Build or buy a boilerplate: the decision framework

"Should I build my own boilerplate or buy one?" is the most common question I get from indie hackers starting in 2026. The short answer: it depends on what saves you more time in the next 90 days. This guide is a decision framework. Not to push you one way or the other: to let you decide with criteria. ## What's really inside a boilerplate Before deciding, you need to understand what a modern boilerplate saves you: - Auth (email/password + OAuth + 2FA + recovery) - Multi-tenancy (organizations, roles, invitations) - Payments (Stripe or Polar + webhooks + plan-based roles) - Transactional email (welcome, reset, verify, invitations) - Admin dashboard (users, ban, impersonate, audit log) - Multi-step onboarding - Base UI (Tailwind + shadcn + dark mode) - i18n (at least EN/ES) - Storage (S3 or equivalent) Building all of that from scratch, with tests and no obvious bugs, is **200-400 hours**. That's 1-3 months full-time for a senior dev. ## When building from scratch makes sense - Your product **is** the auth/billing system (unlikely) - You'll sell the project's source and want total control - You have 3+ months with no monetization pressure - You enjoy the challenge and are aware of the opportunity cost - Your stack is so unusual no boilerplate fits If you're hesitating on those 5 points, it's not your case. ## When buying is the obvious call - You want to ship a product, not start a library - Your time is worth more than the boilerplate's price - Validating the idea matters more than polishing the code - You don't have 200h spare for non-differential work - You want to learn modern patterns (Next.js 16, Better Auth, Polar) by reading a working repo ## The simple math Cost to build from scratch: 200h × your hourly rate ($50/h freelance average) = **$10,000**. Cost to buy a boilerplate: **$150-300** + 2-4h of customization. Delta: $9,700-9,850 and 2-3 months. If your argument is "I have no money," it usually means you don't have 200h either. And if you do, spending them on generic auth when you could be validating with customers is the worst investment possible. ## Real exceptions There's a case where building wins: if you **already** have a boilerplate from past projects. You open it, tweak it, ship. You're not paying the 200h now; you paid them 2 years ago. Reuse always wins. Another case: if no available boilerplate fits your stack and the adaptation curve exceeds what you save. Rare, but possible. ## Traps when buying **1. Outdated boilerplate**: if it's unmaintained (last commit 6 months ago) and uses Next.js 13 with Pages Router, don't buy. The debt you save turns into different debt. **2. Boilerplate with 50 features you don't need**: if it ships i18n for 7 languages, video uploads, realtime chat, marketplace, blog, podcast — and you only need auth + billing — you'll spend hours deleting. **3. Boilerplate with no docs**: useless. Buying a closed repo without a starting point is burning money. **4. Boilerplate with an exotic stack**: if it uses Astro + Drizzle + tRPC and you've never touched any of those, the 2 saved months turn into 2 months learning someone else's stack. ## How to evaluate a boilerplate before buying 1. What framework version does it use? (current or previous) 2. When was the last commit? (≤ 30 days) 3. Does it have CLAUDE.md or equivalent for AI? (signal that it's 2026-ready) 4. What's the update policy? (one-time vs subscription) 5. Is there a live demo and/or public sample repo? 6. Active community (Discord, answered GitHub issues)? If 4-5 out of 6 are yes, it's serious. ## Bottom line In 2026, with very specific exceptions, buying a mature well-maintained boilerplate is the obvious call. Your time is worth more validating the idea, talking to customers, and building what actually differentiates you. The fastest builder isn't the one who best codes auth from scratch. It's the one who best picks what to build and what to buy.