How to Run a Brand Audit Before You Commission a Rebrand

A brand audit is a two-to-four-week diagnostic that scores five areas — visual consistency, verbal tone, market position, digital presence and internal alignment — and ends in one ranked, scored gap list. That list answers the only question worth asking: leave the brand alone, refresh it, or rebuild it. A studio-run audit costs AED 15,000-60,000 in the UAE (roughly SAR 15,000-61,000 in Riyadh, QAR 15,000-60,000 in Doha); a stripped version you run yourself takes about two days.

Run it before you write the rebrand brief, not after. A brief written without an audit starts from a founder's hunch — almost always "our logo looks dated" — and the studio has no way to test whether the logo is what is costing you deals. Usually it isn't. Either the identity is sound and nine versions of it are in circulation, or the identity is sound and nobody in the market can repeat back what you sell.

What follows is the audit GLDS runs before touching any identity work: five areas, about 25 checks, each scored 0-3 and weighted by how often a customer actually sees it.

Audit first, brief second — the order is the whole point

A rebrand brief is a conclusion. An audit is the evidence that tells you which conclusion is correct. In our experience with regional SMEs, roughly two out of three companies that arrive asking for a rebrand need a refresh plus a set of enforceable rules — not a new name, not a new mark. The other third have a real positioning problem, and for them a new logo without new positioning is money spent on the symptom.

The arithmetic makes the case. A full rebrand for a company with 12 retail doors in the UAE runs several times the audit fee before you touch the physical cost: fascia signage per door, vehicle wraps, packaging dies, uniform reprints, app store assets, a bilingual website rebuild. Spending AED 15,000-60,000 and two to four weeks to find out which problem you actually have is the cheapest decision in the entire project, and the only one that is reversible.

The deliverable is not a mood board or a 60-slide "brand health" presentation. It is a one-page scored gap list, ranked by customer visibility, plus a verdict with a number attached to it. If an audit does not end in something you can argue with, you bought a workshop, not a diagnostic.

Area 1 — Visual: count what is in use, not what the guidelines say

Collect before you judge. Pull 40 to 60 live touchpoints into one folder in a single sitting: the website header, the last 12 Instagram posts, the invoice template, the PDF proposal, three staff email signatures, the Google Business Profile cover, mall fascia photographed straight on, van livery, delivery packaging, the WhatsApp Business display picture, uniform badges, the LinkedIn banner, the trade-show pull-up from last year. Not the brand book — the artefacts a customer met last month.

Then count three numbers, because counting removes the argument: distinct logo files in circulation, distinct hex values used for what everyone calls "our blue", and distinct typefaces. A typical first pull for a 5-to-8-year-old Gulf company turns up 4 to 7 logo variants (the original, the 2021 redraw, a stretched PNG someone rebuilt in PowerPoint), 3 or 4 near-identical blues, and 5 to 9 typefaces once Canva picks and the Microsoft default — Calibri or Aptos — are counted honestly.

In the Gulf, add the Arabic check, because it is where most systems break. If nobody commissioned an Arabic lockup, the sign shop chose one: fascia set in Traditional Arabic or a default Naskh at a weight that does not match the Latin, sitting next to Instagram captions in a third face. Score each item 0-3, then weight it: anything a customer sees weekly counts triple, anything annual counts once.

Area 2 — Verbal: read 20 recent pieces out loud

Sample the writing nobody proofreads. Take the last 10 Instagram captions, 3 outbound sales emails, 5 customer-service WhatsApp replies, the About page and the careers page, and read them aloud in one session. The test is blunt: could a stranger tell these came from the same company? Most drift shows up between the site (written once, by an agency) and the channels where the business actually happens (written daily, by whoever is on shift).

The regional version of this check is register. A Dubai or Riyadh company will often publish formal Modern Standard Arabic on the website, switch to Khaleeji dialect on Instagram, and answer WhatsApp in English. None of that is wrong — but it has to be a decision that is written down, not an accident of who is typing. The break to look for is when MSA corporate boilerplate on the site contradicts the dialect voice that is actually closing sales in DMs.

Output one page: three words you are, three words you are not, and the two rules broken most often with the screenshots attached. That page is worth more than a tone-of-voice chapter nobody opens.

Area 3 — Market position: get described back to you, unprompted

Ask 8 to 12 people one open question with no options offered: "If a friend asked what we do, what would you say?" Split them 5 current customers, 3 lost prospects, 2 partners or suppliers. Record verbatim, do not correct them, and count how many use your own positioning words. Fewer than 3 in 10 means the message is not landing, and no amount of new typography fixes that. Lost prospects are the most valuable ten minutes in the audit — they will tell you exactly what they thought you were.

Then set your words beside the competitive set. Take 5 competitors and copy what each says about itself in three places: the meta description, the first screen of the homepage, the LinkedIn About. If three of six companies in your category claim premium quality and personalised service, nobody owns those words and yours are doing no work. Mine the unprompted language in your Google reviews too — customers describing you in their own words is free positioning research most companies never read.

Score the gap between how you describe yourself, how customers describe you, and how the category describes itself. A brand where all three match does not need a rebrand; it needs distribution.

Area 4 — Digital: speed, mobile, search, and whether AI answers know you

Four measurable checks, in this order. Speed and mobile first: test Largest Contentful Paint under 2.5 seconds and Interaction to Next Paint under 200 milliseconds — Google's Core Web Vitals thresholds — on a mid-range Android over mobile data, not on your laptop on office Wi-Fi. In most Gulf consumer categories 75-85% of sessions are mobile, so the mobile number is the only number.

Search next. In Google Search Console, split clicks into branded and non-branded queries. If 80% or more of your clicks come from people typing your company name, you are harvesting demand somebody else created, not creating it. Check Arabic-language queries as a separate line — plenty of UAE and Saudi sites rank in English and are invisible in Arabic for the same terms.

Then test AI answer presence directly, because it is the fastest-moving of the five checks. Run 10 prompts a real buyer would type into ChatGPT, Perplexity and Google AI Overviews — "best branding agency in Dubai", "interior design studio Riyadh", "who designs bilingual packaging in the GCC" — and log two things: whether you are named, and whether the facts about you are right. What feeds those engines is boring and fixable: Organization schema markup, a Google Business Profile whose name, address and phone match the site exactly, and pages that answer a specific question in their first two sentences.

Area 5 — Internal alignment, then the scored gap list and the verdict

Message 6 to 10 staff separately and in writing — sales, ops, front of house, one junior hire from the last 90 days — and give them 30 seconds for three questions: what do we sell, who buys it, why us instead of the obvious alternative. The founder answers separately, without seeing the others. Divergence here explains most of what you found in the other four areas: a team that cannot describe the brand the same way cannot execute it the same way, and no guideline PDF has ever fixed that.

Now score. About 25 checks, each 0-3, weighted by customer visibility (weekly ×3, monthly ×2, rare ×1), sorted by weighted gap. The verdict falls out of the table. Leave it alone and write rules when position is clear and only execution drifted — the fix is a template set and one person who owns approvals. Refresh — typography, colour discipline, a proper Arabic lockup, templates — when the name and position still hold but the system does not scale past the channels it was drawn for. Rebrand, meaning name and position, only when the audit shows the market believes you are something you no longer sell, or the name blocks the next market legally or phonetically.

You can run a light version yourself in two days: the 30-touchpoint folder, the three counts, 6 staff answers, 5 customer sentences, one PageSpeed Insights run on mobile, 10 AI prompts. You will lose objectivity — nobody hears their own tone of voice, and everybody scores their own logo generously — but you will surface the top three gaps and walk into any studio with a brief built on evidence instead of a hunch.