The number you want does not exist until the scope exists. That is not a dodge — it is the whole problem. Two founders can describe the same product in the same sentence and get quotes that differ by 5x, because one of them means a booking form and the other means a multi-tenant platform with role-based permissions and an audit log.

So instead of one number, here are three tiers with what actually sits inside them. Find the row that matches what you are trying to prove, and you have a planning budget you can defend to a board.

Tier 1 — the signal MVP: $35,000–60,000

Six to ten weeks. One core workflow, executed properly. Authentication, billing, a usable admin panel, and the single loop that proves people will pay. No custom integrations, no bespoke reporting, no mobile app.

This tier answers exactly one question: will anyone use this? It is the right budget when you have a hypothesis and no paying customers yet. It is the wrong budget when you already have a signed pilot with an enterprise client who expects SSO.

Tier 2 — the market-ready MVP: $70,000–140,000

Three to five months. Multiple user roles, two or three real integrations, a design system rather than a set of screens, analytics wired to actual events, and engineering practices that will survive a technical review.

Most funded seed-stage products belong here. You are not proving demand any more — you are onboarding customers who will churn if the product feels fragile.

Tier 3 — the funded build: $150,000–350,000+

Five to nine months. Multi-tenancy done properly, granular permissions, data migration from whatever the business runs on today, compliance groundwork, uptime commitments and an on-call rotation.

This is where regulated industries start, and where anyone replacing a system a business already depends on has to start. Trying to do this on a Tier 1 budget is the single most reliable way to burn a round.

The four factors that actually move the number

• Roles and permissions. One user type is cheap. Four user types with different data visibility is not — it multiplies every screen, every endpoint and every test.
• Integrations. Each external system is a build, a sandbox, an error-handling path and a support burden. Budget two to five weeks each, not two days.
• Data gravity. Starting empty is cheap. Migrating years of messy data from an existing system routinely costs more than the features around it.
• Regulated data. Health, financial or identity data raises the bar on encryption, logging, access control and review — typically 20–35% on top of the same feature set.

The line nobody quotes: what it costs to keep running

Infrastructure for an early SaaS is rarely the problem — expect $300–1,500 a month until you have real traffic. The costs that surprise founders are the per-seat and per-event ones: transactional email, SMS, error monitoring, analytics, payment fees, and, increasingly, model inference. Model those before launch, because they scale with usage while your pricing may not.

How to spend less without creating debt

Cutting cost is a scope decision, not a quality decision. These reductions are safe:

• Ship one role first and add the others once the workflow is proven.
• Replace a custom admin panel with an off-the-shelf one for the first six months.
• Use a managed platform instead of bespoke infrastructure until traffic justifies the move.
• Defer the mobile app — a fast, responsive web app buys you a year.

These reductions are not safe: skipping tests, skipping access control, skipping the design system, and skipping documentation. Each one is a loan at a punitive rate, and you repay it in the middle of your next fundraise.

A cheap build is not a smaller version of an expensive build. It is the same build with the invoice deferred to year two.

If you want a number for your specific scope rather than a range, our estimator walks through the same factors above and returns a bracket in a few minutes — and every engagement we take on transfers full IP to you on day one.