I'm developing a Progressive Web App (PWA) that helps users find coffee shops based on various needs, such as quiet places for work or locations with good outlets and romantic atmospheres, utilizing community tags and real-time data. My tech stack includes Next.js 14 with Tailwind CSS for the front end, Supabase for the backend with PostgreSQL and Auth, Cloudinary for image uploads, and the Browser Geolocation API for location services. As I plan to expand beyond Houston, I have several technical questions: 1. For Geolocation, I'm using the browser's geolocation and the Haversine formula for distance calculations. I'm concerned about potential better approaches, especially when users deny location permissions. 2. Regarding scaling to multiple cities, should I set up separate databases per city, use a single database with city filters, or consider a microservices approach? 3. Is PWA still the best choice for location-based apps in 2026? I appreciate the lower friction, but it comes with limited features—is it worth it? 4. I'm currently using Cloudinary's free tier but want advice on image optimization as I scale—what's a better approach? 5. Lastly, how do you suggest tackling the cold-start problem for content in location-based social apps? Should I seed content myself or rely on organic growth? I built this project solo without code reviews, so any technical critique would be highly valued. I'm open to sharing code snippets or diving deeper into specific implementation details.
3 Answers
Honestly, have you thought about using bunny.net for image storage instead of Cloudinary? It'll save you some cash. For your database concerns, I'd recommend starting with a single database and city filters. It keeps things simpler and helps you avoid duplication issues as you expand. You can scale to microservices later on if needed.
I get where you're coming from with the PWA question, but I'd still suggest sticking with it. It offers flexibility and a broader reach than native apps, even with its limitations. If you truly need offline capabilities down the line, you can always pivot.
I checked out your live demo at beanscene.cafe—great start! Your geolocation setup looks solid, but consider using PostGIS with PostgreSQL for better geolocation searches. It might make your life easier, especially as you scale up. As for the cold-start problem, definitely seed some content initially to give users a reason to stay engaged. It's tough to wait for organic growth when you're just starting out!
I agree with seeding initial content. Without it, users might find the app empty and lose interest quickly. Plus, having some tags and reviews ready goes a long way in building trust!
Good call on bunny.net! It's often overlooked but really effective for what you're doing. Keeping a single database is also a smart move, easier to manage long-term.