Malaysian Festive Ad Ramps, Measured (2026)

Chinese New Year copy is close to absent until 8 weeks out, when it jumps from almost nothing to 1.354% of that week's new Malaysian ads. It climbs for four more weeks and peaks at 8.199% in the week beginning 2026-01-18, which is 4 weeks before the festival. Then it falls, through the last month of the run-up, and by the week after the festival it is at 0.488%. The whole event is over before the day it is named after. That shape has a plain commercial reading. Chinese New Year buying is preparation: a reunion dinner booked, cookies and hampers pre-ordered, a new appliance delivered before the visiting starts. The advertising has to land while there is still time for the thing to arrive, so the creative peaks a month early and the last fortnight is a closing window rather than an opening one. One consequence is worth pulling out for anyone planning a Malaysian calendar. Content aimed at this window has to be indexed before the creative ramps, not alongside it, which means live in early December for a February festival. That is the guidance our own seasonal calendar already carried, and this is the first measurement that supports it.

Hari Raya behaves nothing like it. Festive copy is already measurable 9 weeks out at 0.28% and climbs steadily rather than switching on: 0.558% at six weeks, 1.565% at four, and up through the fasting month to peak at 2.571% in the week of the festival itself. Afterwards it decays slowly, still at 0.809% two weeks later and only falling away by week 4. The asymmetry is the finding. Raya is a month of Ramadan buying followed by a fortnight of open house, so the occasion continues after the date in a way Chinese New Year does not, and the creative follows it. An advertiser who switches Raya creative off on the day has left the tail of the season on the table; an advertiser who does the same at Chinese New Year is merely late to a party that ended a fortnight earlier. One caution about the ramp. Our own published guidance puts Raya content live around six weeks out, and the measurable switch-on is earlier than that. On this evidence the calendar is a little conservative for Raya and about right for Chinese New Year.

Inside each festive cohort the vertical mix is compared against every Malaysian ad starting in the same weeks, so both sides of the comparison carry the same collection behaviour and it cancels. At Chinese New Year, electronics and technology indexes at 234 and home and garden at 224, both far above the market rate, with consumer goods at 160. Clothing and apparel sits at 107, which is to say at the market rate, and that is worth stating because the received wisdom, our own planning documents included, leads the Chinese New Year story with fashion. At Hari Raya the picture is different. Transport indexes at 226, the highest of any vertical at either festival, which is balik kampung showing up as advertising: millions of people move across the peninsula in one week and the businesses that move them advertise it. Services follow at 141 and home and garden at 138. One limit on that comparison, and it is a real one. The cohort is defined by a single leading phrase, chosen because it is the largest, and a Chinese-language apparel ad may carry a Chinese phrase this cohort does not. So the apparel reading is soft and the transport reading, in a vertical that advertises in Malay, is firm. Every alternative phrase and its count is in the CSV so the sensitivity is checkable rather than asserted.

The obvious way to build this page is to compare each festival's vertical mix against a non-festive period, and that is what the first version did. It reported that software over-indexes at Chinese New Year and that the winning creative angle is a product screenshot. That is not a festive finding. It is the archive moving. This is a rolling scrape of ads that were running when they were seen. Its weekly intake across the observed year runs from 470 new ads to 62,911, a factor of 134, and what it collected changed along with how much. Any comparison between two separated blocks of time measures that drift first and advertisers second. The second version kept the vertical mix and controlled for the drift: a median baseline drawn from non-festive weeks on both sides, plus a decay test, on the reasoning that a real festive lift returns toward normal afterwards and a composition shift does not. It found no vertical lift at either festival that both cleared a meaningful threshold and decayed. Apparel indexed above the baseline during the Chinese New Year ramp and higher still in the weeks after it, which is drift wearing a festival's clothes. So the measure changed to the thing itself. A vertical label is a slow-moving property of an advertiser, and an apparel business is tagged apparel in June as much as in February. A phrase is an act: nobody writes kuih raya by accident. Every figure on this page is a share of the ads that started in the SAME WEEK, which is the normalisation our mega-sale timing study established, applied to a measure that can carry it.

Each festival was probed with a set of phrases across Malay, English and Chinese rather than one, and the phrase driving the weekly curve is simply the largest of them. For Chinese New Year that is the abbreviated form, at 4,335 ads across the window, ahead of the Chinese 新年 at 3,430 and the full English phrase at 1,474. For Hari Raya it is the festival's own name at 4,855, with ramadan close behind at 4,637 and the specific mechanics beneath: 1,516 ads say baju raya, 1,338 say balik kampung and 1,071 say duit raya. Probes are never added together, because one ad can carry several and there is no honest union available on this field. Every curve is also published against the phrase's own off-season floor, the level it idles at in the weeks belonging to neither festival. That floor is what makes a jump legible: it is the difference between a phrase becoming common and a phrase becoming visible at all.

It cannot tell you how many Malaysian advertisers ran festive creative. Every count here is a floor. The probes read ad body copy only, so a festive ad whose only signal is on the artwork is invisible, and a festive ad is short-lived by design, so it is less likely to have been caught by a rolling scrape than an evergreen one. That second point deserves precision, because it decides which claims on this page are safe. Short creative life depresses the LEVEL of every curve here. It does not move the TIMING, because it applies equally to every week in the window. So the onset, the peak week and the decay are publishable, and the absolute share in any single week is a floor rather than a rate. One more thing shapes this particular cycle. Chinese New Year fell 31 days before Hari Raya in the observed year, so the two ramps overlap for about a month. Measuring the language rather than the vertical is what separates them at all: an ad saying baju raya in the second week of February is unambiguous where an apparel advertiser starting a campaign that week is not. In the following cycle the two are five weeks apart, so a planner should not carry this year's collision forward as a permanent feature. And there is exactly one observed cycle of each. The archive's Malaysian slice is recent, so the previous year's festivals sit before it opens. No year-on-year claim is available from this data to anyone, ourselves included.

The whole dataset is one CSV under CC BY 4.0, free to reuse including commercially with attribution to AdPlay.ai. It carries the archive's weekly intake curve, every festive phrase with its language and its count, the full weekly series for each leading phrase with its own denominator, the off-season floor each phrase idles at, and the per-vertical cohort comparison, with the method note on every row.