2-Sided Marketplaces, Platform Risk, and Runaway Apps
Plus my thoughts on pricing in a 2-sided marketplace
The last couple of weeks weren’t great for several of my products.
Let’s start with the elephant in the room:
Not great, Bob.
I’m still trying to figure out what I’m going to do going forward. I’ve got a ton of developer accounts that I used to do a number of things, but the vast majority of those will likely be shut down.
Depends on the monthly cost vs the allowed access. But more and more, I’m feeling like it will be unlikely that I’ll be doing much with Twitter in the near future.
And that brings me to the first topic:
Platform Risk
Is it worth it to build on a platform? The simple answer is that there isn’t a simple answer. It really depends on your goals.
Let’s start with the benefits of building on a platform.
Existing user base - there’s a lot to be said for having a pre-existing customer pool to choose from. As I was building on Twitter, I was spending time on Twitter and seeing what people on Twitter wanted. There’s a lot of value there.
Established models - It’s nice to not have to think about what to name things (we all know that naming things is one of the hardest things in Computer Science) With building on a platform, you know what you have to work with: Tweets, Users, etc.
Examples - Something that I don’t think most people appreciate enough: having examples to work from. When you build on a platform, there are lots of other people building on that platform too, and they share examples of how to do certain things. That tends to make development faster and you can likely find someone to ask if you need help.
Now for the cons.
Access can be pulled away at any time. Do something the platform didn’t like? Buh-bye!
Price changes. Have that business model figured out? Think again!
Platforms can die. Not sure if Twitter will survive its Elongation, but that risk is real. Time will tell. What then? I hope you have a backup plan.
I’ve been focusing the majority of my attention away from Twitter. My passion for Twitter hasn’t been there since Elon took over. It’s been too volatile, so I’ve been thinking about all sorts of safer options.
Like 2-sided marketplaces.
2-Sided Marketplaces
2-sided marketplaces are notoriously hard. It comes down to trying to keep 2 sides happy at the same time because if one side is unhappy and leaves, the other side leaves too.
If I were only working on the marketplace, I might not move forward, but podcasts are an interesting model, because I don’t need both sides to start. I can actually offer benefits to both sides individually, and then when they come together in the marketplace, magic can happen.
The key that I’m counting on for success: RSS feeds. These are a gold mine, because every single podcast has one. It’s an open standard that they are built on, and that means I can grab any podcast’s RSS feed and get access to the information I need. When I first started, I was going to focus solely on the podcasters, because I thought my tools might be beneficial. But that didn’t go as well as I wanted, because the biggest selling point was searching podcasts, but I didn’t have a place where that happened yet. So instead, I decided to seed the marketplace by grabbing a bunch of podcasts and processing them, so I can start trying to get traffic to the site. If I can go to podcasters and say “Hey, there are already people searching here and finding you, here’s some data you could get access to and offer them ways to enhance their listeners’ experiences, that becomes a much more enticing offer.
I’m still working on the business model, but I’m looking at offering multiple ways for people to pay for the same product, depending on their needs. I’m thinking about it this way:
The Personal Plan - the podcast listener. Needs a limited number of searches per month, should be affordable. Initial Price Point- $10/month, $100/year
The Podcaster Plan - The actual podcaster. They benefit from offering search to their listeners, so instead of paying for their own search, they can pay to offer free search to their audience members. I’m thinking that this will be mostly larger podcasts interested, so setting the initial pricing at $50/month, $500/year.
The Podcast Network Plan - They manage groups of podcasts. They might be interested in offering the podcaster plan for all of their managed podcasts. AKA the “Enterprise Podcast” plan. Pricing custom based on needs.
Then I can continue to add in some small offerings for podcasters alongside the search, but the search is the main focus.
If you want to see what the search app looks like, check it out here. Still polishing things up a little bit, so if you see weird pieces, those are probably part of the template I used and still need to update.
Runaway Apps
I had a fun notification last week. I check my email, and see that I’ve got a notice from OpenAI. Turns out, as of Feb 6, I’d already hit the soft spending limit that I’d set. It seems that OpenAI’s requests were failing and my app was deciding to retry them over and over and over again (this part still confuses me honestly. I couldn’t find anything that said it should retry them if they failed). I was still running tests for the Who Followed Me? app that summarizes new Twitter followers’ profiles for you and it somehow just kept repeating the requests, taking my daily spend from ~$1.50/day to $60/day.
Between that and the news of the new Twitter API charges, I just turned it off for now. I’m not sure if I’ll bring it back or not, but not worth the costs. I think there’s some value there, but I’m stepping away from it until things with Twitter settle down a bit.
On the other hand, I did discover a new service: Helicone. They help track your OpenAI usage and were really easy to integrate into the app. Now, I can get much better visibility into the requests than the system I had created, which didn’t help a ton. I had a chat with one of the founders and really like where they are going with things.
Miscellaneous Notes
One other cool thing I’ve discovered: Chroma. It’s an open-source vector database that makes a lot of things easy (like how I’m creating search for podcasts). I’m still using Pinecone, but I chatted with the founder and might end up switching over at some point. They just got V1 released and I’m excited to see where it ends up going.