to test and what to fix first

> A brand audit has two halves: an inventory of what you publish, which you can measure, and research into what people make of it, which you cannot judge yourself. Measure colour drift, logo proportions, contrast and names. Test message and recognition on outsiders. Rank fixes by reach times damage.

By LAIN, 30 Jan 2026. 12 min read. Category: Branding. Canonical: https://lainagency.com/resources/brand-audit-checklist

Most brand audit guides end in a spreadsheet of opinions. This one rests on a split. Some things can be **measured**: colour, logo proportions, contrast, the spelling of your name. Others cannot be judged by the people who made the brand and have to be **tested** on outsiders: whether the message lands, whether the look is recognised, what customers say about you. Measure the first group, test the second, then rank the fixes by how many people see the problem and how much damage it does.

## Two halves, and why you cannot grade yourself
The academic frame is Kevin Keller's. A Utah State University poster abstract cites Dev and Keller (2014) for the definition, "a comprehensive examination of a brand to assess its health, uncover its sources of equity, and suggest ways to improve and expand on that equity", and splits the work in two. The **brand inventory** profiles everything you put out: names, logos, slogans, products, prices, communications. The **brand exploratory** is research into what customers think and feel. We read this through that abstract and lecture slides based on Keller's *Strategic Brand Management*, not the book.

Most small-business guides stop at the inventory. The reason to do both is a measurement problem. A study by Brus, Hartnett, Faulkner and Driesener in the *Journal of Brand Management*, as reported by Marketing Week (we saw the report, not the paper), found that marketers tend to **overestimate how famous their brand assets are and underestimate how many other brands share them**. So you may measure your own files, but you may not score your own recognition.

## Step 1: write the question and the scope
Write one question at the top of the page. "Are we consistent across the website, social and documents?" is a good one. "How is our brand?" is not. Then set the scope, and bring one person who does not work in marketing and has never seen the guidelines. That person is your first outsider.

## Step 2: build the inventory, then measure it
Put every place the brand appears in one folder, as dated screenshots or photos: website (home, a service page, contact), each social profile and its last twelve posts, email signature, a proposal or quote, a deck, an invoice, packaging or signage, and third-party pages such as directories and partner sites. Four things in that folder can be measured.

### Colour drift
Sample the main brand colour from each screenshot with an eyedropper. Do not compare hex codes by eye: two codes that look close can be far apart in perception. Convert to CIELAB and compute the CIEDE2000 distance from your master value. The published thresholds are rough. Wikipedia reports a just noticeable difference of about 2.3 in the older CIELAB formula (attributed to Gaurav Sharma) and a typical printing tolerance of 2.0 under D50 light, up to 5.0 for some processes (attributed to X-Rite). We turn them into three triage bands, which are our judgement and not a standard.

| Touchpoint (invented brand "nova") | Sampled | CIEDE2000 from master | Verdict |
|---|---|---|---|
| Website buttons | #1F4FD8 | 0.0 | The master |
| Email signature | #1E50DA | 0.4 | Same |
| Instagram highlight covers | #2B5BDB | 3.8 | Look side by side |
| Proposal PDF, after a trip through CMYK | #2D4FA8 | 6.3 | Different colour |
| Pitch deck, colour picked by eye | #2E5BFF | 6.6 | Different colour |
| Partner page logo | #0B3AA8 | 8.7 | Different colour |

We computed these values ourselves, and checked the code against two CIEDE2000 test pairs from Sharma, Wu and Dalal's published test data (2.0425 and 27.1492). Under 2 counts as the same, 2 to 5 needs a side by side look on one screen, over 5 is a different colour. Sample everything on one screen with one colour profile, and never from a photo of a printed card, because the lighting changes the result.

![A bar chart of six touchpoints and how far their blue sits from the master colour, from 0.0 for the website to 8.7 for a partner page.](https://lainagency.com/images/blog-art/brand-audit-checklist-fig2.webp)

### Logo proportions
Measure the logo's width and height in every placement and divide. If the master is 4 to 1, a placement at 180 by 50 is 3.6 to 1, which is 10 per cent squeezed, and 240 by 54 is 4.44 to 1, which is 11 per cent stretched. Treat any deviation above 2 per cent as a file error, not a taste choice. The usual cause is a layout cell that forces the width while the height was set by hand. Count the logo versions in circulation too. Extras beyond the master set often come from someone re-exporting a screenshot.

### Contrast of every pair you use
List each text and background pairing and compute the WCAG ratio. WCAG 2.2 asks for at least 4.5 to 1 for normal text, 3 to 1 for large text (18 point, or 14 point bold) and 3 to 1 for the parts of interface components and graphics needed to understand content (criteria 1.4.3 and 1.4.11). Text that is part of a logo has no requirement. Our example blue passes on white at 6.63 to 1, but the same blue on a near black ground (#121212) is 2.83 to 1, which fails the 3 to 1 test for a button outline, and its light tint (#8AA2F0) on white is 2.47 to 1. A colour that works on one ground tells you nothing about the others. For the full rules see our [accessibility checklist](https://lainagency.com/resources/website-accessibility-checklist).

### The name as written
Write the business name exactly as it appears in the website footer, the invoice, each profile and each listing. Google's Business Profile guidelines say the name should reflect "the real-world name, as used consistently on your storefront, website, stationery, and as known to customers", and they bar taglines, location text, phone numbers, all capitals, and legal terms such as LLC unless you can show real-world proof of them. Adding extras "could result in the suspension" of the profile. A name that differs between the listing and the invoice is an audit finding and a risk.

## Step 3: test the message on outsiders
Show the homepage to five people who have never seen it, for five seconds, then hide it and ask two questions: what does this company do, and who is it for? This is the five second test that usability teams use for first impressions. It has no single standard, so treat five people as a smoke test, not a statistic. Code each answer as right, partly right or wrong. Our rule: if fewer than four of five state both what you do and who for, the problem is the message, and no amount of visual tidying fixes it.

## Step 4: test recognition without the logo
Romaniuk and the Ehrenberg-Bass Institute define a distinctive asset by two numbers. **Fame** is the share of category buyers who, shown the asset without the brand name, name your brand. **Uniqueness** is whether only your brand is linked to it. Crop the name and logo out of three of your assets (a post, a slide, a pack) and show them to ten people in your market. Ask "whose is this?" with no list of brands to choose from. In a Romaniuk study published in the *Journal of Advertising Research*, cueing respondents with the brand name inflated fame by up to 20 percentage points, so prompting makes your brand look better than it is.

Sources disagree on which number matters more. One Ehrenberg-Bass summary says uniqueness is the more important. A Marketing Architects article says "fame beats uniqueness". Both are needed. For a small brand fame is close to zero, so the number you can influence is uniqueness. Print four to six competitors' homepages, cover every logo and put yours among them: if you cannot pick yours out, the look is shared with the category. Ten people is directional, not proof.

## Step 5: interview with a sample you can defend
"Five to ten customer interviews" is repeated everywhere. The research behind interview numbers is about **saturation**, the point where new interviews stop producing new themes. Guest, Bunce and Johnson (2006) analysed 60 interviews with women in two West African countries and found 80 of 114 themes (70 per cent) in the first six interviews and 100 (92 per cent) in the first twelve. Hennink, Kaiser and Marconi (2017) analysed 25 interviews with patients at one clinic and found **code saturation** (the range of issues) at nine interviews, but **meaning saturation** (understanding why) took 16 to 24. We read the second paper in full and the first through summaries of its abstract.

Two things follow. Both samples were homogeneous, so the numbers apply to one group at a time, and your audit has at least three: customers who chose you, people you lost and people who have never heard of you. Plan about six per group. And a small sample finds the issues but not the reasons, so ask for reasons.

- How would you describe us to a colleague?
- What were you comparing us with?
- What almost stopped you from choosing us?
- Which three words fit us, and which three do not?

Ask what people did, not whether they like your logo, and write down exact words. Stop a group when the last two interviews add no new reason. Our guide to [UX research for startups](https://lainagency.com/resources/ux-research-for-startups) covers asking without leading.

## Step 6: rank the fixes
Give each finding two scores. **Reach**: 1 if few people see it, 2 if customers see it sometimes, 3 if it is a first contact (website, profile, proposal, listing). **Damage**: 1 if cosmetic, 2 if inconsistent, 3 if misleading or blocking (wrong name or contact, unreadable text). Multiply. This is our scheme, not a standard, but it forces the question spreadsheets skip: how many people meet this?

| Finding from the example | Reach | Damage | Score | When |
|---|---|---|---|---|
| Proposal PDF blue is 6.3 away from the master | 3 | 2 | 6 | This week |
| Email signature logo squeezed by 10 per cent | 3 | 2 | 6 | This week |
| Instagram covers blue 3.8 away | 3 | 1 | 3 | This quarter |
| Partner page logo 8.7 away | 2 | 2 | 4 | This quarter |

Score 6 or more: this week. 3 to 4: this quarter. 1 to 2: at the next refresh. Findings that live in the message or the positioning, not the surface, go to a separate list: that is a strategy question, and possibly a [rebrand](https://lainagency.com/resources/when-to-rebrand).

## The statistic you will see, and what it is
Audit articles often say consistent brand presentation raises revenue by 23 per cent. It traces to Lucidpress's State of Brand Consistency survey of brand managers at more than 200 organisations. One trade report says respondents in 2018 estimated 23 per cent and in 2019 expected 33 per cent; other sources date the 23 to 2016, and a later update is cited at 10 to 20. These are estimates of what respondents expect, not measured revenue, and the figure drifted as it was re-run. We did not find the original report. Do not quote it.

## Where audits go wrong
**Everything reads as clear.** Insiders know the answer, so the message seems obvious. Fix: the outsider tests in steps 3 and 4.

**Recognition looks better than it is.** Cause: prompted questions. Fix: unprompted, no list of brands.

**One sample of mixed people.** Saturation applies to a homogeneous group, so a pool of customers, lost leads and strangers blurs each group's themes. Fix: recruit by group.

**Tidying the symptom.** If one problem appears in three or more places, the cause is missing master files, templates or rules. Fix the cause with [brand guidelines](https://lainagency.com/resources/brand-guidelines-what-to-include), or it returns in a quarter.

**Treating drift as a rebrand.** A colour distance of 6 is a file problem. Keep the rebrand question for the message, the positioning or the name.

## The checklist, with pass and fail
- **Question and scope:** one sentence written down. Fail if you cannot state it.
- **Touchpoints:** website, three social profiles, email signature, a proposal, a deck, an invoice and every third-party listing captured and dated.
- **Colour:** every sample within 2 of the master (CIEDE2000), or logged as an exception with a reason.
- **Logo:** aspect ratio within 2 per cent of the master in every placement, one current file set.
- **Contrast:** every text and background pair at 4.5 to 1, large text and interface parts at 3 to 1.
- **Name:** identical in the website footer, invoice and each listing.
- **Message:** at least four of five outsiders state what you do and who for.
- **Recognition:** three cropped assets shown unprompted to ten people, results written down.
- **Interviews:** about six per group, stopped when new reasons stop.
- **Fix list:** ranked by reach times damage, each fix with an owner and a date.

## When to bring in help
You can run this yourself. A designer helps most when the audit points deeper: a message that does not hold, or an identity that was never a system. A LAIN [branding project](https://lainagency.com/about#logo-brand-identity) typically takes 4 to 6 weeks across four phases, starting with Alignment. To see what one identity looks like when it holds across a website, a deck and social, see [Montreval](https://lainagency.com/works/montreval), where each piece "takes only what it needs from the system", [Ordin](https://lainagency.com/works/ordin) and [Urbanova](https://lainagency.com/works/urbanova). If the audit turns up more than you want to fix alone, [send us the folder](https://lainagency.com/contact).

## Frequently asked questions

### What should a brand audit include?

Keller's framework has two halves: an inventory of everything you publish, and research into what customers think and feel. In practice, capture every touchpoint, measure colour, logo proportions, contrast and the spelling of your name, test your message and recognition on outsiders, and interview customers in groups. End with a ranked fix list.

### How many customers should I interview for a brand audit?

Interview research suggests about six per homogeneous group. Guest, Bunce and Johnson (2006) found 70 per cent of themes in the first six of 60 interviews and 92 per cent in the first twelve. Hennink and colleagues (2017) found the range of issues by nine interviews but needed 16 to 24 to understand why. Treat customers, lost leads and non-customers as separate groups.

### How do I check whether my brand colours are consistent?

Sample the colour from each touchpoint with an eyedropper and compute the CIEDE2000 distance from your master value. Published figures put a just noticeable difference near 2.3 in the older CIELAB formula and a typical print tolerance at 2.0, up to 5.0 for some processes. As triage: under 2 is the same, 2 to 5 needs a side by side look, over 5 is a different colour.

### Does consistent branding raise revenue by 23 per cent?

That figure comes from a Lucidpress survey of brand managers, and it records what respondents estimated or expected, not measured revenue. One trade report places 23 per cent in 2018 and 33 per cent in 2019, other sources date it to 2016, and a later update is cited at 10 to 20 per cent. Do not quote it as a result.

### How do I test whether people recognise my brand without the logo?

Crop the name and logo out of three of your assets and show them to about ten people in your market. Ask whose they are with no list of brands, because Romaniuk found that cueing with the brand name can inflate fame by up to 20 percentage points. Count who names you and who names someone else, which tells you fame and uniqueness.

### What should I fix first after a brand audit?

Score each finding by reach (1 to 3, how many people meet it, with 3 for first contacts such as the website, profiles, proposals and listings) times damage (1 to 3, from cosmetic to misleading or blocking). Fix scores of 6 or more this week, 3 to 4 this quarter. Problems in the message or positioning belong on a separate strategy list.

## Sources and fact check

Checked on 28 Oct 2026.

- [Utah State University: The Retail Brand Audit, Conceptualization and Application (poster abstract, 2015)](https://digitalcommons.usu.edu/researchweek/ResearchWeek2015/All2015/226)
- [Hennink, Kaiser and Marconi: Code saturation versus meaning saturation (PubMed Central copy)](https://pmc.ncbi.nlm.nih.gov/articles/PMC9359070)
- [Guest, Bunce and Johnson: How many interviews are enough? (Field Methods, 2006), abstract record](https://library.alnap.org/node/58582)
- [Wikipedia: Color difference (JND and printing tolerance, citing Sharma and X-Rite)](https://en.wikipedia.org/wiki/Color_difference)
- [Sharma, Wu and Dalal: The CIEDE2000 colour-difference formula, implementation notes and test data](https://hajim.rochester.edu/ece/sites/gsharma/ciede2000/)
- [Marketing Science (Ehrenberg-Bass): Mastercard does some branding](https://marketingscience.info/mastercard-does-some-branding/)
- [Marketing Architects: Your brand assets may be working against you](https://www.marketingarchitects.com/blog/v41-n3-your-brand-assets-may-be-working-against-you)
- [Marketing Week: Marketers judgement of brand asset strength rarely accurate](https://www.marketingweek.com/marketers-judgement-strength-brand-assets-rarely-accurate/)
- [Google Business Profile Help: Guidelines for representing your business on Google](https://support.google.com/business/answer/3038177?hl=en)
- [W3C: Web Content Accessibility Guidelines (WCAG) 2.2](https://www.w3.org/TR/WCAG22/)
- [ASI: Survey, brand consistency can boost sales (on the Lucidpress report)](https://members.asicentral.com/news/newsletters/promogram/december-2019/survey-brand-consistency-can-boost-sales/)