Arum Kang and the trouble with too much choice

Thirty million dollars is an inconvenient number to have attached to your name. People tend to remember it before they remember anything else. For Arum Kang, co-founder of Coffee Meets Bagel, it became the sum that followed her out of a television studio and into years of conversations about a business she still wanted to build.

In 2015, Kang and her sisters, Soo and Dawoon, appeared on Shark Tank. They wanted $500,000 for a five percent stake in their dating company. Mark Cuban raised the possibility of buying the whole business for $30 million. They declined. A proposal to finance their next chapter had become a proposal to close the book.

Their original product offered an unusually modest answer. One introduction, at noon, each day. A person to consider without an entire catalogue competing for attention. The business that turned down more money had begun by asking its customers to consider fewer people. There is a certain consistency in that.

Three sisters before there was a startup

Kang grew up in South Korea with entrepreneurship close enough to watch across the dinner table. Her father ran a metal recycling business with his brother. Building a company alongside family was already a working arrangement in the household, long before the daughters tried it themselves.

Their parents sent the sisters to the United States for school. Arum and Dawoon, her twin, attended high school in Hawaii; a family photograph places them at their Honolulu graduation in 2000. Distance from their parents made the sisters more independent and more reliant on one another. The future founding team had years of practice before it needed a business card.

The paths eventually separated. Arum and Dawoon attended the University of Pennsylvania. Soo studied at Parsons and worked in design. Arum built experience in consumer businesses, including Avon and Amazon, then entered Harvard Business School in 2009. Dawoon’s route took her to Stanford’s business school and J.P. Morgan.

A shopping idea, with rather different stakes

The spark came from an unexpected neighbour of romance: online retail. Kang noticed how Gilt Groupe curated a small selection and gave customers something fresh to anticipate each day. She wondered whether that rhythm could work for dating. A regular appointment might make an introduction feel welcome rather than make searching feel endless.

In an early interview, she described the appeal in practical terms: members could spend about a minute on the service each day. The first version matched people through their Facebook networks. It had a daily delivery time and a limited decision window. Its simplicity was a product decision, rather than an apology for a missing feature.

Even the name tried to take the temperature down. Coffee was something people enjoyed during a break. Bagels belonged to the New York setting where the company began. Together they suggested an ordinary pleasure. A dating business could have chosen a name promising destiny. The sisters chose breakfast, an event with a somewhat better record of arriving on schedule.

First, find someone who can build it

The credentials did not produce software by themselves. After graduating from Harvard in 2011, Kang began working on the company with Soo. Neither was a technical co-founder. The task of finding somebody to build the product became an education of its own, involving meetups, flyers, conversations and prospective collaborators who never joined.

Kang’s March 2012 account of those months is refreshingly specific. The sisters used a developer found through oDesk to create an initial product. A technical advisor helped assess candidates and deal with problems. They launched a private beta on Valentine’s Day. Romance, at this stage, depended on testing contractors and asking better questions.

Advisors suggested extra features. Kang and Soo put many of those suggestions aside so they could get a simple service into users’ hands. Kang also discovered that understanding the product’s construction could not be delegated completely. If she did not ask enough questions, misunderstandings surfaced later. Knowing the customer was useful; knowing what the developer meant was useful too.

That early essay makes the work sound available to a novice without making it sound effortless. Its advice is about small tests, technical help and meeting people. The founder who would later reject a famous investor first had to persuade less famous strangers to spend time on her idea.

The introduction was only the beginning

Coffee Meets Bagel launched publicly in New York in April 2012. Its first experience reached beyond the match itself. Mutual interest opened a private texting line for seven days. Local venues could offer a complimentary item, such as a cheese plate or prosecco, to give the pair a reason to meet.

This was a curious blend of social matching and the daily deals business. It also showed what Kang meant by paying attention to the whole dating experience. Two profiles expressing interest still left plenty of work: start talking, suggest somewhere, choose a time, actually arrive. The service tried to make the next step easier.

The early virtual currency was called coffee beans. Users could buy or earn beans and spend them on features, including revealing shared friends. These details belong to the launch-era product, which has changed since. They show a founder testing how a small daily ritual could support a business without requiring an enormous daily search.

By autumn 2012, the team had moved to San Francisco and raised $600,000 in seed funding led by Lightbank. Asked about the largest obstacle and then the plan for the money, Kang gave substantially the same answer: building a team. The app was about introductions. So was much of the work behind it.

A television audience has opinions

Three years later, the sisters’ television pitch made their business a public argument. The proposed $30 million buyout became the defining anecdote. Some viewers sent hostile messages after the refusal. People who had never helped build the company were suddenly quite certain about how its founders ought to sell it.

Arum pointed out the gendered language of the response. In the 2015 follow-up interview, she suggested male founders might have been described as bold, visionary or brave for refusing. Her remark challenged the audience’s choice of adjectives as much as its preferred financial outcome.

The sisters kept building. By early 2016, Coffee Meets Bagel had raised $11.2 million in venture backing and appeared in a ranking of MBA-founded startups. That figure offers a useful sense of what followed the broadcast. It is company financing, not Kang’s personal wealth, and it does not settle the counterfactual of what a sale might have produced.

Learning to hear the room

Kang’s later leadership stories include decisions she wished she had handled differently. In 2018, she recalled two early employees becoming fathers before the company had a parental leave policy. They said they could manage without special arrangements. She accepted that answer, then later regretted failing to set a better example.

The experience sharpened her argument for a varied team. Different backgrounds reveal needs that somebody else might overlook. In a business serving a varied group of customers, that changes what gets built. Her point came from an omission she could identify, rather than a general declaration about the virtues of inclusion.

She has also described getting discouraged by rejection early on and giving up after a first refusal. Over time, she learned to keep useful conversations going. Some eventual hires and advisors came from that persistence. A first response could contain information, rather than serve as the final verdict on an idea.

Asked which skill she wanted to improve, she chose listening. It is a small answer for someone whose work had required so much pitching. Yet the two activities fit together. Convincing people becomes difficult when you are too occupied with your own agenda to hear what they need.

“Listening.”Arum Kang, on the skill she wanted to improve

Fewer days, more trust

The work arrangement changed too. Coffee Meets Bagel became remote and adopted a four-day workweek. Kang’s account of this shift puts trust near the centre: a distributed team needs room to manage its projects and time. Giving employees autonomy also meant surrendering some of the familiar signs that a manager was in charge.

There is a parallel with the product. A full calendar can look productive, just as a full screen of profiles can look promising. Neither tells you much about what happens next. In both cases, the harder question concerns the result people came for. How much activity helps, and when does another round of activity simply get in the way?

Her public advice has become more interested in that distinction. She urges founders to establish product-market fit before spending heavily on expansion. She values peers who have already encountered similar business problems. She also says she would have assembled a capable leadership team earlier. The company needed people whose expertise could guide her.

The next choice was to hand over

In 2025, Kang moved into an advisor role. Coffee Meets Bagel appointed Quincy Yang to lead the United States and Canada, while Shn Juay led its other markets. The change divided operating leadership by region and gave the co-founder a different relationship to the company she had spent years running.

In her public handoff message, Kang praised Yang’s financial discipline and analytical thinking. What made her confident in his leadership, she added, was his empathy and commitment to learning. She said she would remain involved as an advisor and would share more about personal projects later.

Declining to sell and deciding to stop being CEO can occupy the same life. Each involves a different question about control. A founder can want a company to continue without wanting every decision to remain hers. The skills Kang had once wished to find in a leadership team were now part of her explanation for handing responsibility to someone else.

Four days to leave the screen

By April 2026, Kang was still talking publicly about the problem that had drawn her into dating. Coffee Meets Bagel’s analysis suggested its Gen Z users exchanged contact information roughly four days after matching, compared with five for millennials. That measures exchanging details, rather than proving a date occurred. The distinction matters.

The company also reported that 92 percent of surveyed users sought marriage or a serious long-term relationship. These are figures about its own community, rather than a census of a generation. Still, they fit a familiar ambition: help people find someone worth meeting, and make it easier to leave the screen.

In Orinda, where she described living with her husband and two sons in 2023, Kang found something she had enjoyed growing up: neighbours who knew one another, children playing together, a community close at hand. She had once insisted she would never leave the city. A backyard proved a persuasive counterargument.

The famous number in her story remains $30 million. The other numbers are quieter: one match, noon, a handful of days before exchanging details. They concern the scale at which people can pay attention to one another. Kang built a company around that problem, then learned that running it required attention of much the same kind.