Meta Ad Library Malaysia: Rival Ads (2026)

The one asymmetry that explains everything you see

Start here, because almost every frustration a Malaysian marketer has with the Ad Library traces back to a single design decision Meta made.

The library is broad. When Meta expanded it in March 2019, the point was that it would bring in all active ads any Page is running, not just ads related to politics or issues. That is the foundation of commercial competitor research, and it is genuinely generous: your rival's live creative is public, free, and searchable, whether they like it or not.

The data attached to those ads is narrow. The Meta Transparency Center is explicit that additional information, including spend, reach and funding entities, is provided for ads about social issues, elections or politics. Not for everything. For that category alone.

Put the two together and you get the Malaysian marketer's daily experience. Every ad, no money. You can see exactly what your competitor is saying, showing, and offering. You cannot see what it costs them or how many people saw it.

Malaysia makes this unusually vivid, because we are a country where the political side of the programme is switched on. When Meta opened political advertising here for GE15, it required advertisers to complete an authorisation process and carry a "Paid for by" disclaimer, and those ads appear in the Ad Library showing who saw them and how much was spent. So a Malaysian can go and look at real spend and reach figures right now, on a political ad, and then search a homegrown skincare brand and find nothing of the sort. Same tool. Same country. Different category.

That is not a gap you can work around with a cleverer filter. Once you accept it, the library stops being a disappointing analytics product and becomes what it actually is: the best free creative-intelligence feed in the market.

Set the market before you read a single ad

The Ad Library is organised by country, and the country you are looking at is the first thing to fix. It sounds obvious. It is skipped constantly, usually by someone who searched a global brand's name, got hundreds of ads, and started drawing conclusions from creative that will never run in a Malaysian feed.

There is a reason to be scrupulous here rather than just directionally careful. We could not find Meta documentation stating whether the country view reflects where an ad is delivered or where the advertiser is based. Those are different questions with different answers for any brand running regional campaigns out of Singapore or Jakarta into Malaysia. Absent a clear answer, do not assume one. Treat the country view as a strong narrowing signal and verify anything load-bearing by reading the ad itself: the currency in the offer, the language, the delivery promise, the landing page. An ad quoting RM and promising West Malaysia delivery in three days is running here regardless of what any filter told you.

A caution on the rest of the controls, including from us. Meta reworks the Ad Library's interface and relabels its filters regularly, and Meta's own help pages describing the current control set are not reliably readable from outside the product. What can be said honestly: as of a 2022 Social Media Examiner walkthrough, the library offered a keyword and business search bar alongside filters for platform (Facebook or Instagram), media type such as image or video, and language. That is directional, from 2022, and the live tool is the authority. Open it, see what is actually there, and build your process around the controls you can see rather than the ones a guide told you to expect. Any article that gives you a confident dropdown-by-dropdown tour of this tool is describing a UI that may have changed twice since it was written.

Why the market matters more than the tool here

The reason to narrow hard to Malaysia is that the population you are studying is enormous and the behaviour inside it is specific.

DataReportal's Digital 2026 Malaysia report puts Facebook's ad reach here at 23.0 million, equal to 63.7 percent of the total population, with Instagram at 16.1 million, or 44.6 percent. There were 30.7 million social media user identities in October 2025, equivalent to 85.0 percent of the population, against 35.4 million internet users at 98.0 percent penetration. Practically every Malaysian with a phone is reachable, which means your competitors are advertising to more or less the same people you are, at the same time.

Malaysia trades online at scale to match. Total e-commerce income across the whole economy reached RM937.5 billion in the first nine months of 2025, up 1.9 percent year on year, according to the Department of Statistics Malaysia as reported by Bernama. Read that number for what it is and no more: it is a national accounting total spanning business-to-business, consumer and government transactions, so it is not a pot of consumer spend your skincare ad is competing for. It tells you commerce here runs through screens. It does not size your market, and anyone using it to imply otherwise is padding.

The reach figures are the ones that carry the argument. In a market where practically every adult is reachable on the same two platforms, the differentiator is rarely the audience you found. It is the creative you put in front of an audience everyone already has. Which is precisely the thing the Ad Library shows you for free.

Build the shortlist before you open the library

The most common way to waste an afternoon in the Ad Library is to search whoever came to mind, scroll their ads, feel vaguely informed, and close the tab. The fix is to decide who is worth watching before you look at anything.

Three tiers work well for a Malaysian SMB.

To make the tiers concrete rather than abstract, the examples below name real advertisers from our Malaysian ad archive, using a homegrown skincare seller as the worked case. They are representative of how each tier behaves, not a claim about what any of them is running today.

Tier one: your actual peers. Businesses selling a comparable product at a comparable price to the same buyer. Three to six of them, no more. These are the only Pages whose tactics you can lift more or less directly, because their constraints match yours. If you sell a RM89 supplement direct to consumer, a rival selling a RM79 supplement direct to consumer is your tier one. A pharmacy chain is not. For a homegrown skincare brand, the archive shows this tier is where the value-led locals sit: Glad2Glow running a five-piece bundle at RM58.50 into Ramadan, CLEF Beauty and Dr. Liorae Skin Lab working the same price-aware shopper with fairness and even-tone claims. Their constraints are yours, which is why their offer mechanics transfer.

Tier two: the brands you want to become. One rung up in budget and maturity. You are not copying these, you are reading direction: which categories they are expanding into, what claims they are willing to make, how their production quality has moved. SAFI is the instructive example for a local beauty seller, a homegrown name competing on halal credibility at a scale a young brand cannot buy, alongside the clinic advertisers like New York Skin Solutions whose before-and-after machinery runs on budgets you should not try to match. Expect to find brand-led work you cannot afford and should not imitate.

Tier three: adjacent categories sharing your customer. This is where new angles come from, and it is the tier most people skip. Your buyer is not only shopping in your category. A hijab brand watching only other hijab brands will discover, at best, the same restock announcement everyone runs. Watch across the basket instead. The same Malaysian woman buying a serum is being sold to by Afabelle Hijab bundling four instant awning hijabs at RM99 with free cash on delivery, by Naelofar on seasonal lookbooks, and by the hijab-friendly corner that Asma Harun Hijab Hair Care and Anaaka Halal Skincare Malaysia quietly own. That last pairing is the tell: a skincare seller learns more about wudhu-friendly and under-hijab claims from a hair-care brand than from another serum. The Raya sequencing that works in one of those categories usually transplants.

Keep the whole list to about a dozen Pages. A shortlist you check weekly beats a comprehensive one you check never.

Two selection filters save you from bad picks. First, prefer Pages that clearly run offers, not just brand posts: an advertiser you can learn pricing psychology from is worth more than a prettier one you cannot. Second, be suspicious of a competitor everyone in your category already watches, because whatever they are running is already priced into the auction by the time you copy it. The unglamorous tier-one rival nobody talks about is usually the better teacher.

What to read when there are no numbers

Here is the discipline that separates useful Ad Library work from screenshot collecting. You have no spend, no impressions, and no reach for any commercial ad. So read what is left, which is more than people assume.

Repetition. One ad tells you a brand tried something. The same angle appearing in four variations tells you they believe in it enough to build around it. Variation counts are the closest free proxy you have to conviction.

The offer mechanics, in ringgit. This is where a Malaysian feed gets specific fast. Is the hook a price cut, free shipping, a bundle, cash on delivery, a voucher stack timed to a double-digit sale? Malaysian ecommerce buyers are trained on a particular rhythm of offers, and reading twenty rival ads tells you what the going rate for attention is in your category right now. When three tier-one competitors have all moved to free shipping above RM100, that threshold is the market, and your RM150 threshold is a tax on your conversion rate.

Language treatment. Not which language, but how seriously it is handled. A brand that rewrites its hook natively in Malay, with an offer built for that audience, is investing in the segment. A brand running an obviously machine-translated English headline is hedging. That difference is strategy, visible for free.

Who is in the frame. Founder to camera, a paid creator, a stock model, a flat product shot. Production choices reveal budget and, more usefully, reveal what the brand thinks earns trust with a Malaysian buyer. The shift from polished studio work to a phone-shot testimonial in someone's kitchen is a decision, and when several rivals make it in the same quarter, it is a signal.

The seasonal clock. Malaysian creative is calendar-locked to an unusual degree: Ramadan, Raya, Merdeka, the 9.9 and 11.11 sale dates, Deepavali, Chinese New Year, the year-end run. Watching when a competitor starts a festive push, not just what it says, tells you the lead time they have found works. That is a hard-won piece of local knowledge and it is sitting in the library for anyone patient enough to note the dates.

None of these give you a number. Together they give you a defensible read on what a competitor believes, which is what you were actually trying to buy.

The archive problem, stated honestly

This is the point most guides get wrong, and it deserves precision rather than confidence.

Meta documents exactly two retention regimes. Ads about social issues, elections or politics are stored in the Ad Library for seven years, visible whether active or inactive. Ads that deliver impressions in the EU are displayed while active and archived for one year upon the delivery of their last impression. Meta documents no equivalent archive for an ordinary commercial ad in Malaysia.

The reasonable inference is that a Malaysian competitor's ad is available to you while it runs and gone once it stops. You will see this asserted flatly all over the internet as a Meta rule. It is not one. Meta has not stated it in those words, and the honest position is that we know what Meta has documented and we are reading the silence.

That uncertainty is academic, because it does not change the correct behaviour. If there is any real chance the record disappears when a rival hits pause, you cannot rely on looking it up later. So capture on a schedule. Screenshot the ad, note the date you saw it, note the offer and the format, and keep it somewhere with a timestamp. Do that for six months and you own something the live library cannot sell anyone: a dated history of what your category actually ran, including the tests that quietly died. A searchable archive of Malaysian ads captured over time, such as AdPlay.ai, is one way to hold that record, but a disciplined folder and a weekly calendar block does the same job for a team of one.

The competitor you most want this record for is the one running short tests. Long-running ads are easy, they wait for you. The three-week experiment that got killed is the interesting one, and it is exactly the ad you will never find again.

Where the extra data does exist, and why it is not for you

Two Meta tools sit alongside the library and both are worth understanding once, so you stop hoping.

The Ad Library Report gives an aggregated view of ads about social issues, elections or politics in a selected country for a given time period, and Meta says it is available in all countries that require authorisation to run those ads. Malaysia requires that authorisation, so the Report logically covers us, though we could not verify Malaysia on a Meta country list directly and Meta's country-list pages are difficult to read from outside the product. Either way, the Report aggregates political advertising. It will never mention your competitor's face serum.

The Ad Library API enables deeper analysis of ads about social issues, elections or politics, ads targeting the EU, and branded content, for authorized users. Malaysia falls into none of those buckets, which is exactly why there is no supported export path for a rival's Malaysian product creative. It is not a general commercial-ad research API, and if your plan involves pulling a competitor's serum ads through it programmatically, that plan has no route.

One more piece of the political-category machinery matters only if you are in it, and it is worth pulling apart because it is reported wrongly almost everywhere. Two separate obligations are in play. Advertisers running ads about social issues, elections or politics already have to disclose AI-generated or AI-edited media that depicts a real person saying something they did not say, a realistic-looking person who does not exist, or an event that never happened. That duty is live and Meta rejects ads that ignore it. What changed on 1 June 2026 belongs to Meta, not the advertiser: since that date Meta has additionally used automated detection to identify ad media created or edited with third-party generative AI tools, then applies an "AI Info" label itself, with no advertiser action required. Neither half makes this a general rule covering every commercial ad. And because Meta does not document how wide the automated net is cast, the absence of an AI label on a rival's ad is weak evidence about how that ad was made, not proof of anything.

Finally, there is the Ad Targeting dataset, which covers social-issue, electoral and political ads from August 2020 onward across more than 120 countries, with entries remaining for seven years. Read that scope carefully, because stripped of it the headline sounds like a general targeting database for 120 markets. Same story as the rest: it exists, it is real, and it is not a window into a Malaysian retailer's audience strategy.

A worked example: twenty minutes on one competitor

Abstract advice is easy to nod at and hard to act on, so here is the pass itself, run on a fictional but ordinary situation: you sell a RM99 hair serum direct to consumer, and you have picked a tier-one rival selling a RM89 equivalent.

Minutes one to three: set the market and take stock. Look at their Malaysian ads and count. Not a precise count, a bucket: are you looking at three ads or thirty? Three means a brand dripping one message. Thirty usually means either a real testing programme or a catalogue campaign spitting out variants, and you can tell which by whether the creative differs or only the product shot does. Write the number and the date down before you scroll, because next week's number is the actual insight.

Minutes four to ten: cluster by angle, not by ad. Ignore the individual ads and sort them into messages. You will typically land on three or four: a hair-fall problem hook, a founder story, a before and after, a price offer. Now count how many ads sit in each cluster. If eleven of their fifteen ads carry the same problem hook with slight variations, that is their bet. The single lonely founder-story ad is a test, not a strategy. This clustering step is where the free library quietly gives you the conviction data Meta refuses to publish as a number.

Minutes eleven to fifteen: read the offer in ringgit. What exactly is being promised? RM89 flat, or RM89 with free shipping, or two for RM150, or RM89 with a cash-on-delivery option? Note the shipping threshold and the payment mechanic specifically, because those are the two levers Malaysian ecommerce buyers respond to most predictably and the two your own checkout can change this afternoon. If their offer has moved since last month, that is worth more than any creative observation: an offer change means something in their numbers forced it.

Minutes sixteen to twenty: capture and conclude. Screenshot the two or three ads that represent each cluster, date them, and write one sentence. Not "they are running before and afters." Something like: "As of 22 October they have eleven of fifteen ads on hair-fall anxiety, moved to free shipping with no minimum, and dropped the founder angle they ran in August." That sentence is your asset. It is dated, it is specific, and in three months it will let you see a trajectory that nobody reading the live library today can see.

Notice what never appeared in those twenty minutes: a spend figure, an impression count, a CPM. You did not need one. You now know what your competitor believes, what they are willing to give away, and what they abandoned, which is enough to decide what you test next.

Troubleshooting the common dead ends

Turn it into a habit that survives a busy quarter

The library only pays out over time, so make the routine small enough that it happens.

Fifteen minutes a week across your tier-one Pages. Note what is new, what disappeared, and what is now in its fourth variation. Once a month, spend a longer session on tier two and tier three, looking for direction and borrowed angles rather than tactics. Before each seasonal peak, run a heavier pass two to three weeks out, because that is when rotation is fastest and when the ad you meant to save vanishes.

Write down what you conclude, not just what you saw. "Three rivals moved to free shipping above RM100 in September" is an asset. A folder of screenshots is homework you never marked.

And then do the thing the research was for. Competitor reading is upstream work: it tells you which angle is worth testing, it does not test it. The team that wins the category is not the one with the fullest swipe file, it is the one that turned a read on a rival's Raya sequencing into their own ad, in market, in a week. Our companion guides on reading the Meta Ad Library and seeing competitor Facebook ads go deeper on the mechanics, and estimating competitor ad spend covers what you can and cannot infer about their budget from the free signals.

The Ad Library will not tell you what your competitors spend. In Malaysia, it never will, unless they run for office. What it will tell you, for free, every week, is exactly what they think works. That is the more useful half anyway.