Anyone putting together a budget for a new digital product almost always ends up running the same search: how much does an MVP cost. The answers you find online range from a few thousand euros to more than one hundred and fifty thousand dollars, and taken at face value they don't help you decide anything.
The gap has a simple explanation. Eric Ries, who popularised the term, describes the MVP as the version of a product that allows a team to collect the maximum amount of validated learning about customers with the least effort. The definition is on the Lean Startup website, and it follows that the cost depends on what you leave out, far more than on the technology or the vendor's rates. If you have to present a figure to management at the start of the year, it pays to begin with the scope and only then ask for quotes.
The price ranges in circulation and how to read them
Almost all published estimates come from software houses, so they should be taken as orders of magnitude rather than independent statistics. For the Italian market, SystemForge puts the spend between €18,000 and €90,000, with timelines of 6 to 20 weeks, and places a simple MVP between €18,000 and €45,000, to be built in anywhere from six weeks to three months. A guide published on Il mio business plan goes as low as €5,000-20,000 for a working digital MVP.
International sources raise the bar. CMARIX and Resourcifi both cite a range from $15,000 to over $150,000; the latter adds that most projects wrap up in two or three months and that costs and timelines depend on scope more than on hourly rates. Softermii breaks it down by type: a simple MVP costs between $30,000 and $55,000 and takes 5-8 weeks, a standard SaaS needs $55,000-140,000 and 8 to 14 weeks of work, and a product with AI features runs from $140,000 to over $300,000 over 3-6 months.
What are the cost items of an MVP
Discovery
This is the part many people would like to skip, and it is also the cheapest. Its purpose is to put in writing who the users are, the hypothesis to test, the metric that will tell you whether it holds up and what stays out. A few hours of workshop are enough if the decision-maker is in the room; it stretches to weeks if every choice goes through three meetings. The resulting document protects both sides, because every feature added verbally during development becomes a budget change.
UX and UI
Wireframes, interface design, a couple of review rounds. The cost grows with the number of screens and with the states of each one (empty list, error, loading), which are often missing from quotes and then reappear once the project is under way. For an MVP, a polished design built on standard components works well: the brand comes through, and custom animations can wait.
Development
This is the heaviest item and the one that varies the most. It covers sign-up and login, the core features and the admin panel for managing users and content. The admin panel is often underestimated: without one, every data correction goes through the developer and costs keep rising even after launch. When the product has to run on iOS and Android, a cross-platform framework such as Flutter or React Native lets you write a single codebase instead of two.
Integrations
Payments, transactional emails, maps, AI models. Every connection has a development cost and a usage cost. Well-documented APIs, such as those from Stripe, email delivery services or OpenAI and Anthropic, can be integrated in predictable timeframes. A corporate ERP or an outdated management system is another story: sparse documentation, data that needs cleaning, test environments that sometimes don't exist.
Infrastructure and recurring costs
This covers the expenses that continue after delivery and that are best budgeted from the start: hosting, domain, third-party services. A mobile app requires developer accounts. The Apple Developer Program costs $99 per membership year, while Google Play Console charges a one-time fee of $25. If you take payments online, every transaction carries a percentage fee plus a fixed amount (current rates are on Stripe's pricing page). AI features add API usage, which grows along with your users: at MVP stage the figures stay modest, but it should be monitored from day one.
A practical tip: have hosting, domain and accounts registered in your company's name, not the vendor's. The day you want to switch partners, it will save you a lengthy handover.
Testing and acceptance
Manual testing of the main flows, checks on real devices, bug fixing, acceptance testing with the people who will use the product. In a six-week project, dedicating one week to this phase is a reasonable proportion. Also ask how long the post-release bug warranty lasts, because the strangest problems surface when real users arrive.
Realistic development timelines
For a small, well-defined scope, six weeks is a sustainable timeframe, in line with the lower end of the estimates above. This is the schedule we follow in our projects:
- Week 1: analysis and wireframes.
- Week 2: design, with two review rounds.
- Weeks 3 and 4: feature development.
- Week 5: admin panel and testing.
- Week 6: acceptance testing and release.
A demo at the end of each week lets whoever commissioned the work see the real state of the product and correct course before a wrong choice becomes expensive. When deadlines slip, the cause is rarely technical. The reasons we see most often are late decisions, content (copy, images, test data) delivered after development has started and features added halfway through. That is why, before signing, it's wise to name one person who can approve screens within a day or two.
Three scope examples: B2B portal, Flutter app, AI SaaS
The examples below are hypothetical, but they reflect the kind of requests we receive. They show how the cost changes as screens, features and platforms vary.
B2B portal for reseller orders
A component distributor wants to find out whether its resellers would order online rather than by email and phone. The MVP includes a private login, a catalogue with a dedicated price list, a cart with order submission and order history. There is a single integration, confirmation emails. The admin panel lets the sales office enable accounts and see what comes in. The metric to set during discovery could be the share of orders received through the portal in the first two months from a pilot group of customers.
Left out are automatic ERP synchronisation (at first the price list is uploaded from a file), multiple discount tiers and the mobile app. We're talking about eight screens and three core features: it's the smallest scope of the three and sits below the lower end of almost every market range.
Flutter app for booking classes
A chain of sports centres wants to test an app for booking classes. It needs a web application for staff and an app for customers, published on the App Store and Google Play and written in Flutter to have a single codebase. There are five features: class calendar, booking with cancellation, membership with online payment, user profile and QR code check-in. Integrations go up to two (payments and push notifications for reminders) and so do the roles, instructor and customer, for a total of around fifteen screens.
A loyalty programme, chat with instructors and statistics for management wait for phase two. Compared with the portal, the cost roughly doubles, because features, interfaces and platforms increase, and developer accounts and store review times come into play.
SaaS with an AI feature
A consulting firm wants to sell other firms a service that reads uploaded contracts and extracts deadlines and key clauses. The heart of the product is a single feature: document upload, extraction through an LLM from OpenAI or Anthropic, human review of the result, export. Around it you need sign-up, account management and a subscription.
Market estimates for AI MVPs are the highest, as we saw with Softermii. To validate an idea like this, though, you rarely need to train anything custom: an existing model called via API, a carefully crafted prompt and a screen where the user corrects the output are enough. That keeps the cost close to a web MVP with one integration, plus API usage. If the product generates content or interacts with people, also check the transparency obligations under the AI Act.
What to postpone without undermining validation
Spending little doesn't achieve much if the starting hypothesis is wrong. A CB Insights analysis of 101 startups, cited by DigitalOcean, identified the lack of a market need as the cause of failure in 42% of cases. If the hypothesis isn't clear yet, it's worth going through a market testing phase before asking for quotes, as we explained when discussing how to validate a digital product before building it.
The rule we use for cutting is this: a feature goes into the MVP only if you can't measure the hypothesis without it. Everything else waits for the data collected from users. These are the items that most often end up on the postponement list.
- The second platform: if your audience works at a desk, start with the web; if they're on the move, go for a mobile-optimised web version or a single app.
- Store publication, when the web version is enough to gather initial feedback.
- Push notifications and marketing automation.
- Granular roles and permissions: at first, user and administrator are enough.
- Full integration with ERP and management systems, which can be replaced for a few months by a file import.
- Advanced reports and dashboards, when a CSV export is sufficient for the chosen metric.
- Multiple languages, customisable themes, animated onboarding.
- Optimisations for traffic volumes you don't have yet.
Some things, on the other hand, are worth investing in from the start. Login and data security, legal requirements (privacy and, if you fall within its scope, the European Accessibility Act) and a minimum of analytics to read the agreed metric. An MVP without usage data is cheap, but it leaves the same questions open as before.
How we set up an MVP at DevInterface
For MVP development we chose a fixed scope at a locked price, and that is what allows us to stay below almost all the ranges above. The Web or Mobile MVP costs €5,000 + VAT and is delivered in six weeks. It starts with a two-hour workshop that produces a Technical Annex listing users, hypothesis, success metric and what stays out. It then includes up to eight pages with your brand, sign-up and login, up to three core features, one integration of your choice among payments, email, maps or AI, the admin panel and a 60-day bug warranty.
The MVP Plus costs €10,000 + VAT, with the same timeline. It adds smartphone apps published on the stores to the web app, goes up to five features, two integrations and fifteen screens, handles multiple user roles and push notifications, and covers bugs for 90 days. In both cases you pay 40% at kickoff, 30% at the midpoint and 30% on acceptance, and the code and rights pass to you on final payment. Hosting, domain, developer accounts and AI service usage remain in your name and at your expense, as we recommended earlier. After launch you can continue with monthly sprints at €5,000 + VAT, up to 80 hours, deciding new features based on the data collected.
Applied to the three examples, the B2B portal fits the first plan and the class booking app the second. The AI SaaS fits the base plan if subscriptions are activated manually in the first few months, while automatic payment requires the second integration included in Plus.
Before talking to us or anyone else, write down on a single page the hypothesis, the metric and the three features without which you can't measure it: with that page in hand, quotes become comparable.
.jpeg)
