Podcast App CastBox Took Few Wrong Turns — Making Money Will Be the Hard Part | 100 Entrepreneurs

Years at big companies like Google gave the founding team its first leg up.

Wang Xiaoyu has been in a good mood lately. She founded the podcast app CastBox, and at Google's global developer conference this May, Google product manager Jonathan Karmel recommended it to the whole room, saying he opens it every day.

CastBox is a platform app that collects open radio programs and podcasts from around the world. It is a little like Chinese podcast apps such as Lizhi FM and Ximalaya, except that alongside audio made by individual podcasters, CastBox also pulls in a huge number of radio programs in many languages from everywhere.

It has gathered more than 1 million audio programs — radio shows, audiobooks, talk shows and more — from 150 countries and regions. Nearly all of its users are outside China, and a third of its listeners are in the United States. Wang Xiaoyu says she went overseas first because she thought podcast products in China were already more mature.

The company is only two years old, but CastBox has already picked up more than 5 million users and made Google's annual best-apps list two years running.

Putting Their Last Jobs to Work

Two years is all the time this company has actually existed, yet neither the finish of the product nor the speed with which it grew abroad looks like the work of an inexperienced young team.

Wang Xiaoyu is not a serial entrepreneur. She and her team settled on what to build largely out of what they had picked up at their previous jobs.

In late 2015, when she decided to start the company, she was still at Google, in charge of helping developers turn traffic into money in markets worldwide. She spent four years shuttling between Google's Japan and Ireland offices, dealing with developers day in and day out.

Co-founder Li Shibin, who runs marketing, is Wang Xiaoyu's senior from Peking University. She spent nine years at Google on advertising promotion and monetization, providing AdWords clients around the world with localized promotion and market research.

Those years abroad left Li Shibin thinking that "this combination was born to go global."

Liu Yaqi joined CastBox a year ago. He was a front-end engineer at 360, is a serial entrepreneur, and has eight years of Android development behind him. As co-founder and head of Android development, he thought steering clear of the Chinese market was the smart move.

"Audio products in China compete hard, and the traffic has basically been carved up already." In the end the team agreed to go after the overseas Android market.

Wang Xiaoyu is a goal-driven person. She studied psychology at Peking University and taught herself programming in her spare time, for a simple reason: "because it made money."

In college she took on a string of outsourced projects, mostly small games. One of them reached the top three on Thailand's game charts. She was never a trained programmer, but she built the first version of CastBox in three days.

After several years maintaining developer relations in the Android market, Wang Xiaoyu noticed that Android, unlike the iPhone, had no official podcast app. It had well-regarded aggregators such as Pocket Casts and Stitcher, but no podcast product had ever come to dominate.

The market was sending some encouraging signals too.

In 2014 and 2015, Serial, the true-crime podcast from This American Life, became a genuine hit, downloaded more than 68 million times in under a year.

Winning the Platform's Recommendation

For a new app, an early feature in the app store is the kind of cheap, high-yield promotion every team fights for.

CastBox launched on Google Play in January 2016. According to the app tracking service App Annie, less than a month after launch Google Play was featuring CastBox on its home page in 135 countries, and after that it turned up on regional home pages almost every week. Between 10% and 20% of its downloads came from Google Play's official recommendations.

In 2016, CastBox was named a Best of Google Play global trending app.

Years inside Google, and a close knowledge of its culture and rules, helped Wang Xiaoyu enormously. From its first day CastBox followed Google's official guidance to the letter, down to the small stuff in Material Design — animation, touch feedback, navigation. "So a month after launch we put in a request to be featured, and it went through in a week."

Besides Google's recommendation, the team credits its attention to user feedback and the speed of its iteration for the fast growth.

On its Google Play page, CastBox has collected more than 70,000 reviews, and the team has written a long reply to nearly every one. More than 50,000 users have given the app a perfect score.

"Easy to use" is what most users agree on. In Google Play's back end, CastBox scores above its rivals on "ease of use" and "best podcast product."

Co-founder Liu Yaqi told QDaily: "A user once left a comment asking why we didn't have a certain feature. The next day the same user commented again to say they'd found it — but it really hadn't been there before." The CastBox team had added it in the meantime.

A blind woman in Russia once wrote to CastBox to say the podcasts had no voice tags, so she had to open and play each one to find out what it was. The team stayed up all night and added voice tags to 260,000 podcast episodes.

Having worked with developers all over the world, Wang Xiaoyu thinks Chinese developers going abroad have one trait that looks like the least technical of all — they work hard — and that this is exactly what sets how fast a product improves.

"When I used to email developers in other countries, I'd sometimes get a reply saying they were on vacation, and they'd take three months at a stretch." For Chinese founders that is almost unimaginable.

Recommending Audio the Way Toutiao Recommends News

CastBox really is a podcast product that catches the eye in the Android market: a clean interface, features built around how people actually listen, and a good recommendation system.

The first time you sign up for CastBox, you pick three interests out of more than a dozen categories. CastBox then recommends a batch of related stations, and the more you listen, the closer its suggestions get to your taste.

CastBox's core podcast listeners still lean middle class; they want better audio and can tell the difference. Even among the popular content served up purely by algorithm, the quality is generally high.

Behind the recommendations, CastBox uses a wide-and-deep algorithm that Google open-sourced, with various weighting factors added on top. It ranks stations by subscriptions, open-and-exit counts, listening time and the like, then pushes trending content to users.

The pinned column on the home page each day is left to human editors. Calling it human editing is a stretch — it still runs mostly on data. Editors crawl Twitter for what people are arguing about, pick the day's hot topics, and choose podcast episodes to match. Wang Xiaoyu told QDaily that the team is working on speech recognition that will eventually generate a transcript of every podcast. "That solves the problem of information being hard to get at, and makes podcast text much easier to search."

Algorithms and user experience were the two things Wang Xiaoyu came back to most during the interview. "For a product going overseas today, one technical advantage isn't enough to build a wall around you. It has to be everything at once — iterating fast enough, a business model clear and efficient enough, more and faster and better and cheaper."

The Product Is Good and the First Users Are In. Now How Do They Make Money?

CastBox's business model today is simple: gather traffic, then run ads from ad networks inside the app. They usually appear as banners in prominent spots, or slot into the content stream of each episode. "Five seconds after a user opens the playback screen, the track's cover art flips over into an ad. The user can tap the close button in the top left, or wait 30 seconds for the ad to run before it flips back to the cover."

As a business, though, podcasting is nobody's cash machine. The Wall Street Journal reports that the industry took in just $34 million in ad revenue last year — roughly 1% of what advertisers spend on billboards. More awkward still, most of that money goes to the people making the content; platforms like CastBox get a sliver.

It makes sense. With an audio product, people don't spend much time looking at the screen, which leaves advertisers unsure that ads on a podcast platform do anything. The content makers who do end up making money are always the same few top shows.

An intern at FiveThirtyEight once listened through the top 100 podcasts on iTunes and counted 186 ads, 35% of them from five companies.

Almost all of them were "hard ads" — meaning ads read aloud by the host inside the episode.

"Betterman is a smart asset management platform that saves you time and maximizes your returns... To learn more, go to Betterman.com and add ATP after it."

The tech podcast ATP (Accidental Tech Podcast) runs ads like this in the first four or five minutes of every episode. It is the most typical form of host-read ad: advertisers can measure the effect through the special URL listeners type in.

It isn't very efficient, but this is a podcast after all, and people are still more willing to put money into audio ads like these.

Back to the platforms. Companies that produce no original content of their own have run into revenue trouble before, and there is a cautionary example.

Marco Arment, the developer of Overcast, the best-known podcast aggregator on iOS, is Tumblr's former CTO and a blogger with a large following of his own. In its first year Overcast charged: $4.99 to unlock the app's feature limits, and 14.7% of users paid.

Overcast made $160,000 that first year. In the second, Marco tried asking loyal fans to donate, letting everyone use every feature free — but far too few were willing to give. Recently Overcast switched to paid in-app purchases: free users watch ads, and users who pay $9.90 a year lose the ads and unlock other features.

Even a well-known developer like Marco, who arrives with a crowd of fans, has to fight for his app's revenue. Without content as a core advantage, a platform really does have a hard time making money.

In podcasting, the content makers hold the initiative, and listeners stick to the shows more than to the platform.

CastBox wants a new way to make money too: find local podcasters in overseas markets and push paid audio. The team is signing exclusive deals with a number of good top-tier shows to sell audio for money.

In China, paid audio is sweeping through. Qingting FM bought Gao Xiaosong's paid audio show; Luoji Siwei, a top podcast, parlayed its fame into Dedao. Paid audio has at least been proven to work as a market model, though overseas the question is still open.

Li Shibin says she hopes CastBox can one day be the "YouTube" of audio content.

But remember that YouTube, with 1 billion monthly active users, still isn't profitable. Audio or video, the format changes but how to make money is probably still a headache for every platform company.

Cover image from mobilemarketingwatch