Hebrew English website development works best when right-to-left support is designed in from the first layout, not patched on later. Use logical CSS properties instead of left and right, set language and direction per page, plan for mixed Hebrew and English text, and give each language its own URLs and metadata.
One site, two directions: the core problem
Hebrew reads right to left, English left to right, and both often appear on the same screen: a Hebrew sentence with an English product name, a phone number, or a price. Browsers handle this with the Unicode bidirectional algorithm, which is good but not magic. Most RTL bugs come from design and code that quietly assume left-to-right.
Start by deciding the structure. There are two sensible models:
- One codebase, two language versions that share templates and components, with direction switched by a language attribute. This is the usual choice for business sites and web apps.
- Separate sites per language, which only makes sense when the content, audience and teams are genuinely different.
With one codebase, set lang and dir on the html element for every page: lang='he' dir='rtl' for Hebrew and lang='en' dir='ltr' for English. Everything else follows from this.
CSS logical properties instead of left and right
The most valuable habit is to stop writing margin-left, padding-right and text-align: left. Logical properties describe layout relative to the reading direction, so a single stylesheet serves both languages.
| Physical (avoid) | Logical (prefer) |
|---|---|
margin-left | margin-inline-start |
padding-right | padding-inline-end |
text-align: left | text-align: start |
left: 0 | inset-inline-start: 0 |
border-left | border-inline-start |
Flexbox and grid already follow the writing direction: a row of items reverses automatically under dir='rtl'. That is helpful, but it also means you should not hard-code reversed order in your own markup.
Some things do need explicit handling. Icons that imply direction, such as back arrows, chevrons and progress indicators, usually need to be mirrored in RTL. Icons that represent real objects, such as a phone, a clock or a play button, usually should not. Decide this per icon, and keep a short list in your design system.
Finally, test sliders, carousels, transitions and any JavaScript that calculates positions or scroll offsets. Scroll position and swipe direction are common places for hidden left-to-right assumptions.
Mixed Hebrew and English text and numbers
This is where pages break in subtle ways. Typical symptoms: punctuation jumping to the wrong end of a line, a brand name at the end of a sentence pulling a full stop to its left, or a phone number displayed backwards.
Practical rules:
- Numbers, phone numbers, email addresses, URLs and product codes are left-to-right content. Wrap them so they keep their order, for example with the
bdielement ordir='ltr'on that fragment. - For user-generated text where you do not know the direction, such as names in a list or comments, use
dir='auto'or thebdielement so each item decides its own direction. - Do not build sentences by concatenating translated fragments. Word order differs between languages, so use full translated strings with placeholders.
- Keep prices consistent. Decide where the currency symbol goes in each language and format numbers with the browser's internationalisation functions rather than by hand.
- Dates deserve the same care. Use locale-aware formatting and be explicit about which calendar your audience expects.
A quick way to find problems is to paste real mixed content into your components, not placeholder text. Dummy English filler text hides every bidirectional bug.
Fonts and typography
Check Hebrew support before you fall in love with a typeface. Many display fonts contain no Hebrew glyphs, and the browser will quietly substitute a system font that clashes with your design. Pick a font family that supports both Hebrew and Latin from the same designer, or pair two fonts deliberately and compare them at the same sizes.
- Several well-known open-source families, for example Heebo, Assistant, Rubik and Noto Sans Hebrew, include Hebrew and Latin. Confirm the licence of any font you use, and consider self-hosting for speed and privacy.
- Hebrew letters often look smaller than Latin ones at the same font size, so adjust size and line height per language rather than assuming one value fits both.
- Avoid italics for Hebrew; use weight or colour for emphasis instead.
- Limit the number of weights you load, since each one adds to page size.
Remember that the optimal line length and spacing differ between scripts. Review your design in Hebrew at real content length, not only in the English comp.
Forms, validation and email templates
Forms are the place where bilingual sites most often disappoint real users.
- Input direction: email, URL and phone fields should accept left-to-right input even on a Hebrew page. Names and addresses should follow the user's typing.
- Validation: do not reject Hebrew letters in name or address fields, and make sure your database and application use UTF-8 everywhere.
- Error messages and labels should be translated and placed so they read naturally in the right direction.
- Email templates need their own RTL testing, because email clients vary in how they treat direction. Set direction explicitly and test the main clients your customers use.
- Sorting and search: sort Hebrew names with locale-aware collation rather than raw character codes.
If your forms feed into a web application, the related article on RTL mistakes in Hebrew web application interfaces covers the app-side issues in more depth.
URLs, hreflang and metadata per language
For search engines, treat each language as its own set of pages. A typical, clean setup uses a path per language, such as /he/ and /en/, though subdomains also work. Avoid showing different languages on the same URL based on cookies or location, because search engines cannot reliably index it.
Each page should include:
- A translated title, meta description and headings, not just a translated body.
- A self-referencing canonical URL for its own language version.
hreflangannotations that list every language version of that page, usinghefor Hebrew andenfor English, plus an optional default fallback. The annotations must be reciprocal: each version points to all the others.- An XML sitemap that includes both language sets, and a visible language switcher that links to the equivalent page, not just the home page.
Translate URL slugs only if you can maintain them. Latin-script slugs on a Hebrew page are perfectly workable and simpler to manage. Check the official documentation of search engines for current hreflang guidance, since recommendations evolve.
Hebrew English website development: the testing checklist
Before launch, test in real browsers with real content:
- Every template, in both languages, at mobile, tablet and desktop widths.
- Hebrew pages with long strings, short strings, and mixed English and numbers.
- Navigation, menus, dropdowns, modals and tooltips: do they open on the correct side?
- Forms: submit with Hebrew, English and mixed input; check validation messages, emails received and stored data.
- Directional icons, progress bars, sliders and swipe gestures.
- Printing and PDF output, if you offer invoices or documents.
- The language switcher, hreflang tags, canonical tags and sitemap.
- Accessibility basics: correct
langattributes so screen readers pronounce each language properly, visible focus order, and sufficient contrast. - Performance with the Hebrew fonts loaded.
Ask a native Hebrew reader to review the pages. Many layout and wording issues are obvious to them within minutes and invisible to a developer.
Putting it together
RTL is cheapest when it is a design rule from the start: logical CSS, real mixed content in mockups, a font with both scripts, a clear URL structure per language, and a native-speaker review before release. If you are planning a bilingual build and want to talk through scope, see the software development in Israel page, or read how multilingual SEO is set up for another language pair to compare the approach.
Frequently asked questions
Should Hebrew and English be separate sites or one?
For most businesses, one codebase with two language versions is easier to maintain. Share templates and components, switch direction through the lang and dir attributes, and give each language its own URLs. Separate sites only make sense when the content, audience or teams are truly different between languages.
How do I handle hreflang for Hebrew?
Use the language code he for Hebrew and en for English on each page, with every version listing all alternates, including itself. Annotations must be reciprocal. Add an optional default fallback, keep canonical tags pointing to each page's own language version, and check current search engine documentation for guidance.
Which fonts work well for Hebrew on the web?
Choose a family that includes both Hebrew and Latin glyphs so the two scripts look consistent. Open-source families such as Heebo, Assistant, Rubik and Noto Sans Hebrew are common choices. Check licences, limit the weights you load, consider self-hosting, and review sizes in Hebrew because letters can look smaller than Latin ones.