From Twelve Products in Twelve Months to Bannerbear

From Twelve Products in Twelve Months to Talkshow

Jon Yongfook stopped his challenge to build 12 startups in 12 months at his seventh product. The product he chose to focus on then was not an image-generation tool, the forerunner of Bannerbear, but Talkshow, a video message board for remote teams. Only after Talkshow shut down did he take hold of the problem that led to Bannerbear. His founding journey put three abilities to the test in turn: shipping a product, taking a first payment, and keeping customers.

He had already experienced both running a software business and corporate life. From 2013 to 2015 he ran a SaaS called Beatrix, which was earning enough to live on about six months after launch. He then led digital product and design at Aviva, leaving in July 2018 after roughly two and a half years. Regaining the feel of writing code himself and finishing products was an important goal at the time.

The challenge he announced in September 2018 was a mechanism for forcing that goal. Pointing to his habit of stopping products halfway, he decided he would put a result in front of users by the end of every month. When choosing ideas he first defined the customer group and the problem, then assessed whether customers were easy to reach, whether he could charge from the start, and whether he could use the skills he had. By tying a short build deadline to a real public release, he tried to cut down the time spent weighing possibilities only in his head.

During the first four months he launched Zipsell, a platform for selling digital files, Promomatic, a tool for making App Store promotional images, Montage, a wireframing tool, and Tech Jobs Asia, an Asian tech job site. Three of them reached the top spot of the day on Product Hunt and brought in early revenue. In a January 2019 retrospective, though, he judged that he had been too drawn to a culture of building tools other makers would like. He had won quick reactions and praise, but had not built up the customer base and distribution plan needed to grow them into sustainable businesses.

He also started to sort out how he managed several products at once. In June 2019, while focused on Talkshow, he sold Promomatic and Montage to Finland's Respawn Ventures. Promomatic had revenue from its first day, and Montage was still getting steady signups. Handing products that still had potential to another operator was what let him concentrate his time on one business.

Talkshow: Signups Without Retention

But Talkshow, the product he chose to focus on, fell short of expectations. According to a record he published in August 2019, over roughly four months it had 4,750 signups, 680 people who agreed to receive marketing emails, and 7 teams that converted to paid. The more serious problem was that all 7 paid teams churned after the first month. Signup and the first payment happened, but the product gave them no reason to pay again the following month.

User feedback exposed the structure of the problem. For Talkshow to be useful, several team members had to post videos together, and when only one or two of them used it actively, the value on offer was small. Some people were uncomfortable talking to a camera on their own, and recording video in a shared office was a burden. People sympathized with the goal of reducing remote workers' isolation, but the proposed way of using it never settled into daily life. Jon announced the service's shutdown in September 2019.

The criteria he set that month for his next business were far more concrete. The product had to address a problem customers already knew about, and it had to be explainable in one easy sentence. He also had to be able to predict where the first ten customers would come from, and there had to be little the customer needed to keep tending to after adopting it. He especially preferred a service that, once connected, quietly handled work in the background. The burden of Talkshow, which required building a new habit among several people, was reflected in the conditions for the next product.

His time and money were also running thinner. In a later retrospective he said he had prepared two years of savings but began looking seriously at the business's profitability only after spending about a year of it. The field that drew his interest then was automatic image generation. He had a past experience of working at an e-commerce company and of the hassle of making promotional images by hand for every new product. The direction became clear: handle a repetitive task he already knew with software.

Previewmojo and the Pricing Turn

Previewmojo, built in the autumn of 2019, was an early product in that direction. It used set templates to generate the preview image shown when a webpage link is shared. Within days of launch he won his first paying customer on a 9 dollar monthly plan, but that customer cancelled the subscription within days as well. Jon asked why and got no answer, and the experience led him to reconsider pricing and customer segment.

In November 2019 he raised Previewmojo's lowest price from 9 dollars to 49 dollars a month. Existing customers kept their prices, and he changed the standard for what value he should deliver from then on. He moved from a stage of thinking up features an individual developer like himself would pay a small amount for, to looking for work important enough that a company would pay 50 to 100 dollars a month for it. To build a business sustainable enough to include customer support, he reasoned, the economic value of the problem being solved also had to grow.

In early 2020 he changed the name to Bannerbear and built a new marketing site. The application's core features did not change much, and the new name and design brought more visits and interest but did not translate into enough of a sales increase. His monthly recurring revenue for the January and February 2020 stretch was 472 dollars. The narrow use of preview images made it hard to take on the larger automation needs customers had.

The clue to a change of direction came from user requests. Customers wanted more templates, and they wanted to connect Bannerbear to other services such as social-media scheduling tools. At the time Jon had to add templates himself, and implementing connections to outside services one by one would have meant endless development work. In March 2020 he released an editor that let users build their own templates and an API that other programs could call.

The new way of using it was clear. A customer built a base design by placing text and photos, then sent the changing data, such as product name, price, and photo, through the API, and Bannerbear generated image variations. The same feature could produce e-commerce product banners, per-customer email images, and share images per service user. Even if Jon did not implement every use case in advance, customers now had room to connect it to their own work.

Some existing customers left during the transition. Jon kept providing the earlier product's features to existing customers, but the center of development moved to the API. A separately attempted Shopify product also failed to produce results and was shut down. He drew a line at maintaining both everything already built and every sales channel.

No-Code Users and New Distribution

The new growth signal came from no-code users. When the Zapier integration was released around April 2020, users who were not developers began building automations that pulled data from other services and produced images. Some users wrote their own guides and introduced Bannerbear, which was a promotional effect Jon had not expected. A record published on May 18 shows 25 monthly subscribers and 17 separate one-time payments.

Several attempts went into early customer acquisition at the same time. He featured Bannerbear's major overhauls on Product Hunt three times, shared advice in communities, and contacted early users directly for feedback. He also kept piling up posts and newsletters explaining how to use features. The notes that a referral credit scheme did not work and that an affiliate program produced little show he was experimenting with distribution as well.

He also changed his work cycle to run development and promotion side by side. He split his time roughly in half, writing code one week and focusing on promotion the next. The features built and problems solved during the development week became material for the following week's posts and updates, and at the end of the two weeks he reported progress in a newsletter. Promotion was fixed in the schedule rather than left as something to handle with whatever time remained.

The work he considered especially important was documentation. Jon recorded that as tutorials increased, users understood the product better and converted to paid, and that as the quality of the API docs improved, the burden of explaining things to developers shrank. Later the docs included detailed parameters and data examples for each request, plus a collection of request examples aimed at a simpler start. He provided, alongside the product, the explanations customers needed to apply it to their own work.

At the end of 2020 he also reorganized the product's language around customers' purposes. Automation targeted customers who wanted to cut repetitive marketing work with Zapier, while scaling targeted developers and growth teams who wanted to call the API in large volumes. The homepage entry paths and use-case posts were split along these two directions. Even when people bought the same image-generation feature, their expected outcome and usage differed, and the marketing reflected that.

Compounding Growth and a Team

Revenue grew as these fixes accumulated. In June 2020 monthly recurring revenue reached 1,000 dollars, and a growth retrospective published in early 2021 recorded 10,455 dollars. The next growth retrospective, covering up to July of that year, laid out how monthly recurring revenue reached 20,000 dollars. By Jon's account, the first 10,000 dollars took about a year and adding the next 10,000 took about six months.

As the work grew, he adjusted what he handled alone. In May 2021 he hired a freelance writer to help with tutorials, and at the end of that year he added a dedicated content writer to the team. Even at the stage of growing to 20,000 dollars in monthly recurring revenue he still wrote the code himself, but which features to build was increasingly shaped by customer requests. He kept his own ideas on a list and raised their priority when a real customer mentioned the need.

In July 2023 he disclosed the experience of growing to 50,000 dollars in monthly recurring revenue. By then he was also using separate tools that let people try Bannerbear's features directly, such as a free certificate generator and a tool that turns Twitter posts into images, for distribution. Jon explained that even when these tools did not produce an immediate paid conversion, they helped secure outside links and search traffic. He used his ability to build new things to bring customers to the existing business as well.

Direction changes continued after the product grew. In a post from September 2026 he said that while rebuilding Bannerbear V5 he used AI coding agents and also offered an MCP connection that lets agents call Bannerbear's features. Beyond the path where users operate the screen themselves, he widened the path where other software asks for work and receives the result. The central task of automating image work stayed in place, while the way customers reached it kept changing.

Jon reflected that if he were to do the 12 products in 12 months challenge again, he would build several solutions within a particular technology or industry. Launching unrelated products one after another made it hard to carry what he learned into the next attempt. The important change in Bannerbear's path is that his selection criteria grew more concrete with each failure. Avoiding problems that require the whole team's participation, choosing recurring work, and shifting price and usage to match the customers who would buy built up judgments that shaped a business he could sustain.