Two different people are leaving, and they leave for different reasons
Two completely different people are leaving, and mixing them up is why nothing you try works. The first sees your listing on Talabat or Deliveroo, reads it for four seconds, and scrolls to the next place. The second is already on your own ordering page with items in the cart and quits at payment. Open both numbers before you change anything: on the aggregator, menu views divided by orders; on your own site, sessions divided by completed orders. Most Kuwait restaurants land at 8–15% on the first and 1.5–4% on the second. If either is below that, you have a fixable leak, not a demand problem.
Pull seven days, not one. Friday and Saturday nights behave nothing like a Tuesday at 3pm, and one bad delivery hour can drag a whole week's average down. Then split by device: over 85% of Kuwait orders come from a phone, so testing on your laptop tells you nothing real. Finally, look at where people stop — the menu, the cart, the address screen, or the payment screen. Four numbers, one afternoon of work. Every fix below is worthless until you know which of those four screens is bleeding, because the aggregator fix and the checkout fix have nothing in common.
Ready to start your Shopify store?
Start a free trial and try the platform for yourself.
On the aggregator, four things decide whether they tap you
On the aggregator you are a tile, not a brand. Four things decide the tap: the cover photo, the rating, the delivery time, and the minimum order. Shoot your ten best sellers on a plain background in daylight — a 40–80 KD session with a local photographer beats a 500 KD brand shoot for this specific job. Anything under 4.3 stars gets skipped in a crowded Salmiya or Hawally list, so ask your happy dine-in customers for reviews deliberately, not by accident. And check your minimum order: asking 5 KD when the neighbours ask 3 KD costs you more orders than any ad campaign will bring back.
Delivery time is the one you can move this week. Kuwait diners filter hard at 45 minutes, and every ten minutes above that cuts your listing taps noticeably. Look at your prep time, not the driver — most of the delay sits in the kitchen accepting the order late. Fix item names and descriptions too: Chicken Machboos with no photo and no gram weight loses to a competitor who photographed his. But be honest about the trade. Talabat and Deliveroo take roughly 25–30% of every order, so each improvement you make here earns money for them as well. That is what the next section is about.
On your own page, KNET and guest checkout are the whole game
Your own ordering page is where the margin lives, and four things kill it. First, speed: if the menu takes more than three seconds on 4G in Jahra, a third of people are gone before they see a price. Second, KNET. A card-only checkout in Kuwait throws away more than half your buyers — KNET is what people actually pay with, Apple Pay is second, and cash on delivery is still worth keeping. Third, a delivery fee revealed only at the last step. Fourth, forcing an account before checkout. Guest checkout on its own usually lifts completed orders by 10–25% on a restaurant menu.
If your menu today is a PDF, a WhatsApp number, or a builder that cannot do KNET, rebuild it instead of patching it. A restaurant menu on your own store with a proper KNET checkout costs 15–30 KD a month plus a payment gateway, against roughly a quarter of every order on an aggregator. On 300 orders a month at 6 KD each, that gap is 400–500 KD you keep. Do not delete your Talabat listing — it is real discovery and it still feeds you. Use it to acquire, put a card in every bag giving 15% off direct orders, and move your repeat customers across.
Measure the fix, not the feeling. Track completed orders divided by sessions weekly, plus average order value and how far people get before they quit. A Kuwait restaurant page that loads fast, takes KNET, shows the delivery fee on the first screen and allows guest checkout usually sits at 4–8% instead of 1.5–3%. Give each change two weeks on its own before judging it, and never run three changes in the same week. Expect the whole rebuild to take 3–6 weeks and 400–900 KD if you hire help, and less if your menu is small and your photos already exist.