What Malaysian Facebook Ads Promise (2026)

Warranty leads at 2.5% of Malaysian running ads, free shipping follows at 2.2%, then buy-one-free-one at 1.3% and a free gift at 1.3%. The number that will surprise anyone who has read Malaysian ad advice is cash on delivery, at 0.1% of ad bodies. COD is described in the local marketing literature as the mechanic that unlocks Malaysian conversion, and almost nobody writes it in the ad. There are two honest readings and this page cannot choose between them. Either the mechanic is offered at checkout rather than advertised in the creative, which would make the advice right about the market and wrong about the ad; or it is far less central than the advice claims. Both would produce this number. What none of this measures is whether any phrase works. The archive holds creative metadata and no spend, no results and no conversion data, so a common phrase is a common phrase. Prevalence is not performance, and a page like this is at its most dangerous when a reader quietly converts one into the other.

Every mechanic here was probed twice, once in English and once in Malay, and the two counts are published side by side and never added. The split does not run the way a single "write in Malay" rule would predict. Instalment plans are written in Malay far more than in English: 2,218 ads carry ansuran against 1,639 carrying installment. Scarcity runs the same way, with 1,830 ads carrying stok terhad against 1,286 for ready stock. Warranty and the official-store claim run hard the other way, overwhelmingly English. The pattern that suggests itself is that the mechanics tied to trust and to formal guarantees stay in English, while the ones tied to money pressure and urgency switch to Malay. That is a hypothesis this dataset can support but not prove, and it is offered as one. One mechanical limit shapes all of it. The probes run against the ad's own body copy, and the archive's English translation field is not part of the searched surface, so an English probe does not silently pick up its Malay equivalent. That is what makes the pair meaningful. It also means a mechanic phrased in Chinese is invisible to both probes.

National averages flatten a market where a warranty means everything in one category and nothing in another. Electronics and technology carries a warranty mention on 8.4% of its running ads and home and garden on 7%, both several times the national rate. Retail and marketplace leads instead on free shipping at 4.2%, which is the mechanic a marketplace seller actually controls. Beauty and personal care leads on buy-one-free-one at 2.4%. Read the vertical rows as the page's real output and the national row as context. If you sell an appliance, the relevant fact is that a large minority of your category writes a guarantee into the creative. If you sell on a marketplace, it is that shipping is the promise your category makes and warranty is not. Verticals here are model-assigned and multi-label, so an ad can sit in two of them at once and the columns do not sum to anything. A vertical below the reporting floor is measured and shipped in the CSV but kept off the page, because a probe share drawn from a few thousand ads moves too much to publish.

Three limits, each of which pushes every number on this page down rather than up. The probe reads the ad's body copy only. An advertiser whose entire promise is the words FREE SHIPPING laid over the artwork is invisible here, because the only other surface available is model-generated text recognition whose coverage cannot be quantified, and publishing an unquantifiable floor next to a quantifiable one invites a comparison that neither supports. The probe is an exact phrase. An ad offering free delivery in words we did not probe for is not counted, which is why several mechanics carry more than one phrase and why the query string behind every figure is in the CSV rather than described. And the denominator includes every running Malaysian ad, including the ones with no body copy and the ones from advertisers for whom the mechanic is meaningless. A law firm does not offer free shipping. Excluding those would need a filter on a field that is deliberately not filterable, so they stay in, and every share here is therefore a share of the whole market rather than a rate among advertisers who could plausibly have said it.

The whole dataset is one CSV under CC BY 4.0, free to reuse including commercially with attribution to AdPlay.ai. It carries every probe with its language, its exact query string, its count, its denominator and its share, the per-vertical cut, and a validation row for each headline probe showing the count under both of the index's matching strategies, so you can see that a two-word phrase was matched as a phrase and not word by word.