Podcasts are exploding. There are now more than 700,000 podcasts available spanning more than 155 countries and over 100 languages. Many of those shows are of average quality — but there are a lot of companies now focused on producing high-quality audio experiences. Companies like Gimlet Media, Wondery, Pineapple Street Media, and even my own startup Lawson Media, are focused on delivering highly engineered podcasts that people will actually want to listen to. These types of productions are fuelling significant growth in the industry but they still make up a relatively small portion of the overall market.

Any person can create a podcast — but not every person can create a good podcast.

Producing high-quality content takes time and energy that many people aren’t willing to spend. Those people often turn to apps like Anchor that allow users to create podcasts quickly and easily, but aren’t as focused on helping users create engaging shows that will gain a meaningful audience. In fact Anchor claim their service is used in the creation of around 40% of all new podcasts. If that’s true it would mean there’s an awful lot of low-quality podcasts flooding the market.

We also know that there is significant awareness of the term ‘podcasting’. In Australia for instance, Edison’s Infinite Dial research indicates that 83% of people have heard of podcasting. However only 30% have actually listened to a podcast. In the US 70% are aware of podcasting, with 51% actually giving them a try. But when you look at monthly listening figures — in Australia 22% of people listen every month with 32% listening each month in the US. That gap between people hearing about podcasts, giving them a try, and then converting into monthly listeners indicates a problem. Why are a lot of people hearing about podcasts but not trying them? And why are people trying podcasts, but not sticking with them?

The Discovery Problem

There’s a lot of discussion in podcast circles about the issues that exist with podcast discovery. Every conference or podcasting event I attend is filled with people talking about how difficult it is to get your show out to audiences when there’s so much noise that exists in the space. One possible reason for the gap between people trying podcasts and turning into regular listeners might be due to discovery. What if those people aren’t finding shows they actually want to listen to?

Apple control the majority of the podcast market through the Apple Podcasts app and third-party providers who feed off their database — however they are notoriously bad when it comes to podcast discovery. The categorisation of podcasts is limited to some key areas like ‘Business’, ‘Games & Hobbies’, or ‘TV & Movies’.

The search function in Apple Podcasts seems to work largely based on a title search rather than a process that helps surface relevant content on a topic. There is also a ‘Browse’ section of the app — where Apple feature a selection of podcasts which are curated by a small team of individuals. Podcast hosting companies are often the gatekeepers to make sure the Apple team isn’t bombarded by individuals who think their podcast is deserving. I’ve got a great relationship with the team at Apple — however the team there is small and always under pressure from companies to feature their shows. Apple also has dozens of different stores for different regions of the world so it’s hard to stay on top of all of that with a small team.

Other apps like Spotify and Pandora are working hard at solving the discovery issue. Spotify has in the past been very selective with shows that appear on their platform and are working on improving discoverability by putting podcasts on a level playing field as music . Although with the purchase of Anchor the quality of shows on Spotify is definitely changing, and not in a good way.

Pandora are working on the Podcast Genome Project which is their attempt at improving recommendations using both machine learning and human curation. They claim to be analysing more than 1500 data points and using that data to surface shows, then mixing in some human curation to add more metadata to the podcasts. The podcast genome project is, I think, the most likely to yield better recommendation results — however Pandora is a closed system available to US audiences — and if you’re outside the US you are out of luck.

The PodFinder Approach

PodFinder takes a platform agnostic view to our podcast discovery service by connecting listeners with high-quality, curated, shows through a Facebook messenger chatbot. Listeners can search by topic or keyword and PodFinder will make recommendations of shows worth listening to. If users are interested in a particular show they can find out more information or listen to a preview directly in messenger. Then when they’re ready to subscribe they can listen in their favourite podcast app.

PodFinder is a search engine for the best of the podcast industry. As we curate content our team adds a significant amount of additional metadata to aid in the future discovery process. Our platform helps direct new listeners toward major categories however users have the ability to freeform questions and our conversational UI will respond if we have shows that match what they’re looking for.

Our initial prototype development was supported by an innovation grant from The Walkley Foundation for Journalism, and we plan to monetise using a paid self-service advertising system which can scale with our user base.

Why Human Curation?

Podcast listeners will often find podcasts via word of mouth or through recommendations from other shows. That personal recommendation is really important to whether someone will consider a show good enough to actually try out. We know from the data that there is a gap between people trying out a show and becoming regular podcast listeners. We also know that there are large Facebook and Reddit communities with many people asking for the same recommendations over and over because they’re struggling to find new content to listen to within the apps themselves.

At PodFinder we don’t want new listeners to have a bad experience with their podcasts. We don’t want them having to wade through the 700,000 shows available trying to find the gems that they’ll actually enjoy. While other services might use an AI system to try and surface content — these systems don’t actually tell you whether a show has great production quality, or an engaging storyline. These are the types of elements that only a human can determine.

Listeners will not find every show on PodFinder but you can trust that with every show you do discover you’ll have a great listening experience. The more people who try out the service — the better the recommendations become.

Our Longterm Vision

We’ve identified a bunch of key indicators which contribute to whether or not a show will actually be good. These are things like storyline, producers, host, levels, and artwork. Some of these can be identified using AI and others require a human touch. As our technology evolves our team will work hand-in-hand with an AI system to cut through all the noise and surface shows which perhaps you may not have heard of.

We also plan to bring our technology to other platforms — like Google Home and Amazon Alexa. As more and more of our lives involve the use of a voice assistant — we recognise that for voice platforms there is a huge opportunity to help people discover great podcast content. Existing systems rely on you actually knowing what a show name is first before you can listen — however PodFinder makes discovering new content, simple.

Where can I try the PodFinder beta?

The PodFinder beta is available now on Facebook Messenger. Visit podfinder.com or search for PodFinder in Facebook Messenger.

N.B — This is a beta and there will be bugs. If you discover a bug just type ‘bug report’ into Facebook messenger and you can send us the details.