Layla Converts 2.5x More: What a Gulf-Born AI Coach Taught Us About Arabic Retail

In October 2025, Chalhoub Group launched Layla — a generative AI beauty coach built into the faces.com platform. In the months that followed, the performance data came in: 2.5 times higher conversion compared to standard browsing, seven additional minutes per session on average, and a 27% increase in products viewed per visit.

But the number that carries more weight than any of those: 80% of Layla's conversations took place in Arabic — including Arabizi. And 85% of Saudi customers on the platform have engaged with the tool.

Why That Number Outweighs the Conversion Rate

Most AI-driven retail experiences in the region start with a product built in English and layer on translation. The result is an AI that speaks Arabic in clinical, neutral tones — competent, but distant from the way a trusted advisor actually communicates. It answers questions without building rapport. It recommends without understanding preference. Conversion stays low because trust was never built.

Layla was built differently. Chalhoub understood that a Gulf consumer — particularly in beauty, a category built on personal trust and tactile experience — doesn't want an AI that describes products. She wants something closer to the knowledgeable friend at the counter: familiar language, culturally fluent expression, warm in register. That difference in design produces a measurably different outcome.

What Seven Additional Minutes Actually Mean

Seven additional minutes per session is not a vanity metric. In e-commerce, session time directly correlates with purchase intent — but only when the time is driven by engagement, not confusion. A user who spends seven more minutes because she's having a helpful conversation is doing something qualitatively different from a user who spends seven more minutes because she can't find the checkout button.

The 27% increase in products viewed per visit tells the same story from a different angle: Layla is successfully expanding what the customer considers relevant to her. This is the AI equivalent of a good sales associate who says 'given what you told me about your skin, you might also want to look at this' — and is believed, because the advice came from understanding, not from an algorithm running 'frequently bought together.'

The Lesson for Saudi Brands Considering AI

What Layla demonstrates isn't that AI improves conversion — that's already understood. It demonstrates something more specific: AI built for the Gulf consumer performs at a categorically different level than AI imported from Western markets with a translation layer applied afterward. Cultural adaptation is not an optional enhancement. It is the difference between a 10% lift and a 250% lift.

This raises a direct question for any Saudi brand considering interactive AI — whether a chatbot, recommendation engine, or purchase assistant: is the AI you're deploying built for a Saudi user, or are you importing an American experience and hoping it translates? A customer service bot built in authentic Arabic — not translated Arabic — changes how the consumer reads the brand before they've even evaluated the product.

The brand that speaks the customer's real language doesn't need to persuade them — the conversation converts itself.

The numbers Chalhoub published are not just proof that one platform works. They are a reference point for the entire market: when AI is built with the user's culture rather than around it, results don't improve incrementally — they jump. And that's what separates a genuine implementation from an expensive experiment.