Skip to content
Operations November 7, 2026 9 min read

Arabic Names in a CRM: Spelling and Search

Mohammed, Muhammad or Mohamed? How to store Arabic names in a CRM, keep the Arabic script, search across spellings and avoid duplicate contacts on call lists.

D
November 7, 2026

Quick answer

Store each Arabic name twice: once in Arabic script exactly as given, and once in the Latin spelling the customer uses, in separate fields. Do not force first-name and last-name boxes, because Arabic names have more parts. Normalise diacritics, hamza and alef forms when you search, and match duplicates on phone number or customer ID before name, so one customer is not called twice under different spellings.

A contact centre meets the problem on day one. The same customer appears as three records because three agents typed the name three ways. A call list contains the same household twice. An agent reads a transliterated name aloud and gets it wrong. None of this is a language-skills problem; it is a data-model problem, and it is cheap to prevent if you design the name fields before the list is imported.

This guide covers how Arabic names are built, why one name has many Latin spellings, how to store and search them, and how to keep duplicates off your call lists. The examples are illustrative, and the external facts are sourced at the end.

Why Arabic names break first-name and last-name forms

A form with "First name" and "Last name" assumes two parts and a family name that is shared and inherited. That is not how every naming system works. The W3C article on personal names around the world uses an Arabic example to make the point: Abu Karim Muhammad al-Jamil ibn Nidal ibn Abdulaziz al-Filistini, which it translates as "Father of Karim, Muhammad (given name), The beautiful, Son of Nidal, Son of Abdulaziz, the Palestinian". One name holds a reference to the first-born son, a given name, a descriptive epithet, a chain of "ibn" (son of) links, and a word showing origin ("the Palestinian").

The IFLA guidelines on Arabic personal names, written for cataloguers, list the same family of elements under their Arabic terms: ism, kunya, nasab, laqab and nisba, and add the shuhra, the best-known form of a person's name. You do not need to model all of these in a CRM. What matters is the lesson: there may be more than two parts, and the part people go by is not always the last one.

The W3C's practical advice for form design applies directly:

  • Use one full name field where possible, and split only when you have a specific need.
  • If you do split, avoid the labels "first name" and "last name".
  • Add a separate field for how the person wants to be addressed.
  • Do not require a family name, and allow long names.
  • Accept the full name as the person provides it, including spaces, punctuation such as hyphens and apostrophes, and characters outside the English alphabet.
  • Use Unicode (such as UTF-8) throughout pages, databases and code.

One name, many Latin spellings

The given name written محمد is a good example. Customers spell it in Latin script as Mohammed, Muhammad, Mohamed, Mohammad and other ways, and each is correct for the person who uses it. Their passport, bank card or email address may all carry different spellings of the same name.

There is no single romanization. Schemes exist, and the Library of Congress publishes an Arabic Romanization Table (2012 version), part of the ALA-LC romanization tables approved by the Library of Congress and the American Library Association, which the IFLA guideline lists among the schemes in use for cataloguing. Under that table the letter ح is written as ḥ, so that scheme would give Muḥammad, with a dot under the h. But that is one scheme, designed for cataloguing and not for how people spell their own names in daily life. The IFLA guideline likewise discusses romanization schemes in the plural. The same LC table also shows why different spellings of other names arise. For example, it does not write a hamza at the start of a word, or the alif that carries a hamza, and it writes the final ة as h or t depending on the grammar of the phrase. Two people writing the same name from the Arabic can honestly end up with different Latin forms.

What this means for your CRM:

  • Do not "correct" a customer's Latin spelling to a house style.
  • Do not use the Latin spelling as the key to find a person.
  • Keep the Arabic script as the anchor, because it is the original.
FieldWhat goes in itWhy
Full name (Arabic script)The name as the customer or their document writes itThe original; the best thing to show an Arabic-speaking agent
Full name (Latin spelling)The spelling the customer uses in Latin scriptUsed in email, SMS and English-language systems
Preferred name / how to addressWhat the agent should call them on the phoneA person may go by one part of the name only
Phone (normalised)One agreed format, digits converted to 0-9The most reliable match key
Customer IDYour own identifierSurvives every spelling change

Store the Arabic script and the Latin spelling side by side

The W3C says that if you need both native script and a Latin transcription, you should use separate fields for each. In a contact centre that gives you three benefits at once. The agent who reads Arabic sees the Arabic. The SMS or email template that goes out in English uses the Latin spelling the customer chose. And the list import has two ways to land a match.

Practical rules for the import file:

  1. Make sure the file is saved as UTF-8. A spreadsheet saved in an encoding that has no Arabic characters can turn Arabic into question marks, and that damage cannot be undone from the saved file.
  2. Give the Arabic and Latin names their own columns. Do not squeeze both into one cell.
  3. Do not require the Latin column. Many customers will have only the Arabic form on record.
  4. Keep the customer ID and the phone number as separate, mandatory-where-possible columns.
  5. Spot-check twenty rows after the import by opening them on screen, in the language your agents will use.

Search: diacritics, hamza forms, ta marbuta and alef variants

A search for an Arabic name can fail even when the contact is on file, because the typed text and the stored text differ in ways a reader ignores. The Unicode Arabic chart lists each of these as a separate character:

  • Diacritics. Marks such as fathatan (U+064B) and shadda (U+0651) are characters of their own. A name stored with marks and searched without them will not match unless search ignores the marks.
  • Hamza forms. Alef with hamza above (U+0623) and the plain alef are different characters, and yeh with hamza above is another (U+0626). Many people type the plain letter and leave out the hamza, and many records are stored the other way.
  • Ta marbuta. Teh marbuta (U+0629) at the end of a word looks close to the letter ha, and is often typed as ha in quick entry.
  • Elongation. Tatweel (U+0640) stretches a word on screen but is a character in the string.

A sensible search strategy, which you should test with your own data, is to normalise both the stored value and the query before comparing: remove diacritics and tatweel, treat the hamza forms of alef as plain alef, and optionally treat final ta marbuta and ha as equal. Keep the original stored text unchanged; only the comparison key is normalised. This also helps with the Latin side: store a lower-cased comparison key and let the search ignore case and spacing.

Be careful with over-normalising. If you treat every vowel-like letter as equal, unrelated names will collide. Start conservative, review what the search returns for real names on your list, and loosen only where agents tell you a contact was missed.

Duplicates: match on phone number and customer ID before name

Because the same name has many spellings and different people share the same name, a name-only duplicate check is weak in both directions. It will miss "Mohammed" and "Muhammad" at the same number, and it will wrongly merge two different customers who share a common name. Use this order:

  1. Customer ID. If both records carry the same ID from your own system, they are the same customer.
  2. Normalised phone number. Convert every number to one format and every digit to 0-9 before comparing. Two records at the same number are candidates, not automatic duplicates, because a family can share a landline or an office line.
  3. Name, as a supporting signal. Compare the Arabic script after normalisation. Use the Latin spelling only as a weaker hint.
  4. Hold near-matches for review. Do not auto-merge when only the name matches.

For a call list the cost of a missed duplicate is concrete: the same customer receives two calls, or one customer you have already spoken to is dialled again by a different campaign. Check for duplicates on import, and again before a large campaign, as well as in the CRM.

What the agent should say: confirming a name without guessing

Agents mispronounce names when the screen shows only a Latin transliteration and they must guess the Arabic. Showing the Arabic script beside it removes most of the guesswork. For the opening of a call, confirm the name the way the customer would expect, and let them correct you.

السلام عليكم، هل أتحدث مع [الاسم]؟

English note: "Peace be upon you, am I speaking with [name]?" The name placeholder is filled from the Arabic-script field.

عفوا، كيف تفضلون أن أخاطبكم؟

English note: "Excuse me, how would you like me to address you? (polite plural form)" Use this when the record has no preferred-name entry, then save the answer in the preferred-name field so the next agent does not have to ask. The phrases above are generic examples; the longer phrase bank in our guide to Arabic customer service phrases covers more of the call.

If the customer corrects the spelling or pronunciation, update the record during the call or in wrap-up, and never overwrite the Arabic script with an agent's guess.

Arabic-Indic digits in IDs and phone fields

The Unicode Arabic block has two sets of digits besides the ones used in English. Arabic-Indic digits (U+0660 to U+0669) are used with Arabic proper, and Eastern Arabic-Indic digits (U+06F0 to U+06F9) are used in the Arabic-script languages of Iran, Pakistan and India, such as Persian and Urdu. They are separate characters from 0 to 9, so a number typed with them does not equal the same number typed with 0 to 9 unless something converts it.

That affects you in three places:

  • Phone fields. An agent or a customer form may produce Arabic-Indic digits. Convert on input, before the duplicate check and before dialling.
  • ID numbers. National or account IDs pasted from an Arabic document may use the other digit set. Normalise before you compare.
  • Imports. A spreadsheet from a local team may mix digit sets in one column. Check a sample before loading.

For how digits are spoken by an AI voice agent, see our guide to Arabic text-to-speech for numbers and dates. For recognising names and numbers said by callers, see Arabic speech recognition across dialects.

How DialerBee fits

The Arabic RTL page describes an Arabic agent desktop with a right-to-left layout, where each agent picks an interface language independently, so Arabic and English agents can work the same campaign. It states that full-text search across contacts, notes and dispositions handles Arabic diacritics, hamza variations and tashkeel marks, returning results whether or not the query or the stored data includes the marks. Bidirectional text in scripts is handled, so a name in Arabic inside an English script line renders in the right direction.

Contact records, custom fields, imports with a column-mapping preview and do-not-call scrubbing on import are covered on the lead management page. The field set for names is yours to define there. The agent desktop shows the record and the script beside the call, so the agent sees the customer's own name on screen. If your CRM is elsewhere, see the CRM integration setup guide and the integrations page. Whatever system you use, test with your own data: search a real customer under three spellings and see whether the same record comes back.

Frequently asked questions

How should I store an Arabic name in a CRM?

Store the name in Arabic script exactly as the customer or the document gives it, and store the Latin spelling the customer actually uses in a separate field. The W3C advises that where you need both native script and a Latin transcription you should use separate fields for each. Add a free-text field for how the person wants to be addressed, and allow long names.

Is there one correct Latin spelling of an Arabic name?

No. Romanization schemes exist, such as the Library of Congress Arabic Romanization Table, but they are one scheme among several, and people spell their own names in Latin the way their passport, family or bank has always done. Treat the customer's own Latin spelling as the correct one for that person, and do not silently rewrite it.

Should I split an Arabic name into first name and last name?

Only if you have a specific need. Arabic names can carry a given name, a descriptive epithet, a chain of father and grandfather names and a nisba, and the W3C suggests a single full name field where possible, with labels that avoid the words first name and last name if you do split. A separate preferred-name field handles how the agent should say it.

How do I stop the same customer being called twice under different spellings?

Match on a stable key first, such as the normalised phone number or your own customer ID, and use the name only as a supporting signal. Names vary by spelling, diacritics and hamza forms, so a name-only match produces both false duplicates and missed ones. Review near-matches by hand before merging.

Why does searching for an Arabic name sometimes miss a contact?

Because the stored text and the search text differ in ways a person ignores: a diacritic mark, an alef with or without hamza, a final ta marbuta or ha, or an elongation mark. Unicode treats each of these as a separate character. Good search normalises them on both sides before comparing.

Are Arabic-Indic digits a problem in phone and ID fields?

They can be. Unicode has Arabic-Indic digits (U+0660 to U+0669) and Eastern Arabic-Indic digits (U+06F0 to U+06F9), which are different characters from 0 to 9. If an agent or a customer types them, a naive match against a stored 0 to 9 number fails. Convert digits to one form on input and keep the original if you need it.

Sources

  1. W3C Internationalization, Personal names around the world (Arabic name example, form-design advice, separate fields for native script and Latin transcription, use of Unicode). w3.org/International/questions/qa-personal-names (checked November 7, 2026)
  2. IFLA Cataloguing Section, Guidelines for Arabic personal names (2018; ism, kunya, nasab, laqab, nisba, shuhra, romanization schemes in use including the ALA-LC tables). repository.ifla.org/handle/20.500.14598/3933 (checked November 7, 2026)
  3. Library of Congress, ALA-LC Arabic Romanization Table (2012 version): ḥ for the letter ḥāʼ, rule 2 (alif supporting hamza not represented), rule 7 (ta marbuta as h or t) and rule 8(a) (initial hamza not represented). loc.gov/catdir/cpso/romanization/arabic.pdf (checked November 7, 2026)
  4. The Unicode Standard, Arabic block chart (U+0600 to U+06FF): hamza forms, teh marbuta, tatweel, diacritics, Arabic-Indic and Eastern Arabic-Indic digits. unicode.org/charts/PDF/U0600.pdf (checked November 7, 2026)

The search-normalisation steps and the match order in this guide are general data-handling suggestions, not part of any standard. Test them on your own data.

Related reading

Ready to see DialerBee in action?

Request a 15-minute live demo, or start a 14-day free trial — no slides, no commitment. Calling starts once your own SIP trunk or PBX is connected and approved.

14-day free trial · free AI credit · no card · 11 languages · BYOC · compliance-supporting controls