Founders often ask what an MVP costs. The honest answer is that the price is decided by a handful of choices, most of which you control. Here are the ones that matter most.
1. The number of user roles
Every type of user (customer, admin, vendor, teacher, student) multiplies screens, permissions and edge cases. An MVP with one user type and a simple admin is far cheaper than a marketplace with three sides. If you can launch with one side and handle the other manually, do it.
2. Payments and money movement
Taking a card payment for a subscription is well understood. Splitting payments between parties, handling refunds and payouts, or holding funds in escrow adds real complexity. On ChefJRP the payment flow was simple on purpose: pay once, get instant access.
3. Real-time features
Live updates, chat and collaborative editing need WebSockets, careful state handling and more testing. NepseView needed live prices because that is the product. If real-time is not the core of yours, polling or a refresh button is fine for version one.
4. AI features
Calling an LLM is easy. Making answers consistent, adding memory and keeping costs under control is not. Budget for evaluation and cost limits from the start. I wrote more about this in adding LLM features to a web app.
5. Custom design
A distinctive design takes time. A clean, consistent interface built from a good component system is much faster and is usually enough to learn whether people want the product.
6. Integrations
Every third-party system (CRMs, accounting tools, shipping providers, government APIs) brings its own documentation, quirks and failure modes. List them early, and question each one.
How to keep the cost down
- Write down the one job the product must do for its first users
- Move everything else to a “later” list, and be strict about it
- Do manual work behind the scenes where you can, and automate it once it hurts
- Choose proven technology, so nothing is built twice
If you want a second opinion on your scope, I offer MVP development and I am happy to review a feature list before you commit to anything.