A WhatsApp catalog is a product list you upload to Meta and connect to your WhatsApp Business account, so a customer browses items, adds them to a cart and sends you an order without leaving the chat. In the UAE the order arrives inside WhatsApp, but the payment does not close there. There is no native in-chat checkout for a Dubai merchant, so the money moves on a hosted page from a licensed gateway after you drop the link into the same thread. Building the full loop of catalog, bot and CRM takes around two to three weeks and starts at AED 6,000 for setup plus AED 1,200 a month to keep it running. A catalog on its own, typed into the free WhatsApp Business app, costs nothing and takes an evening.
What a WhatsApp catalog actually is
Meta stores the catalog, not WhatsApp. You create it in Commerce Manager or directly in the Business app, and it holds items with an image, a title, a price, a description, a link and a retailer ID that you control. That retailer ID matters more than anything else on the item, because it is the only string that ties an order back to a row in your inventory.
Once the catalog is connected to the account, four things can appear in a conversation, according to Meta's catalog documentation:
| Message type | What the customer sees | Typical use in Dubai | |---|---|---| | Catalog message | A thumbnail that opens the whole catalog | "Browse everything" reply to a cold enquiry | | Single product message | One item with image, price, description | Answering "how much is the black one" | | Multi-product message | Up to 30 items grouped into sections | A shortlist after two qualifying questions | | Product carousel | Horizontally scrollable product cards | Same shortlist, visual browsing |
The cart sits underneath all of them. A customer adds items while browsing and sends one order back to you. Meta's documentation is specific about the shape of that cart: one cart per chat thread, up to 99 units of any single item, no cap on how many distinct items go in, and no persistence across devices. If the customer starts a cart on their phone and opens WhatsApp Web, the cart is not there.
The order that lands on your side is a message containing product line items with their retailer IDs, quantities and the prices as they stood in the catalog. It is an intent to buy. It is not a payment, and it is not a booking.
The Business app or the Cloud API
Most Dubai owners start with the free Business app because it takes twenty minutes. It works, up to a point. WhatsApp's own business resources put the app catalog ceiling at 500 items, with up to 10 photos or videos per item. For a boutique with 40 SKUs that ceiling is irrelevant. For a spare parts trader in Al Quoz it is a wall.
The harder limit is that everything in the app is manual, not the item count. Someone has to see the order, price it, check stock, produce the link and type it out.
| | WhatsApp Business app | WhatsApp Business Platform (Cloud API) | |---|---|---| | Catalog items | Up to 500 | Managed in Commerce Manager, feed driven | | Single and multi-product messages | Manual send | Sent by your system, triggered by intent | | Order arrives as | A message a human reads | A webhook payload your code receives | | Inventory updates | Typed by hand | Scheduled data feed, hourly at the fastest | | Payment link | Copied and pasted by a person | Generated per order, sent in seconds | | Order into CRM | Retyped | Written automatically | | Cost | Free | Meta conversation pricing plus your build |
On the feed side, Meta lets you schedule catalog uploads hourly, daily or weekly in Commerce Manager, with CSV, TSV, XLSX, Google Sheets or XML accepted. Hourly is the fastest cadence available there. Anyone selling you "real-time WhatsApp inventory sync" through a scheduled feed is describing something Meta does not offer. If you genuinely need live stock, you check stock at the moment the order webhook fires, not on a feed timer.
If you want the loop to run without a person in it, you need the Platform. The WhatsApp AI sales agent is the piece that sits on the webhook and does the pricing, the stock check and the link in one pass.
Where the money actually closes in the UAE
Every setup guide skips this part, and it decides whether the catalog earns anything.
WhatsApp's native payment rail only reaches a short list of markets, and the UAE is not one of them. India runs it for consumers on UPI. Singapore and Brazil have business-side versions, and Brazil's own went narrower in January 2026, when Meta dropped direct card payments to businesses there and left Pix and payment links as the fallback. None of that touches a merchant licensed in the UAE. So the working pattern in Dubai is: order arrives in WhatsApp, your system creates a payment link at a licensed gateway, the link goes into the same thread, the customer taps it, pays on the gateway's hosted checkout, and the gateway calls your webhook to confirm.
What that means practically, gateway by gateway:
- Telr, PayTabs, Ziina, Stripe and Network International all issue a hosted payment link that survives being pasted into a chat. Any of them closes the loop. Reported fee bands and the trade-offs between them are covered in our guide to payment links inside WhatsApp, so I will not repeat the fee table here.
- A bank transfer gives you nothing to paste. The customer leaves the chat, opens their banking app, and you find out whether they paid when a screenshot arrives, or when someone checks the statement on Tuesday. Every UAE business I have looked at that runs on transfers has a reconciliation backlog and no idea which chats are actually closed.
- Cash on delivery is still normal in the Emirates and it breaks the loop differently. The order is real, the money is not confirmed for days, and the CRM stage has to reflect that or your pipeline lies to you.
One detail that catches people. The prices in the catalog are the prices the customer saw when they built the cart. If your catalog is a day stale, the order arrives at yesterday's price and you are now negotiating a discount you never intended to give. Price the order server side, from your own data, and treat the catalog figure as a display value.
The full loop, end to end
A catalog makes money rather than work only when this whole chain exists:
- Customer opens the catalog from a message, an ad, or the business profile.
- They add items and send the cart. Your webhook receives the order with retailer IDs and quantities.
- Your system matches each retailer ID to a live inventory row, prices the order, and checks availability.
- If something is out of stock, the agent says so in the chat within seconds and offers the alternative. It does not go silent while someone checks the shelf.
- A payment link is generated for the confirmed total, carrying the order reference.
- The order and the customer are written into the CRM as a deal at the correct stage, with the item list attached.
- The gateway confirms payment on its webhook. The deal moves to paid. Fulfilment gets triggered.
- If nobody pays within a set window, a follow-up goes out. Not a blast. One message, referencing the specific cart.
Steps 6 and 7 are where most Dubai builds are quietly broken. The catalog works, the cart works, the link works, and the CRM knows nothing. Getting that half right is the subject of connecting WhatsApp to your CRM.
Where it breaks in practice
I run a WhatsApp agent on a live business every day, so this list is not theoretical.
The order arrives and nothing happens. This is the most common failure by a distance. The catalog is connected, the webhook subscription for order messages was never added, and orders sit in the inbox looking like ordinary messages. Nobody notices until a customer asks why their order was ignored.
The customer sends a cart and disappears. They built a cart, sent it, got a reply four hours later, and bought somewhere else. On our own studio number, the difference between a reply in under a minute and a reply in an hour is the difference between a booking and a read receipt. Carts are worse than enquiries here, because a cart means the customer has already decided.
Retailer IDs drift. Someone edits the product feed, the IDs regenerate, and every incoming order now references items your inventory system has never heard of. Freeze the retailer ID as your own SKU on day one and never let the feed generate it.
Items get rejected in review. WhatsApp reviews catalog items automatically against Meta's Commerce Policies. Alcohol, tobacco, weapons and medical products get pulled. Downloadable digital goods and subscriptions are prohibited outright. For Dubai this hits clinics, pharmacies, supplement sellers and anyone with a training course as a product. Half a catalog rejected two days before a campaign is a bad Sunday.
The one cart per thread rule bites. A customer who builds a cart, then wants to add something for a colleague, ends up amending one cart. If your process assumes two orders, it will produce one wrong invoice.
Nobody owns the abandoned cart either. A human sales rep decides, on a Thursday afternoon, that a cart "looked like a time waster" and never follows up. The agent has no opinion. It follows up on every cart, at the same interval, at 3am on a Saturday and at 9am on a Monday.
When you should not build a catalog
A catalog is a fixed price list with pictures. If your prices are not fixed, it will misrepresent you.
Skip it when your product is configured rather than picked. Studio rental is the example I know best. The rate depends on the zone, the block length, the time of day and whether a second zone gets held. A catalog would have to publish one number, and that number would be wrong for most enquiries. We do not use one. The agent asks two questions and quotes the real figure.
Skip it when you have fewer than about fifteen items. A well-written text menu inside the conversation converts as well, loads faster on a weak connection, and costs nothing to maintain. The catalog is a maintenance obligation, and an unmaintained catalog with two discontinued items and one wrong price is worse than no catalog.
Skip it when the buying decision needs a conversation. Real estate, aesthetics, high-ticket services, anything with a consultation in the middle. There the catalog is decoration and the qualification is the product.
Build it when you have a real SKU list, stable prices, repeat buyers who know what they want, and volume high enough that manual quoting is eating a salary. In Dubai that usually means retail, food, spare parts, cosmetics, gifting and B2B supply.
Cost and timeline in Dubai
| Scope | What you get | Cost | Time | |---|---|---|---| | Business app catalog, typed in | Browsable list, manual everything | AED 0 | An evening | | Catalog plus scheduled feed in Commerce Manager | Prices stay current, still manual selling | AED 0 plus your time | Two to three days | | Catalog plus AI agent plus CRM plus payment links | Order to paid with nobody in the loop | From AED 6,000 setup, from AED 1,200 a month | Around two to three weeks | | Full agency build | Same scope, longer contract | AED 20,000 to 50,000 in quotes clients have forwarded to me | Two to three months |
Meta conversation charges sit on top of all of it and are billed by Meta, not by us. Prices move, so confirm current figures before you budget.
The AED 6,000 line is our entry point for putting one process on autopilot, wired into whatever you already run rather than replacing it. We do not sell you a CRM or a gateway. We connect the ones you already pay for.
How we run this on our own number
SkyLight's studio number in Dubai Investment Park is answered by an agent, not a person, and has been for months. That business does not use a catalog, for the reason above. It uses everything else in the chain: intent detection, server-side pricing, a payment link in the thread, a deal written to the CRM at the right stage, and follow-up that never forgets. The month I am writing this, that contour produced AED 67,600 net with paid traffic running at roughly 5x return.
The catalog is one component. The plumbing behind it is the thing that pays.
