Stack decisions consume a disproportionate amount of a solo maker's time, and almost none of them determine whether the product succeeds.

That is the honest framing. We have data on 466 shipped indie products, and nothing in the stack correlates with a product doing well. What the data does show is which choices are common enough to be well-supported — which matters when you are alone at midnight with a bug and no colleague to ask.

This guide is about picking quickly and getting back to the part that matters.

The Default Stack, According to 466 Products

If you want the short answer: Next.js on Vercel, Tailwind for styling, a hosted Postgres, and Stripe when you start charging. That combination describes a large share of what gets shipped, and choosing it will surprise nobody.

Being unremarkable is the feature. Every problem you hit has been hit by thousands of people, which means it has been answered on Stack Overflow and is well-represented in the training data of whatever AI assistant you are using. That is worth more than any framework's benchmark numbers.

Hosting: Vercel Unless You Have a Reason

Vercel appears on 27% of products — more than React itself. Cloudflare is second at 23 products, then Google Cloud and AWS tied at 9 each.

The gap is instructive. Vercel has 14x the adoption of AWS among these makers. AWS is cheaper at scale and more flexible in every direction, and almost nobody here is at the scale where that matters. Deployment friction costs a solo maker more than hosting bills do.

Pick Vercel if you are building with Next.js and want to stop thinking about deployment.
Pick Cloudflare if you want cheaper egress, edge execution, or you are already using their other services.
Pick a VPS if you have long-running processes, need persistent connections, or genuinely enjoy this part. A $6 VPS running Docker is a perfectly good answer that the data underrepresents.

Framework: The React Gravity Well

Next.js at 104 products against Vue at 5, Svelte at 4, and Angular at 4. Combined, the non-React options total 13 against Next.js alone.

This is not a quality judgement. Svelte is genuinely pleasant, and people who use it tend to be happier than people who use React. But when you are one person, ecosystem size is a practical constraint rather than an aesthetic preference: more libraries, more answered questions, more example code, and better AI assistance because the models have seen far more of it.

Use Next.js if you want the default and the least friction.
Use Astro if your product is content-heavy and mostly static — it appears on 8 products and is excellent at that specific job.
Use Svelte or Vue if you already know them. Do not learn them for a first product; you will spend your limited hours on the framework instead of the product.

Backend: The Firebase and Supabase Question

Firebase leads 65 to 41 overall, which surprises people. But the picture inverts depending on where you look: among Next.js products specifically, Supabase leads 18 to 10, and among AI products they are tied at 16 each.

Firebase's overall lead comes substantially from mobile apps, older projects, and the .NET contingent. Among new web projects, Supabase is winning.

Pick Supabase if you want Postgres, SQL you can reason about, and the ability to leave later. Standard Postgres means your data is portable.
Pick Firebase if you are building mobile-first, want realtime sync without thinking about it, or need auth working in twenty minutes.
Pick a plain Postgres host — Neon, Railway, or your VPS — if you want a normal database and are comfortable wiring up auth yourself.

One note on lock-in: Firebase's data model is proprietary, so migrating away is genuine work. Supabase is Postgres underneath, so leaving is mostly a matter of moving the database. If that difference matters to you, it is the deciding factor.

Auth: The First Hard Thing

Auth is where most solo makers lose their first weekend. The data reflects how people avoid that: Firebase at 65 and Supabase at 41 are both largely auth decisions as much as database ones. Clerk appears on 4 products, Better Auth on 2.

The practical advice: do not build auth yourself. Not because you cannot, but because it is a solved problem with severe consequences for getting it subtly wrong, and it will not differentiate your product.

Take whatever your backend provides. If you are on Supabase or Firebase, use theirs. If you are not, Clerk is the most common standalone option here and handles the tedious parts — sessions, password resets, social login, and the edge cases you will not think of.

Payments: Stripe, Unless You Sell Internationally

Stripe appears on 13 products; Dodo Payments, Paddle, Creem, and Polar together on 6.

The distinction worth understanding is merchant of record. With Stripe, you are the seller, which means you are responsible for collecting and remitting VAT and sales tax in every jurisdiction where you have customers. With Paddle, Dodo, Creem, or Polar, they are the legal seller and handle that entirely.

For a solo maker with international customers, that difference is worth real money in accountant time and real reduction in anxiety. The fees are higher; the alternative is understanding EU VAT rules yourself.

Use Stripe if you are mostly domestic, or you have someone handling tax.
Use a merchant of record if you are selling globally and would rather not become a part-time tax specialist.

Also worth noting: only 19 of 466 products tag any payment processor. Most indie products never reach the point of charging, which says more about the pricing decision than the payment one.

AI: Gemini for Validation

Gemini leads at 46 products, then Claude at 24, with OpenAI and Anthropic's direct API at 10 each.

Gemini's lead is mostly about the free tier. For validating an idea with no revenue, a generous free tier beats marginal quality differences, because the cost of being wrong about a model choice is low and the cost of a surprise bill during validation is not.

The practical approach: build against whatever is cheapest while validating, and keep your provider calls behind a thin wrapper so switching is an afternoon rather than a rewrite. Or use OpenRouter, which appears on 3 products and makes swapping providers a config change.

Things Not Worth Your Time

Microservices. You are one person. A monolith you can hold in your head beats an architecture you have to draw a diagram to explain.

Kubernetes. One product in the entire dataset uses it, and that is roughly the right number for indie products.

Your own auth. Covered above. The consequences of subtle mistakes are severe.

Premature optimisation for scale. The median product here has fewer than ten upvotes. Build for the traffic you have.

Learning a new language for this project. Rust and Go together account for 11 products. Both are excellent. Neither is where you want to be discovering language quirks while also discovering whether anyone wants your product.

The Underrated Choice

.NET appears on 29 products — comparable to TypeScript — and gets almost no attention in indie maker circles.

There is a whole population of makers shipping profitable products in C# who simply do not post about it on X. If that is your background, the data suggests you should use it rather than learning the fashionable stack. Familiarity beats trendiness by a wide margin when you are working alone.

The Short Version

• Default: Next.js, Vercel, Tailwind, Supabase, Stripe. Nobody will question it.

• Use what you already know. Familiarity beats every benchmark.

• Do not build auth. Do not build payments. Do not build a queue.

• Pick a merchant of record if you sell internationally.

• Skip microservices, Kubernetes, and new languages.

• Spend the saved hours on distribution, which is the part that actually determines the outcome.

You can browse what other makers used by tech stack, or see the full numbers in our 2026 indie stack report.