The Product Is the Marketing

I had a coffee chat with a friend who started a company in Silicon Valley and built a service used by 3 million people around the world. They've built and scrapped more than 40 products so far and are still running hard. They were back in Korea for a bit, so we met up, and I learned a lot from what they've figured out through all the struggle and from the way they keep taking on new challenges.

The biggest thing I took away from the conversation: in B2C today, you have to design the go-to-market (GTM) and marketing into the product from the moment you start planning it. No matter how good a product is, it means nothing if the GTM doesn't work. My friend got to 3 million users with zero dollars spent on marketing. They didn't put extra effort into marketing. The product itself was simply built to spread on its own. Here's what we talked about, broken down by topic.

1. The product is the marketing

My friend's first service was a casual chatbot. In 2024, in the early days of the GPT API, they launched a service that let non-developers easily build their own chatbots. Think a bot that imitates a particular celebrity's way of talking, or a bot that writes in the style of a government report.

It ended up with 3 million cumulative users. Marketing spend was zero, and all the traffic came from SEO. The secret: Jang Wonyoung (a member of the K-pop group IVE) mentioned it on You Quiz on the Block, a popular Korean talk show, and on her livestream, and demoed the website herself. That's a promotion worth hundreds of thousands of dollars, for free. And the same viral moment played out the same way in other non-English-speaking countries too: Brazil, Turkey, Portugal, India. Back then GPT wasn't good at capturing each country's cultural and linguistic quirks (the way people talk, the nuance), and this service got that part right.

So celebrities and influencers in each country showed off the bots built under their own names, and it spread naturally as they promoted it on their own. It wasn't a strategy they planned from day one, but looking back, the structure of the product is what made it go viral.

It's getting harder and harder to get a B2C product in front of people. Seeing this case, I felt strongly that designing and planning a product so the product itself becomes the marketing is the more fundamental work.

2. Casual is the bait, office work is the revenue

After seeing the huge traffic growth on the casual chatbots, they naturally expanded into office use. They launched a second service alongside it: drop in a Word or HWP file (HWP is the Hangul word processor format that Korea's government and schools run on) and AI organizes it for you. On the Word side, Gemini and GPT handled it directly within three months and that business faded right away. But AI didn't support HWP for more than two and a half years, so it kept bringing in steady revenue in the Korean market. It's a case where the lag between Korea and the US in adopting AI created an opportunity in a local market.

The revenue structure was the interesting part. 99% of actual revenue came from office use. The main users were office workers, government employees and teachers, using it for real work like writing student records or internal reports. The casual bots contributed only 1% of revenue. But here's the key: almost nobody came in directly for the office bots. Everyone came in through the casual bots, then discovered the office bots inside and converted. Without anyone intending it, the casual bots became the bait at the top of the funnel.

This is probably the kind of insight you only get by living through it. So their goal is to keep reproducing it going forward. What if we tried to copy it, though? I think the problem might be less about technology or planning and more about whether we have the patience for it. Over the long run, the picture is to gather traffic and then monetize it, but we're always in a hurry, so we start things and then shut them down. What's the answer? I still don't really know.

The more important insight is that the casual users and the office users turned out to be the same persona. If the people using the casual bots and the people using the office bots had been completely different, this funnel wouldn't have worked. But we're all human, and nobody works 24 hours a day. An office worker can play around with a fun chatbot after work, and if that same person goes to work the next day and finds the office features on the same platform, conversion happens naturally. I thought that if you design with a person's whole life in mind, more broadly, you could probably build funnels like this better.

3. Quietly launching 40 products

Right now my friend's main business is AI data solutions: collecting and processing data. They have a fair number of customers in the US, and it's been a stable main source of revenue so far. The company's core vision is to give everyone technical power through AI, and if you want to use AI well you have to collect data, so demand has naturally grown. But they know this business won't last forever. Given how fast AI is moving, there's no telling when something that's working now will get replaced.

So their strategy is to keep building new products, put them online quietly without any announcement, watch how people react, and grow only the ones that survive. They've built more than 40 products so far, and the survivors make up the current business. The rest died or are still being built up. They said that with so many AI products out there now, getting anything to take off is getting harder, but they're not stopping. They keep putting new things out.

4. A product without GTM is meaningless

AI has lowered the bar for development, and now anyone can build a product. So competition has exploded. In this environment, thinking about marketing after you've built the product is already too late.

My friend's casual bots didn't reach 3 million users because they went all in on marketing. They got there because the product itself was built in a way that made people want to share it on their own.

You have to plan the product with GTM and marketing in mind from the very start. Without that, it doesn't matter how good the product is. Building a product with a structure that lets it spread is itself part of building a good product. I felt that strongly today, and it's given me a lot more to think about.

I'm not an AI expert, but I'm always up for a coffee chat. If you'd like to share your story or hear more about other people's experiences, reach out on LinkedIn.

Originally written in Korean on Brunch, March 19, 2026.