The idea of a podcast platform was on our agenda when I was working at fizy. However, we could not start the project due to different strategic priorities. Later, when I joined the Dergilik team, the project came up again, and I was thrilled to be a part of it. 🥳
First of all, we accomplished the podcast project with my excellent colleagues and managers in different organizations such as development, quality assurance, design, business, operation, security, law, Marcom, etc. So many people made this possible, and it was an excellent chance for me to work with them.🙏🏻
Since we began working on the podcast project, the strategy has evolved. The new plan was to establish a podcast platform under a new brand and to make these podcasts available through the Dergilik and fizy applications. The moderation requirement came due to new regulations during the development phase, so we updated the plan and architecture accordingly. This drawing illustrates the fundamentals of architectural design.
Podcast Studion
First and foremost, we wanted to ensure that the podcast platform was ready. In doing so, we could begin to onboard podcasters, and bring on early adopters to test and give feedback on the product. We needed to consider the whole podcasters' flow; registration as a podcaster, verification of podcast ownership, customer support, etc.
After the necessary verification step, the podcast platform parses the RSS feed and gets all the required metadata and episodes' audio files. After that, we host audio files on our CDN. We developed a switch on/off feature to host audio files or deliver from the source. Hosting audio files was an issue because you have to consider performance and security issues for downloading and hosting in your CDN. After four months of hard work, our team was able to launch Podcast Studion. 🚀
Dergilik App
Following the launch of Podcast Studion, we focused on integrating it with the Dergilik mobile application. The improvements to the Audio Article and the Auto-list that I mentioned in previous articles were incredibly beneficial to us. We've enhanced these features even further for podcasts, resulting in a fully dynamic podcast experience for the Dergilik app. Besides, we fed in all of the analytical data to the podcast platform, we were able to prepare the infrastructure for analytics that podcasters could access from the dashboard of podcaststudion.com.
Review Process
During development, new regulations required the podcast review process, so we had to develop an approval process by our editors. We updated the podcast publishing process, added the approval phase, created the essential roles in podcast studion, and made it possible to listen, search, and approve episodes. This development was more complex because we had to modify an existing flow. In addition, together with our AI team, we established an analysis process with STT on all episodes. The AI platform examines each episode and notifies our system if there is a need to control with moderation.
Problems and Optimization
As Podcasters added new podcasts to the platform, we started to identify problems and optimization opportunities. Podcast hosting providers did not have a specific standard, which began to cause issues. After we solved those issues, we focused on optimization tasks. We did not have any performance issues, but we wanted to optimize the platform before happening. So we focused on our crawling process, optimized for frequency of adding, changing, or removing an episode/podcast. We developed e-tag, hash, and other solutions by addressing these issues in order of priority and optimized platform.
Podcast Analytics Dashboard
Analyzing podcast statistics is critical for podcasters. That's why we created a dashboard to monitor analytics data such as the number of followers, the number of unique listeners to an episode, average listen time on an episode, etc. We also designed a monthly insight/summary newsletter for their podcast.
fizy App
As the final stage, we started the integration process with fizy. We agreed on architecture and delivered all podcasts' metadata information and updates via Kafka, and fizy provided podcast analytics for the dashboard using the same infrastructure. After the fizy team launched the podcast feature to their users, Podcasters now have the opportunity to publish their podcasts on two different platforms and reach millions of people.🥳
