Most "bilingual" brands in the Gulf are monolingual brands wearing a second language as a costume. The English version gets the love: the kerning, the grid, the witty headline. Then someone runs it through a translator, flips the layout, drops in whatever Arabic font ships with the CMS, and ships it. The result reads like a tourist menu. Technically correct. Quietly disrespectful.
Designing bilingual properly is not a translation problem. It is a systems problem. You are building one brand that has to feel native when read right-to-left in Arabic and left-to-right in English, with neither version feeling like the afterthought. That takes intention at every layer: type, layout, tone, and the invisible plumbing underneath.
Mirroring is not flipping
The default instinct is to take the English layout and mirror it for Arabic. CSS gives you direction: rtl and logical properties, and that handles most of the structural mirroring for free: the nav moves to the right, padding and margins follow reading order, the visual flow reverses. Good. Do that.
But mirroring is where craft begins, not ends. Some things must flip: navigation, progress bars, the direction an arrow points to mean "next." Some things must never flip: logos, phone numbers, brand marks, anything with a Latin or numeric identity baked in. Photography needs judgment, a portrait facing into the page in one direction now faces off the edge in the other. Charts that read as a timeline left-to-right become nonsense reversed unless you rethink them.
The discipline is to treat each element as a decision, not a transform. Build your design tokens with logical properties from the start, margin-inline-start instead of margin-left, so the system carries direction automatically. Then audit the exceptions by hand. A bilingual layout that has never been opened in RTL by an Arabic reader is a layout you have not finished.
Type is where most brands give up
Here is the part that separates the serious from the lazy. Latin and Arabic type have different anatomies. Arabic has no uppercase, connects its letters, sits on a baseline that behaves differently, and carries far more visual weight per character at the same point size. Set Arabic at the same size as your English body copy and it will look cramped and quiet. You almost always need to bump Arabic up a step and loosen the line height.
Pairing matters more than picking. The goal is a Latin face and an Arabic face that share a temperament, the same sense of contrast, the same warmth or geometry, without pretending to be the same typeface. Some families now ship matched Arabic and Latin cuts designed together, and they save you enormous grief. When you pair separately, judge them side by side at real sizes in real sentences, not in a specimen. A geometric sans in English should not sit next to a calligraphic Naskh that thinks it is at a wedding.
Three rules that survive contact with real projects:
- Set type in context, never in isolation. A headline that sings alone can fight its translation. Design the Arabic and English versions of the same component on the same artboard.
- Respect the numerals decision. Eastern Arabic numerals or Western digits is a brand choice with real consequences for forms, prices, and phone numbers. Decide once, apply everywhere, and make sure your fonts actually contain the glyphs.
- Test the awkward strings. A name, a long compound word, a price with a currency. The places things break are never the hero headline.
Tone is a translation of intent, not words
Arabic and English carry register differently. English Gulf marketing tolerates, even rewards, a clipped, casual, slightly cheeky voice. The same cheek rendered literally into Arabic can land as flippant or off, because Modern Standard Arabic carries a formality the English never implied, and dialect carries intimacy the English never asked for. A clever English pun usually has no Arabic twin. Forcing one produces a sentence that is correct and dead.
So you do not translate the line. You translate the intent and rewrite the line. This is why machine translation and "we'll have someone in the office do it" both fail, they optimize for accuracy when the job is resonance. Brief your Arabic copywriter the way you briefed the English one: here is the feeling, here is the audience, here is what we will never say. Let them write Arabic that a Bahraini or Saudi reader would believe came from a person, not a plugin.
If a native reader can tell which language you designed first, you have not built a bilingual brand. You have built one brand and its understudy.
Build the system, not two websites
The practical answer is to design tokens, components, and rules that hold both languages as first-class citizens. One set of spacing logic that respects reading direction. Type styles that already account for the Arabic size bump. Components tested in both directions before they ship. A content model where every string has both versions, with neither marked "default." When the system is right, adding a campaign is a day of work, not a renegotiation of the whole brand.
Bilingual design done well is invisible. Nobody notices the Arabic looks effortless; they just trust it. That trust is the whole prize, and you only earn it by refusing to treat one language as the original and the other as a copy. Design both directions like they are home. Because for half your audience, one of them is.
