When Malaysian Brands Launch Mega-Sale Ads (2026)

For each sale, every Malaysian ad that started between ten weeks before and three weeks after is bucketed by week. A vertical's centre of gravity is the ads-weighted mean of those week offsets: a value of minus three means its starts cluster three weeks before the sale. The spread between the earliest and latest vertical is 1 weeks at 12.12 and 1.1 at 11.11, across 22 verticals with enough starts to measure. At 9.9 and 10.10 it reaches 1.9 and 1.7 weeks, and those two cycles sit earliest in the archive's life, where the counts are thinnest and a single vertical can swing the extremes. A spread of about a week across an entire market is not a scheduling decision. If you sell electronics and your neighbour sells home goods, the measured difference in when your categories switch on mega-sale creative is smaller than the difference between shipping on a Monday and shipping on a Friday. Advice that tells you to launch earlier because of the vertical you are in has nothing behind it in this data.

The obvious question is how many weeks ahead the market ramps, and this method cannot answer it honestly. The archive holds an ad only if it was collected while running, and collection grew substantially across 2025, so later weeks in every window carry more ads for reasons that have nothing to do with Malaysian advertisers. Two of the four cycles peak in the weeks after the sale date, which is an artefact of that growth rather than a discovery about post-sale advertising. That is exactly why the vertical comparison is the part worth publishing. A vertical's share of its own cycle is measured against all ads starting the same weeks, so both sides of the ratio carry identical collection behaviour and it cancels. An absolute claim about market-wide lead time would need a control window drawn from outside the archive's own growth curve, and there is not one, so this page does not make that claim and neither should anything citing it. The weekly table above is published as counts rather than as a ramp for the same reason. It is useful for seeing the shape of a real cycle and for checking our working. It is not evidence that Malaysian advertising peaks in any particular week.

The CSV carries every week of every cycle for every vertical clearing the reporting floor of 150 starts: the raw count, that vertical's share of its own cycle, the same-week share across all Malaysian ads, and the index between them, over 1,431 rows. The useful way to read it is comparatively. Pick your vertical and a neighbouring one, put their weekly index columns side by side, and you can see whether one genuinely front-loads relative to the other within a cycle where the collection bias is shared. What you should not do is take one vertical's weekly counts on their own and read the shape as a ramp, because that shape carries the growth curve inside it. One label to ignore: the classifier's catch-all bucket swings hardest and means least, so it is excluded from every headline here and marked in the file.

The whole weekly dataset is one CSV under CC BY 4.0, free to reuse including commercially with attribution to AdPlay.ai. Each row states its sale, its week, its vertical, both counts, both denominators, the index and the method note that applies to it. If you are writing about Malaysian sale-season timing, this is a number you can check, which is more than can be said for most of the lead-time advice in circulation.