Korean Celebrity Ad Films on YouTube: Views, Costs and What Brands Get Back

Seven of the ten were uploaded in 2026, one in 2025 and two in 2024, and all sit on the advertiser's own channel rather than a broadcaster's. Most casts are idols or actors, and a few faces repeat: Go Youn-jung and Park Ji-hoon each star in two films, and aespa's Karina is in two on her own, plus Nongshim's film with aespa.¹

Brands also split one film into several uploads, and the shorter cuts tend to win. The Community Chest's 15-second cut has about twice the views of its 30-second version, Venti's 15-second cut beats its 30-second one, and Jinro's 30-second cut has 108.9 million to 1.65 million for the full-length version.¹

The money has moved online. In 2024 Korean advertisers spent 17.13 trillion won, of which 10.10 trillion (59%) went online and 3.22 trillion (18.8%) to broadcast. Broadcast spending has fallen each year: 4.02 trillion won in 2022, 3.39 trillion in 2023 and 3.22 trillion in 2024.¹⁶

That doesn't mean YouTube beats television. A 2025 study for the Korea Advertisers Association, using Nielsen audience data, found TV ahead of digital on awareness (35.2% to 31.4%), preference (16.1% to 11.0%) and purchase intent (13.2% to 8.7%), and cheaper per rating point (1.05 million won against 3.47 million won). On cost per view, though, YouTube was among the cheapest media tested at 21 won per view of six seconds or more.¹⁷

We found no source that explains the choice for celebrity ad films specifically, so the reasons below are our reading of those numbers, not a finding: a film on the brand's own channel needs no airtime, reaches viewers outside any TV schedule, and produces a public view count that can be quoted in a press release, as nearly every source in this post does.

The only cost figures that reach the public are press reports of star fees and campaign budgets, and every one is an estimate or a company's own statement.

All figures are as the outlets reported them. The Segye Ilbo fees are marked "reportedly" in the article.¹³ Compose Coffee's 7.35 billion won is the whole advertising budget for the V contract year, with 5.4 billion won for V-driven branding and 1.95 billion won for other marketing, so it is not V's fee.¹⁴ Segye Ilbo also reports that banks' contracts carry a morality clause under which they seek penalties of two to three times the fee paid if a model's conduct creates a risk.¹³

Production cost: not found. We could not find a reliable published figure for what it costs to shoot a celebrity ad film in Korea. The reports we could open on the state ad agency's survey and the advertisers' association study give media spend and star fees but not production budgets, so this post gives none rather than an estimate.

In the Compose Coffee case the bill is shared: the headquarters pays 60% and franchisees 40%.¹⁴

No brand in the table has published a return figure. What exists is reported views plus a few sales claims, and each is the company's own or a journalist's attribution, not a measurement.

• National Geographic Apparel, 2026: the campaign's Reversible Fleece ranked first overall on Musinsa on its first sales day and the prepared stock sold out. No unit or revenue figure is given.³
• NH Nonghyup Card, 2026: more than 10,000 zgm debit cards were issued within four days of launch. No baseline is given.¹⁰
• Binggrae Super Cone with Son Heung-min, 2019: after the 30-second ad was posted to YouTube on April 11 and reached about 1.2 million views in ten days, a Binggrae official said orders for the previous week were up about 25% on the period before the ad. The same official said that comparing ads with sales usually takes at least one to three months.¹⁵
• Nongshim, 2026: overseas subsidiary sales rose 26.8% in the first half, but the article doesn't connect that to the ad.⁵
• Compose Coffee, 2022 to 2024: across seven franchise industries in a Leaders Index analysis, headquarters sales rose 10.8% while average sales per store fell 7.6%. iNews24 says an individual store can't tell whether a sales rise came from advertising or from local trade and the economy.¹⁴

Research on the other side of the ledger is US-based. A study of 93 celebrity disgraces between 1986 and 2011 found significantly negative abnormal returns for the firm's stock around each announcement, and none when the firm ended the contract.¹⁸ It says nothing about Korea, but it is the risk the banks' morality clauses are written for.

The table is not a ranking of Korea's most-watched celebrity ads. We built it from ads that were reported in the press or that we found on advertisers' channels, so it is skewed towards hits, and a quieter ad from a bigger star may be missing. The view counts are also not all of an ad's reach: TV airings, Instagram and the other cuts are excluded, and the brand-supplied totals quoted above mix platforms. No source we opened gives the cost of producing a film or the sales a given film caused.