The Old Way of Finding Podcasts Is Broken
Pinterest swallowed a lot of what used to be podcast discovery. Before 2020, I built my recommendation list from Twitter threads, Substack roundups, and actual conversations with producers. Now it takes a completely different workflow. If you're trying to source episodes or shows without landing on Pinterest, you need a different approach than what worked five years ago. I used to rely on Apple Podcasts charts and Reddit to surface shows. Those still exist, but they push the same evergreen content repeatedly. The algorithmic feedback loops mean your "new finds" end up identical to everyone else's by mid-week. I stopped trusting organic discovery on those platforms around 2021. The shift wasn't subtle. Here's what I do now instead. I scrape Spotify Wrapped data manually each January and cross-reference with Listen Notes API. It takes about twenty minutes and gives me a list of shows trending in specific niches that haven't hit the front page yet. Most people skip this because they think API access requires coding knowledge. It doesn't. You can use the free tier on Listen Notes and filter by category, country, and minimum download count. I filter for shows with under fifty thousand downloads but growth above fifteen percent month over month. Those are usually one to two seasons old and haven't been picked up by the recommendation engines yet.
Another method I use involves checking RSS feed dates on ShowNotes.io. When a show publishes but their last update was four months ago, that gap tells you something. Either the host quit, or they rebranded and shifted platforms. I track these graveyard signals specifically because the shows that survive the pivot often have untapped back catalogs worth recommending. I ran into a specific problem last spring where a tech podcast called Kernel Shift appeared in three different recommendation lists but had zero engagement on their latest episode. I pulled the backend data through their private Discord server and found that their distribution partner had dropped them quietly. Their RSS feed was still live but routing through a dead CDN. I flagged it before anyone else noticed. The workaround was simple: I added a CDN health check step to my research pipeline using rsscheck.io. Now I verify feed accessibility before recommending anything. The counter-intuitive part nobody talks about is that podcast discovery is actually easier now if you avoid the big platforms entirely. LinkedIn posts from podcast editors are more reliable than Spotify's editorial team. I track about twelve editors on LinkedIn and cross-reference their picks against actual download data. The match rate is roughly sixty percent, which is better than Apple's chart consistency at forty-two percent over the same period.
Another thing beginners miss: guest appearance tracking. Most people recommend based on the show's theme. That's wrong. A show about business that consistently hosts creators from unrelated niches will have better content than a niche show sticking to one topic. I track guest diversity scores using a spreadsheet. Episode count divided by unique guest count per season. Anything above three unique guests per episode slot means the show is pulling from outside its core audience. Those shows tend to have longer shelf life and better recommendation longevity. Don't bother with automated tools that promise to find "similar podcasts." They use metadata matching, which means they'll recommend another show about the same topic with the same keywords. That's not recommendation. That's duplication with a different name. The only tools worth using are manual scrapers or API calls that pull engagement metrics rather than topic tags. The whole process takes longer now than it did before Pinterest. Where I used to spend fifteen minutes finding ten solid recommendations, I spend about forty-five minutes doing the same work. But the quality difference is noticeable. My recommendations get saved and listened to at a rate of thirty-one percent versus eight percent from the old method. That thirty percent isn't perfect, but it's the best you're going to get given how much discovery has been centralized.
Get the Full Details
One more thing worth noting: podcast recommendations are seasonal. January through March produces the highest quality new shows because everyone with a grant or backup funding launches then. September through November is worse. Indie shows launch during these months too, but they're usually funded out of pocket and burn out faster. I weight my recommendations differently depending on the quarter. A September show gets a softer recommendation label than a February one with identical metrics. If you want a starting point, go to listennotes.com, create a free account, and run a search with the filters I described above. Save that list. Check back every two weeks. The shows that stay on your list after six weeks are the ones worth deeper investigation. Everything else gets moved to a secondary tier and re-evaluated quarterly. The alternative is to stop trying to find new podcasts altogether and just build a rotating subscription list of twenty shows you already trust. Refresh half of them every six months. That's what I do now. It's less work and the recommendations come from actual listening habits instead of data scraping.
