Where Malaysian Meta Ads Actually Run (2026)

Meta publishes, for every ad in its public Ad Library, the set of placements that ad is eligible to appear on. That is a rare thing in this dataset: almost everything else we can cut by is a model's reading of the creative, and this is Meta's own record. The limit is in the word eligible. It records what the advertiser permitted when the campaign was built, not where the ad was actually shown, how often, or what any of it cost. An ad eligible on Audience Network may never have received a single Audience Network impression. So nothing on this page is a delivery figure, a reach figure or a spend figure, and it should not be read as one or repeated as one. Coverage is close to total, which is why the shares are trustworthy: 222,699 of 222,921 running Malaysian ads carry a value, 99.9%, and the 222 without one are excluded from every denominator on the page rather than quietly counted as zeroes. That distinction matters more than it sounds: the field defaults to an empty list, and an empty list matches no placement filter, so counting against every running ad instead would have understated every share on this page.

Facebook eligibility sits at 97.1% and Instagram at 96.4%. Effectively every Malaysian advertiser allows both, and the small remainder is mostly ads deliberately restricted to one surface. Below that pair the picture opens up. Audience Network is enabled on 66.9%, so roughly a third of Malaysian advertisers turn off the placement that puts their creative on third-party apps. Messenger sits at 60.8%. The two newest surfaces are adopted unevenly: Threads at 41.7% and WhatsApp at 25.1%, which is what a placement looks like partway through arriving rather than a preference. The practical read for a Malaysian advertiser is that the interesting decision was never Facebook versus Instagram. It is whether you leave the long tail of automatic placements on, and a large minority of your competitors have decided not to.

Counted as exact sets rather than as presence, 30 distinct placement combinations are in use across Malaysian running ads. The most common is Audience Network + Facebook + Instagram + Messenger + Threads at 22.1%. That number is the reason this page exists. Before writing it we set a threshold: if any single combination accounted for more than about 85% of Malaysian ads, there would be no page here, only a sentence saying everyone leaves the default on, and the sentence would go on an existing placements guide. At 22.1% the market is plainly making choices. The second-largest combination is the most telling: 14.3% of Malaysian ads are eligible on Facebook and Instagram and nothing else. That is not a default anywhere in Meta's campaign flow. It is an advertiser going into placement settings and switching the rest off, and roughly one ad in seven has had it done to it.

Cutting by vertical, the Facebook and Instagram floor barely moves: almost every category sits in the mid-nineties on both. Where verticals separate is on Audience Network, which runs from about 69% in the lowest category to about 77.3% in the highest. Read that spread carefully rather than as a recommendation. Verticals here are model-assigned and multi-label, so an ad can sit in two of them and no column sums to anything. And because this is eligibility rather than delivery, a category that keeps Audience Network on is telling you about its tolerance for cheap incremental placement, not about where its results came from. The honest summary is that placement choice in Malaysia varies far more between individual advertisers than between industries, which is the same shape the mega-sale timing study found on a completely different question.

The whole placement dataset is one CSV under CC BY 4.0, free to reuse including commercially with attribution to AdPlay.ai. It carries the field coverage, per-platform presence, every exact combination with its share, and the per-vertical cut, each row naming the source field it came from so you can see which numbers are Meta's own and which are cut by a model label.