Why This Question Gets Vague Answers
Ask five agencies how much it costs to build a web application and four of them will say "it depends." That answer is technically correct and practically useless. You are trying to make a budget decision, plan a fundraise, or evaluate whether to hire a team. You need numbers.
Here are real numbers, based on actual projects, with an honest explanation of what moves the needle in each direction.
Tier 1: Simple Internal Tool or Prototype
Range: $8,000 to $20,000
What fits in this range: a CRUD application with authentication, a dashboard that reads from one or two data sources, an internal admin panel, or an investor-ready prototype with 4 to 6 screens and a basic backend.
What keeps it in this range: a clearly defined scope, no third-party integrations beyond standard auth (Google OAuth, email), a single user role, no real-time features, and a standard deployment on a managed platform like Vercel or Railway.
BeSound's prototype was delivered in 6 weeks at the lower end of this range. The scope was tight: onboarding, content browsing, audio player, and session tracking. Nothing outside the defined brief was built.
Tier 2: SaaS MVP or Serious Internal Platform
Range: $25,000 to $60,000
What fits in this range: a multi-role SaaS application with subscription billing, a customer-facing platform with complex business logic, a data-heavy internal tool with reporting and export, or an application that integrates with 2 to 4 third-party APIs.
What pushes toward the high end: multi-tenancy (data isolation between customers is non-trivial), payment integration (Stripe is manageable, complex billing rules are not), file uploads with processing pipelines, and email notification systems with templates and queuing.
Tier 3: Enterprise Platform
Range: $60,000 to $150,000+
Our CCS platform engagement ran to $77,240 over 1,921 hours. That covered a complete enterprise platform with a React frontend, Go backend, PostgreSQL database, complex role-based access control, data pipelines, and multiple integrations. It was maintained and extended over an extended period, not delivered as a single fixed-scope project.
What defines enterprise work: compliance requirements (HIPAA, SOC 2, GDPR all add engineering time), audit logging, SSO integration (SAML or Azure AD), complex approval workflows, high availability requirements that demand infrastructure work beyond a single server, and the organizational overhead of coordinating across multiple stakeholders.
What Drives Cost Up
- Authentication complexity: Google OAuth is cheap. SAML SSO, multi-factor authentication, and role-based access control with complex permission trees add significant time.
- Third-party integrations: Each integration has its own auth model, rate limits, error handling requirements, and schema quirks. Budget 3 to 5 days per serious integration, more if the API is SOAP-based or poorly documented.
- Real-time features: WebSockets, live updates, and collaborative editing require a fundamentally different architecture than request-response APIs. They are not expensive to build in isolation but change the complexity of everything else.
- Compliance requirements: HIPAA adds encryption, audit logging, data retention policies, and a business associate agreement. Each adds real engineering time and ongoing operational overhead.
- Unclear scope: A project with a well-defined spec costs 20 to 30% less than a project where requirements are discovered during development. The cost of a change in design is cheap. The cost of a change after the database schema is built is not.
What Reduces Cost
- A written specification with wireframes or mockups before development starts
- Reusing proven patterns rather than building from scratch (auth libraries, payment integrations, standard UI components)
- Right-sizing the tech stack: a simple internal tool does not need a microservices architecture
- Phased delivery: shipping an MVP first and adding features based on real usage costs less than trying to build everything upfront
The Real Cost of Getting It Wrong
A $15,000 project delivered by an inexperienced team that needs to be rebuilt is not a $15,000 project. It is a $15,000 sunk cost plus the cost of the rebuild. The total cost of a failed first attempt is always higher than paying for experience upfront.
If you are scoping a web application and want an honest estimate based on your specific requirements, FriendsBit has delivered across all three tiers above. Tell us what you are building and we will give you a real number.