Most ecommerce projects in this region do not go wrong at the design stage. They go wrong three weeks before launch, when somebody finally asks how returns work, or whether the Arabic site needs its own product descriptions, and the answer changes the design.

These are the decisions that cost the most when they are made late. Settle them before anyone opens Figma.

Who is actually buying

"UAE customers" is not a segment. A resident buying weekly groceries, an expat furnishing a flat they will leave in two years, and a GCC customer shipping to Riyadh behave nothing alike. They want different delivery promises, they abandon at different points, and they need different things above the fold.

Pick the one you are designing for first. You can serve the others, but a site designed for everyone converts for no one.

Cash on delivery

Cash on delivery remains common here in a way it is not in the UK or US, and it changes more of your design than you would expect.

• The checkout is not the end. A COD order is a promise, not a payment, and some of them will be refused at the door. Your confirmation flow, your stock handling and your returns policy all have to account for that.
• Fraud and abuse controls become design problems. Order limits, phone verification and address confirmation all appear in the interface, and if they are bolted on late they feel like accusations rather than reassurance.
• Your reporting changes. Conversion rate means something different when a proportion of conversions are later declined.

Decide early whether you offer it. Adding it afterwards touches checkout, order management, fulfilment and the returns journey at once.

Delivery promises

Delivery is the most-read part of most product pages in this market, and the most commonly fudged. Same-day within Dubai, next-day to the northern emirates and three days to Saudi are three different promises, and the page has to say which applies before the customer reaches checkout, not after.

If you cannot commit to a time, say what you can commit to. Vague delivery copy is the single fastest way to lose a basket that was otherwise ready.

Arabic, decided properly

Bilingual is a decision with three levels, and they cost very differently:

• Interface only. Navigation, buttons and checkout in Arabic, product content in English. Cheapest, and obvious to any Arabic-speaking customer.
• Interface and product content. Every product described twice. The design work is similar, the content operation is a permanent commitment.
• Two genuine storefronts. Different merchandising, different campaigns, sometimes different catalogues.

There is no right answer, but there is a wrong one: deciding after the templates are built. Right-to-left changes layout, line length, numerals and form behaviour, and retrofitting it costs considerably more than designing for it from the first layout.

Your product data is the design

This is the part clients least expect. A product grid is only as good as the data behind it. If half your products have no size attribute, filtering by size is useless. If descriptions vary from forty words to four hundred, no card layout will look consistent.

Before designing the catalogue, audit the data. Count how many products have complete attributes, consistent imagery and usable descriptions. Design for the data you have, then improve the data — not the other way round.

Platform: the honest version

Shopify handles most of this market well, and the cases for something custom are narrower than agencies suggest. You genuinely need custom when you have complex B2B pricing, an ERP that must be the source of truth, or a catalogue structure the platform cannot express.

You do not need custom because you want a distinctive design. Shopify will carry almost any design you can draw. Choosing custom for aesthetic reasons buys you a maintenance burden you will still be paying for in three years.

What we would settle first

If you take one thing from this: write down the answers to these five before design starts.

• Which customer the site is designed for, in one sentence
• Whether you accept cash on delivery
• What delivery promise you can make, by destination
• Which level of Arabic you are committing to
• What proportion of your catalogue has complete, consistent data

We have designed retail and commerce experiences for Carrefour and Rivoli, and in both cases the decisions above shaped the design more than any stylistic preference did.

If you want a view on your own situation, or a realistic range for the work, the questions we get asked most covers how we scope. There is also a companion piece on what a website costs in Dubai and why quotes vary so much.