Meta Catalog Ads for Shopee & Lazada (2026)

Why your products live somewhere Meta's catalog does not want to point

Every catalog guide starts you in Commerce Manager. For a Shopee or Lazada seller, the blocker arrives before that, in one sentence of Meta's field reference.

Meta documents the link field like this: "Links must begin with http:// or https://, be valid and be hosted on your business's website domain. Don't provide a link to a Facebook domain (such as your business's Facebook Page) or somewhere else."

Read that against your reality. Your product page is lazada.com.my/products/.... That is Lazada's domain. It is not your business's website domain, and "somewhere else" is doing a lot of work in that sentence. The catalog spec assumes a merchant who owns a storefront and is describing its inventory to Meta. You are a merchant who rents shelf space, and the spec has no field for that.

This is a different problem from the pixel problem, and it is worth separating them cleanly because sellers conflate them constantly. The pixel problem is that you cannot install tracking code on a page you do not control, so purchase events never reach your ad account. That is covered in our guide to Facebook ads for Shopee sellers in Malaysia, and there is no point re-litigating it here. The catalog problem is narrower and comes first: you cannot put a compliant link in a feed you do not control. One is about what comes back. This one is about what goes in.

Here is the honest caveat, and it matters more than a confident answer would. Meta documents the own-domain rule. We could not verify what Meta actually does when a catalog item's link points at a marketplace domain. There is no primary Meta page we found describing rejection behaviour for that specific mismatch, and the search results that look like an answer are almost all Google Merchant Center documentation, which is a different platform with a different rulebook. So the correct posture is: the rule is real and quotable, the enforcement is unconfirmed, and the only person who can tell you what happens to your feed is you. Upload five SKUs, open Commerce Manager diagnostics, read what it says. That is a twenty-minute test that beats a month of forum arguments.

What a compliant Meta feed actually demands from you

Before you decide whether to fight the link rule, understand the size of the job. A catalog item is not a name and a price. Meta's field reference lists nine required fields for a standard product, and every one of them has to be produced from marketplace data that does not export in this shape.

The size field is required only for Shops in certain categories, so most sellers can leave it out of a first pass.

One correction worth making loudly, because it is the exact mistake this guide warns about. You will read, in a lot of places, that Meta requires "at least one of brand, gtin or mpn" as a unique identifier. That is Google Merchant Center's rule, not Meta's. Meta's field reference lists brand in its required set outright, and treats gtin and mpn as recommended classification aids rather than a required identifier trio. The distinction matters for an own-label marketplace seller: you cannot satisfy Meta by supplying a gtin and leaving brand empty, which is precisely what the Google-shaped advice implies you can do. If you sell unbranded goods you buy in and relabel, you still owe the field a real string. Importing the wrong platform's rulebook is the most common way a feed gets built confidently and wrongly, and it is worth checking a rule against Meta's own reference before you build a hundred rows on it.

Three of these deserve more than a table row.

The 65-character title recommendation. Meta allows 200 characters and recommends a maximum of 65 to avoid longer titles being cut off. Your Lazada title probably reads something like "Ready Stock Premium Organic Cold Pressed Coconut Oil 500ml Free Gift Shipping from KL". That is a marketplace search-ranking artifact, and once it is cut off your Meta ad shows a fragment of it that means nothing. Rewrite titles for the feed. This is not optional polish, it is the difference between an ad unit that reads as a product and one that reads as scraped junk.

The ringgit price format. This is the single most Malaysia-specific mechanical detail in the whole exercise, and it is where local feeds break silently. Meta's rule: "Format the price as a number, followed by a space and then the 3-letter ISO 4217 currency code. Always use a period (.) as the decimal point, not a comma (,)." So 59.90 MYR. Not RM59.90. Not MYR 59.90. Not 1,299.00 MYR. The feed is machine input, not shopper-facing copy: Meta's field reference specifies the input format and says nothing about how you would write the price in an ad. If you built the file in a spreadsheet with a currency format applied to the column, your export may carry an RM prefix or a comma separator you never typed. Open the raw CSV in a text editor and look at the characters, not at what the spreadsheet shows you.

The US-English stock vocabulary. availability "must be written in U.S. English. Supported values: in stock, out of stock." That is the whole list. Not "In Stock", not "available", not whatever string your Seller Centre export produces. Same for condition: new, refurbished, used. When a seller hand-maps a few hundred marketplace rows into a feed, this is a common break point, and the fix is a lookup table, not a judgement call.

And images. "Images must be in JPEG or PNG format, at least 500 x 500 pixels and up to 8 MB." Marketplace listing images are frequently composite banners with burned-in voucher text, free-shipping flashes and star ratings, because that is what wins a marketplace search results grid. Those are terrible catalog images and they will look like spam in a Meta feed. You need clean product shots re-hosted somewhere you control, and note a quiet operational trap: changing an image requires a new URL. You cannot overwrite product-01.jpg in place and expect the catalog to pick it up. Version the filename. You get up to 20 additional_image_link values if you want carousel-style variety.

The two routes into a catalog that actually exist for you

Meta documents four ways to get inventory into a catalog: the Commerce Manager UI, a feed with scheduled or manual upload, the Feed API, and the Batch API. Meta describes the UI route as suited to a "small product set, testing, and so on".

For a marketplace seller with no owned storefront and no developer, the two APIs are off the table. That leaves two.

Route one: type it into Commerce Manager. Create the catalog, add products by hand. Genuinely fine at ten or twenty SKUs, and the fastest way to find out whether your link approach survives diagnostics. It stops being fine the moment you have a hundred products or a price that changes mid-campaign, because there is no update mechanism except you, again, by hand.

Route two: build and maintain a feed file. A file with a row per product, uploaded manually or parked at a URL that Meta fetches on a schedule. Meta's field reference publishes a worked example CSV feed and a worked example XML (Atom) feed, so those two shapes are safe to copy; it points at a separate reference for the full list of supported formats, which we have not reproduced here because we could not read it. This scales. It also means you are the integration. There is no connector between Shopee Seller Centre and this file, and none between Lazada Seller Center and this file. Whatever gets into it, you put there.

We looked hard for a Meta-format feed export from either marketplace and found none documented. The careful way to say that: neither marketplace documents a native Meta feed export. If your account manager offers you one, take it. Until then, assume the file is yours to build.

Our general Meta product catalog setup guide covers the Commerce Manager wizard itself, and catalog diagnostics and errors covers reading the warnings once items are in. This page is about the part those guides assume you already have: a domain and a data source.

Where the scheduled feed quietly breaks for marketplace sellers

The scheduled feed is the feature that makes a catalog feel automatic. For you, it is a trap dressed as automation.

Meta's own advice for fast-moving inventory is to "Configure a scheduled feed for daily/hourly updates". Sensible, if something is generating the file. A scheduled feed refreshes what it fetches, and only that. If the file sitting at your URL is a static export you produced last Tuesday, an hourly schedule re-reads last Tuesday sixty times a day and reports success every time.

Nobody is pushing your Shopee stock changes into that file. When a Lazada flash sale drops a price from 89.00 MYR to 59.00 MYR, your feed does not know. When a SKU sells out on Shopee mid-campaign, your feed still says in stock and your ads keep spending on it. The failure is silent, which is the worst kind: no rejection, no diagnostic, just an ad promising a price the listing does not honour and a shopper who bounces.

So match the schedule to how often the file behind it genuinely changes. A weekly hand-rebuilt file on a weekly schedule is honest. An hourly schedule pointed at a monthly export is theatre.

Two things we deliberately will not tell you, because we could not verify them on a Meta primary page: the maximum feed file size (blog sources contradict each other outright) and whether there is a hard minimum refresh interval. Meta's inventory doc says "daily/hourly" and points at the Batch API for anything closer to real time. That is the extent of what is confirmed. If you need a firm number, read Commerce Manager's own limits in your account rather than trusting a blog figure.

The destination fork: your store, or the listing

Now the strategic part, and it is a genuine trade-off rather than a right answer.

Meta's objective set is six under ODAX: Awareness, Traffic, Engagement, Leads, Sales and App Promotion. The fork you face is Traffic versus Sales, and it is not a preference. It is decided by which page the shopper lands on.

Send them to your own store. You own the domain, so the link field is compliant. You can install the pixel, so purchase events flow back. That unlocks the Sales objective, which can optimise toward people likely to buy rather than people likely to click, and it gives you attribution you can actually read in Ads Manager. The catalog is the natural fit and everything works the way the documentation says it will.

Send them to the listing. The link rule is a problem, the pixel is impossible, and you are on the Traffic objective optimising for link clicks or landing page views. Attribution ends at the click. Meta tells you it bought some number of visits at some cost per visit; Shopee Seller Centre tells you some number of orders happened that day; nothing connects the two rows. You are reading two dashboards and inferring.

The reason anyone takes the second deal is that the listing converts. It carries the ratings, the review count, the platform vouchers, the free-shipping thresholds and the checkout the shopper has already used forty times. A cold visitor to a brand-new .my store with no reviews often converts worse than the same visitor landing on a Lazada page that looks established, even though the store gives you perfect measurement. Perfect measurement of a worse conversion rate is still a worse business.

Malaysia's reach pool is big enough that either path has room: Facebook's potential ad reach here was 23.0 million in October 2025, 63.7% of the population, with Instagram at 16.1 million, or 44.6%, against 35.4 million internet users at 98.0% penetration, per DataReportal's Digital 2026: Malaysia report. Reach is not the constraint. Attribution is.

And the marketplace is not going anywhere. Momentum Works put Southeast Asian platform ecommerce GMV at USD 157.6 billion in 2025, up 23% year on year, with Shopee, Lazada and TikTok Shop holding a combined 99% share of the region, and Malaysia among the fastest growers at 48% year on year (directional: the report itself is paywalled and we verified this through Yicai Global's reporting, not the report). One note on Malaysian money figures, since they look contradictory: DOSM reported national e-commerce revenue of RM937.5 billion for the first nine months of 2025, up 1.9%, but that is total national e-commerce revenue and it is B2B-dominated. It is not marketplace consumer GMV and it does not describe what Shopee and Lazada sellers transact. Different scope, different number, no contradiction.

If you are weighing the objective choice more generally, our Facebook ad objective guide walks the six, and attribution windows explains what the reporting can and cannot tell you once events do flow.

Collaborative Ads: the sanctioned route, and its gates

There is a third path, and it is the one Meta actually built for this situation.

Meta's own framing, from a December 2025 post about the Indian quick-commerce market: "Brands can use a retailer's catalog to promote products and drive shoppers directly to the retailer's online store." Meta describes it as connecting discovery on its platforms with conversions on retailer apps. That is precisely your problem stated as a product. The definition travels; the performance figures in that post do not, since they describe Indian quick-commerce partners and say nothing about Malaysia.

Mechanically it runs on a catalog segment. The retailer onboards, creates a segment (Meta: "The segment must contain all the products you would like to share with your brand partner"), and shares it to a brand's business ID, which the brand accepts in the Collaboration Center. The brand then advertises against it. The hard limit is worth internalising: "Brand advertisers cannot edit the catalog segment but they can create their own product sets from it." You never own the product data. You can slice it, not fix it. If the marketplace's title for your product is bad, it stays bad.

The naming is messy and there is no point pretending otherwise. Meta's docs say Collaborative Ads. Lazada and Shopee still say CPAS in their seller-facing material. Both names are live in 2026. Treat them as two labels for the same thing and move on; we found no primary source confirming a dated rename, so we will not narrate one.

Lazada publishes its gate. Lazada Sponsored Solutions states that access is "exclusively for whitelisted stores with a Traffic Enhancement Agreement (TEA) with Lazada". So it is commercial, negotiated, and not self-serve. Lazada also publishes a useful capability alongside it: its Catalog Audience can reach shoppers who "interacted with your store (e.g., visits, add to cart, purchases) within the last 180 days". That 180-day lookback is Lazada's own published figure for its own Catalog Audience. Do not generalise it to Shopee or to Meta at large.

The mechanical acceptance steps, if you get through the gate: after requesting the CPAS catalog in Seller Center, go to Meta Business Manager, open Collaboration Center, then Partners, find "Lazada Media Management", and click "Accept Assets".

Shopee does not. We tried to source Shopee's CPAS eligibility terms and could not, from any Shopee-owned public page. Shopee documents CPAS onboarding in its Seller Education Hub, but the hub renders client-side and returns an empty shell to a fetch, and the underlying endpoints return the same shell. Every specific number in circulation about Shopee CPAS eligibility traces back to agency blogs, not to Shopee. We are not going to repeat figures we cannot source, so the useful instruction is: confirm the terms with Shopee seller support, and treat any confident number you read elsewhere, including a minimum SKU count or an ad-account requirement, as unverified.

That contrast is worth sitting with. Lazada's route is restrictive but legible: you know there is a gate and you know what it is called. Shopee's route may well be more accessible, and you cannot find out without asking. For a seller planning a quarter, legibility has real value.

One prerequisite note, since it surprises people. Meta's get-started page for Advantage+ catalog ads lists a short set: a Facebook Page representing the advertiser (optionally an Instagram account), an ad account with registered payment information, and a catalog. A pixel is not on that prerequisite list, and dynamic product audiences are described as optional. That does not mean you should run without events; events are what makes optimisation work. It does mean the format itself is not gated on a pixel you cannot install, which is a narrower and more useful fact than the general advice implies.

Lazada in Meta's affiliate programme

Lazada announced on 8 July 2026 that it had joined Meta's Facebook Affiliate Partnerships programme, covering six Southeast Asian markets: Singapore, Malaysia, Thailand, Indonesia, Vietnam and the Philippines.

The shape: creators enable Professional Mode, connect their Lazada affiliate account under the Monetisation tab, tag Lazada products in their Facebook content, and earn commission on completed purchases. Instagram integration is described as planned, not live.

For a Malaysian Lazada seller this is a genuinely different route to Meta distribution. It is not a catalog you build and it is not media you buy. It is creator-led, commission-based, and the product data comes from Lazada's side of the wall. The feed problems above simply do not apply, because you are not the one supplying the feed. What replaces them is a different dependency: you are trusting Lazada's product data and a creator's judgement about which of your items to feature, and you have no field-level control over either.

The programme was announced in July 2026 and the operational detail (commission structures, which sellers are included, how the tagging surface behaves in practice) has moved since. Treat it as a path worth asking Lazada about before you build a plan on it.

The hybrid most Malaysian sellers land on

The tidy answer is "build your own store". The observed answer is "run both", and the reason is not indecision.

Signature Snack Sdn Bhd, founded in 2014, is a Malaysian FMCG brand and a certified B Corporation. It runs its own transactional site at signaturemarket.co, and it also runs a "Signature Market Malaysia Official Online Store" on Lazada Malaysia. That is the trade-off in this guide, embodied by one company. The owned site is where a pixel can fire, where the link field is compliant, and where the Sales objective is available. The Lazada storefront is where a chunk of Malaysian shoppers prefer to check out, vouchers and ratings and all.

The pure marketplace-destination case exists too. Senheng, a Malaysian electronics retailer established in 1989 that launched online in 2009 and operates roughly 100 physical stores nationwide, runs an official Lazada Malaysia storefront. A brand like Senheng advertising a specific model into that storefront takes the second deal knowingly: attribution ends at the click, because the listing is where a shopper is comfortable buying an appliance.

A brand like Signature Market typically runs a representative split: catalog-driven prospecting into the owned site where measurement is clean, and separate Traffic campaigns into the Lazada storefront during platform sale windows, when the vouchers are the offer and fighting them is pointless. Representative angles for how the hybrid is structured, not a claim about any ad running right now.

The practical consequence: your catalog is a project for the owned store, not a workaround for the marketplace. Stop trying to make a catalog point at Lazada, and start asking whether the owned store deserves to exist for other reasons. Usually it does.

A decision checklist

Run these in order. Each one closes off a branch.

1. Do you own a transactional domain today? If yes, build the catalog against it, install the pixel, run Sales. Skip to step 6.
2. Do you have fewer than about twenty SKUs? If yes, type them into Commerce Manager, upload a test batch, read the diagnostics. You will learn more from that in an afternoon than from any guide, including whether your link approach survives at all.
3. Can you actually maintain a feed file? Not "could you build one once". Could you rebuild it every week, forever, including the week of a flash sale. If the honest answer is no, do not schedule a feed, because a stale scheduled feed is worse than no catalog: it spends money advertising prices that no longer exist.
4. Is your store on Lazada? Ask your account manager about a Traffic Enhancement Agreement and CPAS eligibility. It is a commercial conversation, so have it early rather than discovering the gate mid-quarter. Ask about the affiliate partnership in the same call.
5. Is your store on Shopee? Ask Shopee seller support directly what CPAS onboarding requires today. Do not plan against a number you read on a blog.
6. Fix your product data before you fix your targeting. Titles rewritten to Meta's recommended 65 characters or fewer. Prices as 59.90 MYR. Availability as in stock in US English. Images clean, at least 500 x 500, no voucher banners, versioned filenames.
7. Decide what you will measure, before launch. If the destination is a listing, write down now that you will read Meta for clicks and Seller Centre for orders and that nothing will reconcile them. Agreeing to that in advance stops the mid-campaign argument about why the numbers do not match.

The catalog is not the goal. A shopper who buys is the goal, and for a Malaysian marketplace seller the catalog is usually a reason to finally answer a question you have been avoiding: whether you are a brand with a store, or a seller with listings. Both are legitimate businesses. They just run different ads, on different objectives, with different amounts of truth in the reporting.