Do AI Paid Ads Drive Indonesian Brands' Global Growth?

As of August 1, 2026, the public record does not support AI-driven paid media as the main engine behind the most cited Indonesian or Indonesia-adjacent brands going global. For Indonesian brands planning cross-border paid ads and AI strategy, the dated case files point somewhere less fashionable and more expensive to fake: app ecosystems, outlet density, marketplace distribution, logistics execution, and local market exposure.

Kopi Kenangan: the richest record, and the weakest case for retroactive AI attribution

Kopi Kenangan is the cleanest test of the claim because the company gives enough dated operating numbers to separate “the brand grew” from “AI ads drove the growth.” The company reached its first full-year group profit of US$17 million in 2025, while revenue rose 45% year over year to US$184 million. Its footprint reached 1,324 outlets across six countries, with 1,136 outlets in Indonesia and about 158 in Malaysia. Malaysia revenue nearly doubled to RM111 million [1][2].

Those are not soft awareness metrics. They are store count, country count, revenue, profit, and market-level revenue. If paid media was the strategic engine, this is where the public record would need to show a bridge from campaign spend to incremental store sales, app orders, new Malaysian customers, or repeat purchase. The available record instead makes the digital ecosystem and the store base visible.

The app numbers matter because they describe a system that can compound beyond a single campaign. Kopi Kenangan’s digital ecosystem added 4.47 million new customers, new-customer acquisition rose 159%, and monthly transacting digital users increased 116% to 1.5 million [1][2]. That does not prove no paid media was used to acquire users. It does show that the measurable asset in the file is the owned digital loop: customers can order, transact, return, and be recognized again.

That distinction is not academic for a first-market budget. An app ecosystem can lower the cost of the next transaction if the product, store coverage, delivery radius, and offer cadence hold together. A campaign can only keep buying attention. The cited Kopi Kenangan record gives a growth lead more reason to fund app acquisition, CRM, order flow, outlet readiness, and local product-market fit than to declare automated media the center of the Malaysia result.

The Satu Kenangan failure makes the attribution tighter, not looser. Kopi Kenangan halted the hyperlocal experiment, and CEO Edward Tirtanata said, “a lower price point does not automatically guarantee success” [1]. That is a useful failure to keep in the file. It says the company did not win simply by discounting or widening distribution. Local proposition, format, customer behavior, and operating model still had to work.

There is a reported Kopi Kenangan marketing-spend reduction in snippet-only material, but it should not carry the argument here. Without the full source record, methodology, category definition, and market split, it is not a safe basis for deciding whether paid media became more or less efficient. The cleaner conclusion is narrower: the verifiable sources document profitable expansion, Malaysia scale, outlet growth, and app-user growth; they do not document an AI-paid-media result that deserves strategic credit.

Malaysia is media-relevant. That still does not make it an AI-ads proof point.

Malaysia is the strongest part of the file for anyone arguing that media investment matters. Campaign Indonesia’s market record cites a Ninja Express survey in which nearly half of Malaysia’s online shoppers bought Indonesian brands in the prior year, BPS data showing 2.27 million Malaysian tourist visits to Indonesia in 2024, and a Vodus 2024 survey showing 73% willingness to try new brands, rising to 80% among regular social media users [5].

That is not a “nearby, therefore easy” market. It is a market with signals of openness, familiarity, and social discovery, but still requiring local work. Dex Yeoh, Associate Director at Ipsos Synthesio Malaysia, put the constraint plainly: Indonesian and Southeast Asian brand penetration “has not yet reached Malaysia’s main market” and “comes down to brand awareness and how much these brands invest in media exposure” [5].

That line deserves attention because it ties the outcome to awareness and exposure. It does not, however, isolate paid social, paid search, creators, marketplace media, PR, offline visibility, delivery-app presence, or AI-optimized bidding. A Malaysia entry plan can justify media. It cannot honestly cite this market record as proof that AI-paid ads are the growth driver.

J&T Express: expansion explained by operations, not consumer persuasion

J&T Express should not be stretched into a consumer advertising story just because it is an Indonesian-founded cross-border name. Its FY2025 record is about logistics execution: revenue rose 18.5% year over year, free cash flow increased 96.1% to US$494 million, and new markets achieved full-year profitability for the first time [3].

The likely strategic assets are visible in the business model: route density, delivery reliability, cost control, and partnerships with e-commerce platforms. If those assets improve, merchants and platforms have reasons to keep volume inside the network. Paid media may exist somewhere in the company’s marketing mix, but the cited FY2025 evidence does not present consumer media as the lever that made new markets profitable.

For a growth lead, the implication is simple. If the product being exported depends on fulfillment, delivery experience, or platform trust, the first budget argument belongs to the operating layer. Paid media can create demand faster than the network can satisfy it; that is not growth, it is a queue of disappointed customers.

Skintific: marketplace dominance, with an identity caution flag

Skintific is useful because it shows what marketplace distribution can do in Southeast Asia, but it should be handled carefully in an Indonesian-brand discussion. TMO Group identifies Skintific as a Guangzhou Feimei brand, even though it has been positioned in ways that make it appear closely tied to Indonesia in market conversations [4]. Any strategy that depends on “Indonesian brand identity” should not use Skintific as a clean ownership example.

As an e-commerce case, though, the numbers are hard to ignore. TMO reports RMB 800 million, approximately US$114 million, in cumulative Southeast Asia sales over two years; TikTok Shop 2022 Global Sales Champion status; and No. 3 skincare brand position across Shopee and Lazada [4]. Those are distribution outcomes inside social-commerce and marketplace systems.

Paid traffic may well be part of a marketplace beauty play. The category often relies on creator proof, platform visibility, livestreaming, search placement, review volume, pricing, bundles, and retargeting. But the cited Skintific record does not break out which share came from paid ads, which came from organic platform rank, and which came from social-commerce mechanics. It certainly does not establish AI bidding or AI creative as the causal layer behind the result.

Driver scores: evidence of contribution, not a ranking of fashionable channels

The scoring below measures the strength of public evidence that a driver carried cross-border growth. It is not a claim that a low-scored driver had no effect. A score of zero means the cited sources do not provide usable evidence for that driver in the named case.

Kopi Kenangan scores highest on app and operations because the public numbers sit exactly there. J&T scores highest on operations because the result being reported is profitability and free cash flow in new markets. Skintific scores highest on marketplaces because the record names TikTok Shop, Shopee, and Lazada positions. The paid-media row is low not because ads are useless, but because the cited evidence does not show ads carrying the outcome.

This is where automated platforms can create a planning trap. Performance Max, Advantage+, AI Max, TikTok automation, and similar systems can reduce waste when an account has enough conversion signal and clean business constraints. They can also overfit to thin first-market data, chase cheap proxy events, or reallocate spend faster than the market team can interpret what is happening. That same pressure shows up in automated bidding under cost stress, where the system can keep optimizing inside a narrow measurement frame while the business problem sits outside it; the mechanics are discussed further in this site’s record on PMax, Advantage+, and AI Max behavior under spend pressure.

The AI adoption record is not an effectiveness record

MMA Global Indonesia’s State of AI in Marketing is useful context precisely because it stops the conversation from pretending the market has already standardized around mature AI-led growth systems. The report shows that 16% of Indonesian marketers have fully integrated AI, while 38% are still experimental and 32% are partially integrated. It also reports that 49% cite content-marketing application as their biggest AI challenge [6].

Those figures support a narrow conclusion: AI is present in Indonesian marketing workflows, but full integration is still limited. They do not show that AI-paid ads caused cross-border growth for Kopi Kenangan, J&T, or Skintific. Adoption is not effectiveness. A platform setting being available is not proof that it produced incremental Malaysian revenue. A vendor case study is not the same as an operator-supplied account record with market split, spend, attribution window, baseline, and holdout logic.

That matters most in a new SEA market, where the first conversion data is usually thin and biased toward the easiest early adopters. The model may learn from tourists, existing fans, expatriate communities, discount seekers, or marketplace bargain traffic before the brand has reached the main market. Signal decay and weak proxy events are not only a youth-audience targeting problem; they also bite when a new market has too little stable conversion history. The measurement risk is similar to the one covered in this site’s discussion of signal decay and ad targeting.

Creative automation deserves the same caution. Default-on enhancements, generated variations, and campaign-level automation may be useful, but they should not quietly rewrite the brand’s first-market learning agenda. If a brand has not yet learned which Malaysian customer, price point, product format, language cue, marketplace page, or delivery promise converts profitably, automated creative rotation can make the dashboard busier without making the strategy clearer. The caveat is similar to the one raised in this site’s note on Advantage+ Creative Enhancements and default-on automation.

What to fund before calling it an AI paid-ads strategy

The first cross-border budget should follow the evidence order, not the recap-deck order. If the brand is a store-led consumer business, fund the route to repeat transaction first: outlet placement, delivery coverage, app onboarding, CRM, local payment behavior, and product-market fit. Kopi Kenangan’s public record gives more support to those lines than to a large AI-media allocation.

If the brand depends on e-commerce, fund the marketplace layer before overfunding external acquisition. That means product pages, platform search visibility, reviews, fulfillment reliability, creator proof, stock discipline, bundles, and pricing tests. Skintific’s record is a marketplace-distribution case first. Paid media can amplify a working shelf; it cannot compensate for a weak one for long.

If the brand depends on delivery or service reliability, fund operations before demand acceleration. J&T’s record makes the point without needing a consumer-ad story. Cross-border growth that breaks the service promise is just paid churn.

Media still belongs in the plan, especially in Malaysia. The market evidence supports awareness-building and exposure. But the first question should be which exposure solves the actual entry bottleneck: local credibility, category education, store discovery, marketplace trust, app installs, trial, or repeat use. Paid media is one possible answer. AI-driven paid media is a narrower answer that needs its own evidence.

A practical allocation frame is therefore sequential. Fund the verified cross-border drivers first. Reserve AI-driven paid ads as a capped test line in the first market, with clean naming, market-specific reporting, and a result that can be defended without borrowing credit from store openings, app growth, marketplace rank, or logistics readiness. Increase the allocation only after the account has enough local conversion signal and the test shows named-market lift on the business metric that matters.

If the planning deck needs a number before the market has stable data, that number should be a test cap chosen from the brand’s own risk tolerance, not a benchmark copied from these cases. The public evidence does not justify making AI-paid ads the first dollar in a Malaysia or Singapore entry plan. It justifies testing automation after the route to purchase is real enough for the algorithm to learn from something other than noise.

References

1. Inside The Indonesian Starbucks Challenger That's Betting On Affordable Premium Coffee, Forbes Asia, Apr 15 2026
2. Kopi Kenangan earns US$17 million profit in 2025 after expansion, IDNFinancials, Jan 30 2026
3. J&T Express Reports 18.5% YoY Revenue Growth for FY2025, PRNewswire
4. How Skintific dominates Southeast Asia's Beauty and Skincare eCommerce, TMO Group, Oct 7 2024
5. Cultural similarities aren't enough: Indonesian brands need a new approach to Malaysia, Campaign Indonesia
6. [Indonesia] State of AI in Marketing, MMA Global