
An AI-personalised shopping app shows each customer the products, search results and offers that match what they search, browse and buy. For a UAE store with repeat customers, the sequence matters: start with search and recommendations, which run on data you already have. Running costs then grow with your traffic.
Key takeaways
Start with the data you already have. Search runs on your product descriptions from day one; recommendations improve as orders build up.
Some recommendation services need a minimum. Amazon Personalize needs at least 1,000 recorded interactions, such as product views or purchases, from 25 customers before it can make suggestions.
Test search in Arabic and English. Use the searches your customers really type, typos included, before you choose a tool.
Running costs follow traffic. Services a developer connects charge by use, mostly per 1,000 requests, and some bill a minimum around the clock.
On Shopify, Salla or Zid, start in the platform's app store. A custom app fits when one record of each customer must work across your app, website and shops.
Planning a shopping, marketplace or loyalty app? See how we build e-commerce and retail apps for UAE brands.

Four store situations and the AI feature each one starts with, from Arabic and English search to a WhatsApp assistant.
What does an AI-personalised shopping app do?
It changes what each customer sees in your store, based on what they search, browse and buy.
In this guide, your store means your online store and, if you have one, your mobile app. Platform apps are add-ons you install from the Shopify App Store, Salla's app store or Zid's App Market, and a custom app is software built for your store.
Personalisation in an online store usually comes down to five features, each running on its own data:
Feature | What the shopper sees | What it runs on |
|---|---|---|
Search | Results that understand typos, words that mean the same, Arabic and English | Your product titles and descriptions in both languages |
Recommendations | Suggested products on each product page and in the cart | Your order history and product descriptions |
Personal offers | Emails, app notifications or WhatsApp messages with products the customer is likely to want | Each customer's purchase history, and their consent |
Loyalty | Points or rewards shaped by what each customer buys | Each customer's purchase history |
Shopping assistant | A chat that answers questions and suggests products | Your catalogue, policies and stock |
Which AI feature should a UAE store build first?
Build first what your data supports today: search can start now, and recommendations and offers grow with your order history.
Done well, personalisation pays. McKinsey's 2021 analysis found that personalisation most often lifts revenue by 10 to 15 percent, with company-specific results from 5 to 25 percent. The annual e-commerce report from EZDubai and Euromonitor International, reported in September 2026, puts UAE e-commerce at AED 42.2 billion in 2025 and names AI and data-driven personalisation among the trends shaping its next phase.
The right first feature depends on your store's situation:
A new store, or a small catalogue. Search that handles typos and words that mean the same, in Arabic and English, plus suggested products from your platform's own app. With a short order history, your product descriptions do most of the work.
A growing store with repeat buyers. Recommendations on product pages and in the cart, then personal offers and loyalty rewards for customers who have agreed to receive them. Our guide to loyalty apps covers the rewards side.
A large catalogue. Search that puts first the products each shopper is most likely to want, because browsing no longer gets customers to the right product.
Customers who already order on WhatsApp. A shopping assistant: an AI agent, a program that answers customers and carries out tasks such as suggesting products and sending re-order reminders. Our guide to AI agent cost in the UAE sets out its setup and monthly bill.
If two situations apply, start with the one your data supports today. Search runs on your descriptions from day one, while recommendations and offers need order history. Our e-commerce app guide puts search in the first version of a store app, and recommendations, segmented offers and smarter search in the third, built on the purchase data the first versions collect.
How much data do you need before recommendations work?
It depends on the service: some set a minimum of recorded activity, while Shopify's free app can fall back on your collections.
Amazon Personalize sets a clear floor. It is a recommendation service from Amazon Web Services that a developer connects to your app or website, separate from the Amazon marketplace. It needs at least 1,000 recorded interactions, such as a customer viewing or buying a product, from at least 25 customers with two interactions or more each before it can make suggestions, and Amazon recommends 50,000 interactions from 1,000 customers for good-quality suggestions.
Shopify's free recommendations can start before a store has purchase history. Its Search & Discovery app suggests products bought together and, for stores with an English storefront, products with similar descriptions; when neither is available, it suggests products from the same collections. You can also add your own suggestions by hand, and orders imported from another store or platform aren't used, so a store that moves to Shopify starts its purchase history again.
Complete product descriptions help a new store. Until your order history builds up, search and some recommendation tools lean on titles and descriptions. Complete ones in Arabic and English, with the product's name, type, material, colour and use, are preparation you can do yourself.

Search runs on your catalogue from day one; purchase-based recommendations need customer history, and some services set a minimum.
Can AI search handle Arabic and English?
Yes, but tools handle Arabic differently, so test each one with the searches your customers type.
Each tool describes its Arabic support in its own terms. Salla's app store lists AI Finder, a smart-search app built for Arabic-speaking shoppers that corrects typos. Algolia, a search service a developer connects to your store, documents "basic" handling of Arabic plurals.
Five searches to run on each tool, taken from your store's search report:
The same product in Arabic and in English. Both searches should find it.
A common typo. The right product should still come first.
Two local names for one product, such as kandura and dishdasha. Both should find the same products once they're in the tool's synonym list; Shopify's app lets you create these groups yourself.
A singular and a plural. Both should return the same products, in both languages.
A search with no exact match. The page should suggest close products and avoid an empty result.
How do AI personalisation costs grow with the store?
Apps you install from an app store set their own monthly price, and Shopify's own app is free; services a developer connects charge by use, mostly per 1,000 requests, so their bill follows your traffic.
The prices below are as each provider publishes them, checked on 11 November 2026.
Apps you switch on yourself:
Shopify Search & Discovery. Free, built by Shopify, with suggested products on product pages, synonym groups and a report of what customers search for. It works inside a Shopify store.
Apps from Salla's and Zid's app stores. Each app's developer sets its price. Alfinder, for example, lists plans from SAR 249 a month for up to 4,000 products on its page for Zid stores, with a free trial.
Services a developer connects to your store or app:
Algolia search. A search request is one call to Algolia's search, and with search-as-you-type every keystroke counts as one. The Grow plan includes 10,000 requests a month, then charges $0.50 per 1,000; Grow Plus, which adds AI ranking and personalisation, charges $1.75 per 1,000 after the same 10,000.
Amazon Personalize. Prices depend on the kind of model. Its two newest custom models charge $0.15 per 1,000 recommendation requests, a request being one call for suggestions, and its older custom models charge $0.0556 per 1,000 real-time requests for the first 72 million a month, and less above that. For real-time suggestions, each active campaign (Amazon's term for a live recommendation model) is billed at least one request per second, even when no one is browsing. Amazon's own example, a homepage carousel (a scrolling row of suggestions) with 36,000 visits an hour for 10 peak hours a day, comes to $2,016.80 a month on the newest models and $723.35 on the older ones, data storage and training included.
A store with modest traffic can pay for requests it doesn't make. Amazon's minimum runs around the clock, so free allowances and installed apps usually suit a smaller store, and per-request services suit a store busy enough to use what it pays for. These are the providers' published prices; Applify Lab prices each project after a diagnosis of the store.
Two other lines can join the bill. A shopping assistant pays for the AI that writes its replies and, on WhatsApp, Meta's message fees, which our guide to the WhatsApp Business API in the UAE sets out. Building the features into a custom app is a one-time cost that depends on your platform, the systems to connect and the languages, covered in the same e-commerce app guide.
Want to know what your store's data can support?
Tell us your platform, catalogue size and monthly orders, and we'll tell you which AI feature fits your data today.
Shopify, Salla, Zid or a custom app: where does AI fit?
On Shopify, Salla or Zid, start with an app from the platform's app store; a custom app fits when one record of each customer must work across your app, website, shops and systems.
Platform apps run on the store's existing products and orders. Shopify's own Search & Discovery app is free, Salla's app store lists AI Finder for Arabic search and suggestions based on each shopper's behaviour, and Zid's App Market lists an AI-powered search service, Kashaf. A custom app can work alongside your Shopify, Salla or Zid store, using its products and orders, or run on systems of its own.
A custom app makes sense in three cases:
One customer across channels. Your app, website and physical shops share one record of each customer, so suggestions follow the customer.
Your own systems. Stock, loyalty or the tills in your physical shops (point-of-sale
systems) feed the suggestions and offers.
Where the data sits. You choose each service and where it stores customer data; storage locations differ, and Algolia's Grow plan, for example, offers hosting in the US, the UK or the EU. The UAE's Personal Data Protection Law sets separate rules for sending personal data to countries with and without an adequate level of protection (Articles 22 and 23).
Our guide to Salla, Zid, Shopify and custom apps compares the platforms themselves.

Platform apps work inside one store platform; a custom app connects one customer across your app, website, shops and systems.
If you hire a company to build it, what should you ask?
Ask how each feature will use your data, how you'll know it works, and what it will cost to run as traffic grows.
Six questions to ask:
Which feature first, and why? The answer should start from your order history and your customers.
What data does it need, and do we have enough? Ask for the minimum of each service they propose, and how they'll handle the months before you reach it.
How will we measure it? Agree before launch: sales from suggested products should rise, and searches that find nothing should fall.
How does search handle Arabic? Ask to see the results of your own test searches in both languages.
What will it cost each month as traffic doubles? Ask for the unit prices, the minimum charges and the plan you'd move to.
Where is customer data stored, and how is consent collected? Personalisation uses personal data, so settle this before launch (see the FAQ on the UAE's data law).
FAQ
Do I need a custom app to use AI recommendations?
No. On Shopify, the free Search & Discovery app adds suggested products to each product page. Salla's app store lists AI Finder, which recommends products from each shopper's behaviour, and Alfinder offers recommendations for Zid stores. A custom app makes sense when one record of each customer must work across an app, a website and your own systems.
How much data do I need before recommendations work?
Each service sets its own minimum, counted in interactions such as product views or purchases. Amazon Personalize needs at least 1,000 interactions from 25 customers with two or more each, and recommends 50,000 from 1,000 customers for good-quality suggestions. Shopify's free app can suggest products before a store has purchase history, using similar descriptions on English storefronts and products from the same collections.
How do I test a search tool in Arabic before I buy it?
Take ten real searches from your store's search report, in Arabic and English, with typos, plurals and local names. Run them on the tool's free plan or trial: Shopify's Search & Discovery app is free, Algolia's free plan includes 10,000 searches a month, and Alfinder lists a free trial. The right products should come first in both languages.
How much does AI personalisation cost to run?
Apps from a platform's app store charge a monthly price set by their developer, and Shopify's own Search & Discovery app is free. Services a developer connects charge by use: Algolia includes 10,000 search requests a month, then $0.50 per 1,000 on its Grow plan, and Amazon Personalize charges $0.15 per 1,000 recommendation requests on its newest models, or $0.0556 on older ones, with a minimum billed around the clock for real-time suggestions.
Does AI personalisation work with the UAE's data law?
Yes, when it's set up for it. The Personal Data Protection Law treats profiling, using someone's data to predict their preferences or behaviour, as automated processing. It requires consent unless one of the cases it lists applies, such as performing a contract with the customer, and lets customers stop profiling for direct marketing. Collect consent for personal offers and know where each service stores data. This is general information, not legal advice.
To go further
Explore our apps and AI agents for retail and e-commerce.
Discover our custom app and AI development in Dubai.
See how our AI agents for UAE businesses answer customers on WhatsApp.
See how we personalised an AI travel planner that builds the itinerary for you, from each traveller's preferences.
Planning AI features for your store?
In 30 minutes, we'll look at your platform, your catalogue and your orders, and tell you which feature to build first.
RELATED GUIDES

AI Agents & AI Automation
AI Agent Cost in the UAE: Setup and Monthly Prices (2026)
13 min

Retail & E-commerce
Ecommerce App Development Dubai: Cost and Features (2026)
10 min

AI Agents & AI Automation
WhatsApp Business API in the UAE: Cost and Setup (2026)
16 min

Retail & E-commerce
Loyalty Program App Development for Ramadan Ecommerce
7 min

Retail & E-commerce
Fashion Ecommerce App Development in Dubai: UX & Growth Tips
12 min

Retail & E-commerce
Salla vs Zid vs Shopify vs Custom App: A Dubai Guide (2026)
7 min
