On this page 6 sections
Fixed-price MVP development works when the scope is fixed as a boundary - auth, payments, one core workflow, at most one AI feature, one production deploy - and fails when it’s fixed as a feature list. Price it only after scoping, and a 5-6 week fixed sprint can ship without scope creep.
The industry line is that fixed price never works in software. Agencies that sell hourly work argue it openly, and they have data on their side. But founders keep asking for fixed prices for a good reason: pre-seed runway doesn’t survive an open-ended invoice. What decides it is what has to be true about the scope.
Fix the shape, not the feature list
A feature list looks precise and isn’t. “Users can invite teammates” hides a permissions model, an email provider, invite expiry, and what happens when the invitee already has an account. Nobody prices those correctly from a one-page brief, because nobody knows yet what they are.
Basecamp’s Shape Up names the inversion that fixes this: “Estimates start with a design and end with a number. Appetites start with a number and end with a design.” A fixed-price MVP is an appetite. The price and the calendar are the constraint, and the design is whatever fits inside them.
That only works if the thing being fixed is a boundary. Shape Up’s test for shaped work is that it’s rough, solved, and bounded: “Shaped work indicates what not to do. It tells the team where to stop.” A feature list tells the team what to build. A shape tells them where to stop. Only the second one can carry a fixed price.
For almost every B2B SaaS or AI MVP, the shape is the same five parts:
- Auth. Sign-up, sign-in, one role model. SSO and extra roles wait.
- Payments. One pricing model on a hosted checkout. Stripe Checkout already handles one-time and subscription payments across more than 125 local payment methods, and its subscription quickstart pairs it with a Stripe-hosted customer portal for plan changes and billing details. Building your own billing screens is a phase-two problem.
- One core workflow. The single job the product exists to do, end to end, with real data.
- At most one AI feature. The one that makes the core workflow worth paying for.
- One production deploy. Real hosting, error tracking, a database with backups.
Everything else is a candidate for version two.
Why the “fixed price never works” camp is half right
The opposing view deserves a fair hearing, because most of its evidence is real.
Early estimates are bad. The cone of uncertainty, from Barry Boehm’s work and popularized by Steve McConnell, puts estimates at the start of a project at roughly a factor of 4 off in either direction. Price a feature list at that stage and you’re pricing a guess.
Scope moves. DECODE, an agency that argues for time and materials, cites figures that requirements change around 25% during a typical project and that 52% of projects experience scope creep. It also names the incentive problem: vendors typically add a 15 to 30% risk buffer to fixed quotes, and when a fixed project starts losing money, quality becomes the variable - testing gets compressed and senior engineers get swapped for juniors.
Change orders stall work. Leanware describes the pattern: under a fixed contract, a change pauses work while the agency re-estimates, and the timeline slips two or three weeks. And HouseofMVPs says the quiet part out loud: “Most agencies that claim fixed-price actually run hourly with a fixed-price wrapper, using change orders to add hours mid-project.”
All of that is true of fixed-price feature lists. None of it is an argument for time and materials on a pre-seed budget. The same Leanware piece shows the other side: a build estimated at four months and $70,000 that finishes in six months at $105,000. On hourly billing, every overrun is yours. The honest conclusion is that contracts only decide who carries the scope risk. Scoping discipline is what reduces it.
The scope rules that keep a fixed sprint fixed
These are the rules that let a 5-6 week fixed sprint ship without creep. Each one closes a specific leak.
Price after scoping, never before. Shipkit puts it bluntly: “Without discovery, a fixed-price quote is just a guess.” The fix doesn’t need to be a separate multi-week engagement. It needs the data model, the architecture and the out-of-scope list written down before a line of product code exists.
Write the out-of-scope list first. The in-scope list is easy to agree on. The out-of-scope list is where the arguments hide. “No mobile app, no SSO, no custom reporting, no second AI feature” is the line that protects your launch date.
Swap, don’t add. A new idea mid-sprint is fine. It replaces something of equal size already in scope, or it goes on the post-launch list. That one rule removes most change orders, because nothing grows.
One AI feature, bounded by its evaluation. AI features are where fixed prices go to die, because “make it smarter” has no end. Define the input, the output, and what “good enough to ship” looks like on a small set of real examples. Once it passes, it ships.
Integrations need documented APIs. An integration with a well-documented public API can be scoped. An integration with a partner’s undocumented internal system can’t. The second kind becomes a time-boxed spike or a phase-two item.
The engineer who scopes it builds it. Most fixed-price failures are handoff failures: a salesperson scopes, a PM translates, a junior builds. When the senior engineer who wrote the scope is the one shipping it, the cone of uncertainty narrows fast, because the person making the estimate learns the most from being wrong about it.
What fixed-price MVP development costs in 2026
Published pricing varies widely because “MVP” means different things to different shops. The ranges below come from vendors’ own published numbers.
| Option | Published price | Timeline | Scope risk sits with |
|---|---|---|---|
| Productized fixed-price shop (HouseofMVPs) | $3,999 to $14,999 | 1 to 4 weeks | Vendor, via tight templates |
| Fixed-price agency, simple MVP (Shipkit) | $15,000 to $30,000 | 4 to 6 weeks | Vendor, after discovery |
| Milestone-based agency build (Leanware) | around $70,000 to $80,000 | multi-month | Shared, per milestone |
| Zeroic MVP Sprint | from $12,000 | 5 to 6 weeks | Zeroic, after free scoping calls |
The cheapest options get there by reusing a template, which is fine if your product fits it. The most expensive ones include a separate paid discovery phase, often $5,000 to $15,000 over 2-4 weeks by Leanware’s numbers, before the build is even priced.
How to decide if a fixed-price MVP fits you
Fixed price fits when:
- You can name the one workflow a paying customer would use in week one
- Your AI ambition fits in one feature with a clear input and output
- Your integrations have public, documented APIs
- You’re willing to write the out-of-scope list and hold to it
- Runway is your binding constraint, more than flexibility
Time and materials or a retainer fits when:
- You’re still discovering what the product is, and scope should change weekly
- The core value depends on research-grade AI with no clear “good enough”
- You already have a product in production and need ongoing capacity - that’s what a retainer is for
Red flags in any fixed-price quote:
- A precise price from a one-page brief, with no scoping step
- No written out-of-scope list
- A change-order rate that’s higher than the base rate
- No named engineer - just “the team”
If your MVP passes the first list, a fixed price is the most honest contract you can sign.
How QuizBot shipped on one workflow, then got acquired
Our MVP Sprint is built around the rules above. Scoping happens on free calls before you pay anything: we agree the smallest product that proves your idea and fix the scope and the price. Week 1 goes on the data model and architecture. Weeks 2-5 are the build, and you see working software every week. Week 6 is the production launch, followed by 6 weeks of post-launch support. Senior engineers only, one or two on the project end to end, and you own 100% of the code, infrastructure and credentials.
QuizBot is the shape in practice. The product did one thing: turn any text or URL into a structured quiz. We built the AI pipeline that parsed the input, identified the key concepts and generated questions with plausible wrong answers, plus an API layer so other teams could embed it. One core workflow, one AI feature, shipped. The platform grew to 2,500+ users and was acquired.
ComplyAssist is the same discipline applied to a heavier domain: a regulation graph, task and document layers, one predictive AI layer on top, and role-based access, shipped with a tiered pricing model that scales from a small-team plan to Enterprise. The rest of the MVP work is on /work, and every other engagement shape is on /services.
Frequently asked questions
- Does fixed-price MVP development work?
- It works when the scope is fixed as a shape - auth, payments, one core workflow, at most one AI feature, one production deploy - and the price is set after a short scoping phase, not from a one-page brief. It fails when the contract is a long feature list priced before anyone has designed the data model.
- Is a fixed-price MVP more expensive than hourly?
- Often, yes, on paper. Vendors typically build a 15-30% risk buffer into fixed quotes. The trade is that you stop carrying the overrun risk: on time and materials, a build estimated at four months and $70,000 can finish at six months and $105,000, and that overrun comes straight out of your runway.
- What happens if I need to change the scope mid-sprint?
- In a well-run fixed sprint, a new idea is swapped, not added. If it's more important than something already in scope, the lower-priority item moves to the post-launch list. If it's a different product, it goes into the next phase, usually a retainer or a second sprint, with its own price.
- How long should the scoping phase before a fixed-price MVP take?
- About a week for a typical SaaS MVP. Some agencies run paid discovery phases of 2-4 weeks. For our MVP Sprint, scoping happens on free calls before you pay anything: the scope and price are fixed first, and week 1 goes on the data model and architecture.
- What should never be in a fixed-price MVP?
- Anything whose effort you can't bound before the build starts: a second AI feature, custom ML training, three-way integrations with undocumented APIs, a native mobile app alongside the web app, and admin reporting nobody has specified. Each of those belongs in a later phase or a time-boxed spike.
References
- Shape Up (Basecamp): Set Boundaries
- Shape Up (Basecamp): Principles of Shaping
- Cone of Uncertainty (Wikipedia)
- DECODE: Why time and materials contracts deliver better outcomes for custom software
- Leanware: MVP development cost - fixed price vs milestone contracts
- Shipkit: Fixed-price MVP development in 2026
- HouseofMVPs: Best fixed-price MVP development agencies 2026
- Stripe docs: Build a payments page with Checkout
- Stripe docs: Prebuilt subscription page with Stripe Checkout
Prashant Abbi