Taiwan Unicorn Appier Bets On AI-Driven Software For Digital Marketing

Academics-turned Yu Chih-han and Winnie Lee led Appier to become Taiwan’s first listed unicorn. Now they’re betting on global demand for the marketing firm’s AI insights.

A

s Yu Chih-han navigated a spot in a Boston parking garage in 2010, he knew there was a better way. Years earlier the computer science student had designed AI software for a self-driving car for a university competition. “That’s the moment I felt we had to make AI not just a thing in academia, but more broadly available for business,” says the 43-year-old cofounder and CEO of Taipei-based SaaS firm Appier.

Together with his wife, COO Winnie Lee, they have done just that—representing a new generation of tech talent in Taiwan that has found success outside the island’s mainstay hardware industry. The pair have scaled Appier into a billion-dollar software company (the only others to achieve unicorn status in Taiwan are electric scooter developer Gogoro and software firm 91App). A public offering on the Tokyo Stock Exchange raised $270 million last year, valuing the firm at roughly $1.4 billion.

Now the company heads are eyeing further growth in the U.S. and new ways to expand their product portfolio, says Yu, who spoke with Lee from their office in Taipei. The company specializes in combining machine learning with big data to build a presence in digital marketing—using AI to predict customer behavior and personalize messaging across devices.

The company’s financials have tracked increasing demand for digital marketing services, touted as a high-value approach to improving returns on ad investment and reducing customer turnover. Revenue rose 41% in 2021 to ¥12.7 billion ($111 million) from a year earlier, marking its second straight year of growth. Its operating loss shrank to ¥1.1 billion and Ebitda turned positive for the first time at ¥42 million. And there’s huge potential for further growth: The digital marketing software market reached $57 billion in 2021 and is expected to expand at a CAGR of 19% over the next decade, according to U.S. researcher Grand View Research.

Still, it’s been a bumpy ride for investors. After a strong start—Appier’s shares closed up 19% on their first day of trading March last year—the stock has dropped 43% over the past year to have a market cap of ¥108 billion (as of April 8), much further than the Nikkei 225 index’s 8% decline over the same period. Yu attributes the tumble to “corrections” over six months, while Brady Wang, a Taipei-based analyst with market intelligence firm Counterpoint Research, notes tech stocks worldwide are under pressure from financial market fluctuations. Lee shrugs it off. “Whether or not [Appier] is a unicorn doesn’t matter,” she says. It’s better to be a “dragon,” she adds, because “when the investors invest in you, they’re looking for a company that can bring them returns.”

Appier was advanced from the get-go, says Wang. It was an early mover in AI marketing in Asia and has developed what the analyst calls a coveted database of behavioral patterns. That’s key in helping companies find new sales, predict how customers will act and automate digital campaigns with relevant messaging and buying incentives across devices and multiple channels, including social media and apps. Turning data into insight is important but turning that insight into action will be critical for most companies, Lee said during a media interview last year.

“Advertisers are in desperate need of new ways to target their advertising in the face of the retirement of the cookie,” says Wang, which are now being increasingly blocked by tech products. “Nowadays, consumers often use different devices, such as PCs, smartphones and tablets to access information. However, many precision marketing companies tend to analyze only one device, so it isn’t easy to achieve the benefits,” he says. That edge gives Appier leverage in an increasingly crowded market using AI to drive advertising that includes competition from software giants Adobe and Salesforce.

Yu says the firm’s deep-tech software helps it reach 15 billion users daily across nearly 2 billion mobile devices in Asia, and the firm’s tech generates 51 billion predictions daily. Its biggest markets are Japan, Singapore and Taiwan, with a 1,088-strong client list that includes Carrefour and Google, along with online travel agencies, digital gaming companies and others. Its growth reflects broader trends in Taiwan’s startup scene. Last year, AI and big data firms made up nearly 12% of all startups (retail and wholesale topped the list at 22%), according to PwC’s 2021 Taiwan Startup Ecosystem Survey. Appier had just 700 clients in 2019.

Appier got its start 12 years ago in Malden, Massachusetts, a short drive from Harvard University where Yu was studying for his doctorate in computer science. He shared an apartment with Lee (they had met at Stanford several years earlier while pursuing master’s degrees) and Joe Su, also a postgrad computer science student at Harvard. All three are from Taiwan, says Lee, and were inspired by the American startup culture.

Led by Yu, the trio brainstormed at their dining table on ways to commercialize AI in a mass market. They had nine ideas all told and started a gaming company called Plaxie in 2010 that used AI to control an avatar when the player went offline. But the trio found it hard to monetize Plaxie’s technology. “We don’t give up easily,” Lee recalls. They pivoted to digital marketing and baking AI into big data to help companies better understand customers. After graduating Yu returned to Taiwan and set up Appier in 2012, joined by Su as a cofounder and chief technology officer and Lee who had just finished her Ph.D. in immunology at Washington University in St. Louis. For startup capital, each put between $100,000 and $150,000 of their own money into the venture.

Lee, 41, who debuted on Forbes Asia’s Power Businesswomen list last year, at first did “random things” for Appier including recruitment. Her studies had nothing to do with AI, but she found synergy. “Coming from a research background where I constantly studied novel genes, I have an ability to be resilient,” she says. “It’s okay when your hypothesis goes wrong, because that’s part of the experiment.”

Fueled by venture capital raised over the next seven years, Appier expanded outside Asia, delving ever deeper into AI. Sequoia Capital India became its first investor with $6 million in 2014, Yu says, and it was notably the fund’s first investment in Taiwan. Several more funding rounds followed that attracted the likes of Jafco, SoftBank and UMC Capital, among others. In total the company racked up $162 million in funding before its IPO in Japan, following its aggressive expansion there. It was the first Taiwanese company to list there in over 20 years.

The company specializes in combining machine learning with big data to build a presence in digital marketing.

The capital raise went toward developing new products and investing in talent. Nearly a fifth of its some-570 employees are in sales, says Yu, and they spend anywhere from six weeks to six months pitching clients, including those who manage marketing budgets. “All these decisions and stakeholders need to be satisfied in order to move forward,” he says. Appier aims to grow revenue 38% to ¥17.5 billion this year—while Ebitda is projected to increase nearly 1,270% to ¥575 million. The company sees higher demand in the U.S. and is also targeting investment there to prepare servers and inventory capacity. While the U.S. only contributes about 4% to Appier’s top line, it saw 50% quarter-on-quarter growth over the past three quarters, Yu says.

Last May, the company acquired Taiwan-based conversational AI chatbox BotBonnie for an undisclosed amount, following its purchase of Japanese AI startup Emotion Intelligence in 2019 and Indian content marketing company QGraph a year earlier. Still, Yu doesn’t see M&A as a major driver of future business, rather it’s harnessing new technologies that mirror the human brain’s ability to learn from experience. “If we can achieve that, then I think [artificial] intelligence can evolve by itself,” he says. “We don’t have to do a lot of programming across different tasks.”