Bilingual Agent QA Scorecard for UAE Contact Centres
A bilingual agent QA scorecard is a weighted checklist that lets a UAE contact centre grade both Arabic and English calls on the same criteria — compliance with CBUAE conduct rules, resolution quality, and customer treatment — so quality scores stay comparable across languages, teams and campaigns.
Contact centres in the UAE run on two languages at once. A single collections or servicing team may handle an Emirati customer in Arabic, an expat professional in English, and a small-business owner switching between the two mid-call. Because the population is an expat majority yet Arabic remains the official language, a QA programme that only scores English calls silently leaves a large share of interactions unmonitored. A bilingual scorecard fixes that: the same weighted criteria apply whether the agent speaks Arabic or English, and CBUAE conduct expectations apply either way.
What should a UAE agent QA scorecard measure?
A useful scorecard groups criteria into a few categories, assigns each a weight, and scores every call out of 100. The categories below map to what regulators and customers in the UAE actually care about: regulatory conduct, accuracy, resolution, and communication quality in whichever language the call was held.
The sample weights are a starting template — adjust them to your line of business. A collections desk will weight compliance and fair-treatment more heavily; a sales line will weight disclosure and consent.
Category | What it checks | Sample weight |
|---|---|---|
Compliance & conduct | Correct disclosures, no coercive pressure, fair treatment per CBUAE Consumer Protection Standards | 30 |
Identity & verification | Caller verified before sharing account data; Emirates ID / security questions handled correctly | 15 |
Accuracy | Correct product, fees, dates and next steps stated; no misleading claims | 20 |
Resolution | Customer's issue actually resolved or a clear commitment given | 15 |
Communication | Clarity, tone, active listening, correct language handling and code-switching | 15 |
Documentation | Call summarised and logged accurately for audit | 5 |
Total |
| 100 |
How do you keep Arabic and English scores comparable?
The risk with any bilingual programme is that two evaluators grade to different standards, so an Arabic call and an English call with identical quality end up with different scores. Three habits keep the scale honest:
- One rubric, two languages. Write the scoring guidance once and translate it, rather than letting each language team invent its own criteria. A "fair treatment" failure means the same thing in Arabic and English.
- Native-fluency evaluators or verified transcripts. Whoever scores an Arabic call must understand Arabic natively, or work from an accurate Arabic transcript. Never score a language you cannot follow.
- Calibration sessions. Have Arabic and English evaluators score the same set of calls periodically and reconcile differences, so the two scales stay aligned over time.
Which compliance criteria are non-negotiable in the UAE?
Some scorecard lines are not stylistic preferences — they reflect CBUAE conduct rules that apply to licensed financial institutions and their agents. The CBUAE Consumer Protection Regulation and its Standards require fair treatment of consumers, clear disclosure, and prohibit coercive or abusive collection pressure. Your scorecard should treat a breach of these as an automatic fail regardless of how the rest of the call went.
Auto-fail trigger | Why it matters |
|---|---|
Coercive, threatening or abusive language | Prohibited under CBUAE consumer-protection conduct rules |
Sharing account data before verifying the caller | Data-protection and fraud risk (UAE PDPL, Federal Decree-Law 45/2021) |
Misstating fees, profit rates or repayment terms | Breach of disclosure duties |
Refusing a lawful partial payment on a debt | Runs against fair-treatment expectations and cheque-law changes (Federal Decree-Law 14/2020) |
A single auto-fail sets the call score to zero and routes it for coaching — this keeps agents focused on the lines that carry regulatory and reputational risk.
How often should you sample and score calls?
Manual QA rarely reviews more than a small fraction of calls, so the sample must be deliberate rather than random-only. A practical mix for a UAE contact centre:
- A random baseline sample across every agent, in both Arabic and English.
- A targeted sample of high-risk call types — collections, complaints, and any interaction touching a security cheque, salary transfer letter or account closure.
- A review of any call that generated a customer complaint or an escalation.
Balancing the sample across languages matters: if most monitored calls end up being English simply because English transcripts are easier to read, your Arabic-speaking agents are effectively unsupervised.
How AI helps
Because human QA can only listen to a slice of calls, the coverage gap is where problems hide. YuCI applies call intelligence to every conversation — Arabic and English — transcribing, scoring against your rubric, and flagging compliance risks such as coercive language or missing disclosures for a human reviewer to confirm. The concrete outcome is bilingual QA coverage that no longer depends on which language an evaluator happens to read, so an Arabic call gets the same scrutiny as an English one. This also aligns with the CBUAE guidance on the use of AI and machine learning by licensed financial institutions (Feb 2026), which expects human oversight and explainability rather than fully automated decisions.
Frequently asked questions
What is a bilingual agent QA scorecard? It is a single weighted scoring template used to evaluate contact-centre calls in more than one language — in the UAE, typically Arabic and English — so that quality and compliance are measured consistently no matter which language the agent used.
Why do UAE contact centres need Arabic and English QA? The UAE has an expat majority but Arabic is the official language, so a real customer base is served in both. Scoring only English calls leaves a large share of interactions, and the agents handling them, unmonitored.
Which CBUAE rules should the scorecard reflect? Primarily the Consumer Protection Regulation and its Standards — fair treatment, clear disclosure, and no coercive collection pressure. Data handling should also respect the UAE PDPL (Federal Decree-Law 45/2021).
Can I automate bilingual call scoring? Yes, call-intelligence tools can transcribe and score Arabic and English calls at scale. CBUAE's February 2026 AI guidance expects human oversight, so treat automated scores as a first pass that a QA analyst confirms.
How many calls should we score? There is no fixed figure — combine a random baseline with a targeted sample of high-risk calls and every complaint, and keep the sample balanced across both languages.
What should count as an automatic fail? Conduct breaches: coercive or abusive language, sharing data before verifying the caller, misstating terms, or refusing a lawful partial payment. Any one of these should zero the call regardless of other scores.
Building or refreshing your quality programme for the UAE market? Explore more UAE-specific guides on the YuVerse UAE hub.
References
- CBUAE — Consumer Protection Regulation. https://rulebook.centralbank.ae/en/rulebook/consumer-protection-regulation
- CBUAE — Consumer Protection Standards. https://rulebook.centralbank.ae/en/rulebook/consumer-protection-standards
- CBUAE Rulebook (AI/ML guidance for licensed financial institutions, Feb 2026). https://rulebook.centralbank.ae/
