How much does it cost to develop a web system in 2026? Real ranges by company size

Every week we get a CEO or IT director on the phone asking "how much does a web system cost?". The honest answer is awkward: we can't answer properly without understanding what you need. But that doesn't prevent you from showing up better prepared for the conversation.

In this article we open real ranges we practice in 2026, with everything that weighs on the budget — code, design, infrastructure, integrations, training, maintenance. No vague market average, no "it depends on scope". You leave knowing what order of magnitude to think about and how to handle the next conversation with any vendor.

"How much does it cost" is the wrong question — and why this matters to you

When someone asks "how much does a web system cost", the question hides at least six variables. Each one can multiply the budget by 5 or 10. Asking for a number before aligning those variables is unfair to the vendor and dangerous to you — you'll hear the number they think you want to hear, not the real number.

• What type of system? A marketplace MVP and an internal ERP cost orders of magnitude apart.
• What's the expected volume? System for 50 users vs 50,000 users completely changes the architecture.
• How many integrations? PIX, existing ERP, WhatsApp, external marketplace — each one costs in hours and in risk.
• Custom design or ready-made components? System designed from scratch vs using Material Design / Bootstrap changes the UI budget by up to 30%.
• Does your internal team collaborate? If you have product, design and QA in-house, the external team needed is smaller.
• Is the deadline tight? A larger team working in parallel costs more — non-linearly.

The good news: it's possible to nail ranges if you can name the project category. That's what we do next, with real numbers we practice today and that we see colleagues in the sector practicing in Brazil.

Real ranges by project type (2026)

Values practiced by Reiko and sector colleagues on custom projects with a qualified team in Brazil. Everything already includes design, development, QA, initial infrastructure and 30 days post-launch warranty. Does not include ongoing maintenance — that goes in a block further ahead. (Values converted to USD at approximate 5:1 rate — Brazilian Real prices in parentheses for transparency.)

MVP / initial SaaS — USD 6k to USD 16k (R$ 30k–80k)

To validate an idea or test a small product: 4 to 8 screens, basic authentication, CRUD for 2-3 entities, 1 simple integration (Stripe, PagSeguro, PIX direct) and a lean admin panel. Usually no separate mobile app — just responsive.

• Typical timeline: 6 to 12 weeks
• Team: 1 fullstack dev + 1 part-time designer
• Ideal for: testing product-market fit before heavy investment

Internal system (intranet, control, workflow) — USD 10k to USD 30k (R$ 15k–30k)

A system that automates a process inside the company: expense approval, production control, ticket management, scheduling, asset tracking. Typically 10 to 20 screens, user profiles (admin/manager/operator), PDF/Excel reports and basic dashboards.

• Typical timeline: 10 to 20 weeks
• Team: 1-2 devs + designer
• Ideal for: replacing an Excel spreadsheet that became Frankenstein

Custom ERP / management platform — USD 30k to USD 100k (R$ 30k–60k)

Backbone operation system: financial, inventory, sales, HR integrated in one place. 40+ screens, multiple profiles, mass imports, mandatory tax integrations (electronic invoicing, fiscal reporting), executive dashboards. These projects usually have the client's internal team collaborating (controller, operations manager).

• Typical timeline: 20 to 40 weeks
• Team: 3-5 people (devs, designer, PM, QA)
• Ideal for: company that grew beyond what off-the-shelf ERP handles

Marketplace or 2-sided platform — USD 40k to USD 120k (R$ 40k–80k)

Platform with sellers and buyers: catalog, advanced search, cart, split payment between marketplace and seller, ratings system, basic anti-fraud, seller panel. Usually also needs mobile version (PWA or native app). Development time explodes with each specific business rule.

• Typical timeline: 24 to 48 weeks
• Team: 4-6 people (frontend, backend, mobile, design, PM, QA)
• Ideal for: aggregating offers from multiple suppliers in a single channel

Scalable multitenant SaaS — USD 60k to USD 160k+ (R$ 50k–100k+)

SaaS product that will serve many clients in isolated environments, with recurring billing, granular plan control, white-label, public APIs and SDK. The ceiling here is high: robust products for large teams (think Pipedrive, Linear, ContaAzul style) easily go past USD 200k just on MVP. The ROI comes from the recurring revenue.

• Typical timeline: 30 to 52 weeks
• Team: 5-10 people with continuous dedication
• Ideal for: product you'll sell as subscription to 100+ clients

These numbers assume custom development with qualified team in Brazil. For context about alternatives that come up often:

• Off-shore (India, Vietnam, Argentina): can cut 30 to 50% of direct cost, but has hidden costs (remote management, timezone, rework time, legal difficulty if something goes wrong). Pays off in medium-large projects where you have a dedicated bilingual PM.
• Low-code (Bubble, Retool, Glide, Webflow): delivers part of the MVP at a fraction of the cost. Limited to cases without heavy flow customization or complex business rules. Good for cheap MVP, bad for scaling.
• Single freelancer: cheaper on small projects, but doesn't scale — when they go on vacation or drop the project, you're left with no one who knows the context.

What goes into the price — the real composition

When you see USD 40k in an ERP quote, that number is the sum of these components (with project-specific variation):

• Discovery and planning (5-10%): business immersion workshops, product roadmap, navigable prototype, requirements documentation.
• UI/UX Design (10-15%): wireframes, design system, high-fidelity final screens, usability adjustments.
• Development (50-60%): frontend, backend, database, internal and external APIs, integrations.
• QA and testing (10-15%): automated tests, exploratory manual testing, bug fixes.
• Initial infrastructure (5-10%): cloud setup (AWS / Google Cloud / Azure), CI/CD, monitoring, backups, SSL.
• Management and communication (5-10%): weekly meetings, documentation, risk management, continuous stakeholder communication.

If a quote comes with 90% in "development" and nothing in design or QA, watch out. Either those costs got swept under the rug (and will turn into production bugs later), or the vendor simply won't deliver those stages.

Total cost: year 1 vs subsequent years

The development price is just the entry ticket. The right question is "how much will it cost to keep this system running for 3 years?". Many companies invest USD 50k in an ERP and then discover they need another USD 30k to keep it alive. It's not a scam — it's the nature of software.

Year 1 (with launch)

• Development: the number from the previous section
• Hosting and infra: USD 40 to USD 1,000/month (varies a lot by load and architecture)
• Post-launch maintenance: 15 to 25% of the project's original value (fixes, adjustments requested by users in the first 90 days)
• Licenses and SaaS: transactional email (SendGrid, Postmark), monitoring (Sentry, Datadog), CDN, upload antivirus — USD 20 to USD 400/month
• User training: USD 1k to USD 6k, depending on team size

Years 2 and 3

• Maintenance: 15 to 20% of the original value annually (fixes + small evolutions included)
• Hosting: tends to grow with usage (10 to 30% per year)
• New features: quoted separately when demanded — not included in "maintenance"
• Stack updates: every 18-24 months, there's a cost to update framework versions (React, Node, etc.) to keep things secure

Rule of thumb: plan 50 to 70% of the development value over the first 3 years for maintenance and operation combined. Whoever ignores this discovers it when the system breaks and no one in the company remembers how it works anymore — or worse, the original vendor vanished.

"The biggest cost of software isn't development — it's the maintenance of a poorly-built system. The company that saves USD 10k in development and spends USD 40k fixing it over the next 2 years is the typical client who comes to us asking for a rewrite."

How to request a quote that makes sense

To get comparable proposals from different vendors, come prepared. What looks like "wasting time writing a brief" saves months later.

• Project type and business objective in 1 sentence: "B2B freight quote platform to shorten negotiation from 5 days to 1 hour".
• Essential features vs nice-to-have (separate lists): prevents nice-to-haves from entering the scope and inflating the price.
• Expected volume in the first year: number of users, transactions per day, simultaneous integrations.
• Mandatory integrations: current systems (which ERP? which CRM?), external APIs (PIX, WhatsApp, e-invoicing), single sign-on.
• Desired launch deadline and what changes if delayed: marketing event, end of contract with current system, seasonality.
• Budget range you have: yes, say it. Those who don't say it get proposals in the dark — the vendor will guess high to cover risk.
• Who decides and in what timeframe: align expectations on the evaluation timeline.

With this brief, you get truly comparable proposals. Without it, you'll compare apples to oranges — each vendor will assume something different about scope and prices will vary 3 to 5x without real technical reason.

Signs of a trustworthy vendor (and who to avoid)

Beyond price, some signs help separate serious vendors from those who will become a headache in 3 months:

Positive signs

• Asks hard questions in the first meeting (business rules, critical flows, plan B if something fails)
• Shows portfolio of similar projects at the same size, with complete cases (problem → solution → result)
• Offers sprint/milestone model instead of "fixed price by rigid scope"
• Has their own team, not a freelancer intermediary
• Technically documents what they deliver (you or another vendor can maintain it later)

Warning signs

• Fixed price delivered in 24h without detailed brief
• Refuses to sign NDA before the commercial meeting
• Promises unrealistically short timelines (3 months for a full ERP = lie)
• Wants 100% upfront payment
• No active company registration, verifiable physical address, or contactable references

Conclusion

A well-built web system isn't a cost, it's an investment that pays back 2 to 5 times in productivity or new revenue channel over 3 years. The price varies from USD 6k (MVP) to USD 160k+ (robust SaaS), and what most decides the final number isn't the chosen technology — it's how clear your company is about what to deliver and what to measure.

If you're at that decision point and want a frank conversation (without the "it depends on scope"), talk to us. In 30 minutes we understand your case and nail a real range.

Frequently asked questions

Is it possible to build a system for less than these ranges?

Yes, using low-code tools (Bubble, Retool, Glide) or hiring individual freelancers. Works for small scopes and ideas to validate (up to 10 users, no major integration). Above that, technical limits show up fast and the initial savings become double spending when you need to rewrite.

Why do two quotes for "the same project" come out so different?

Because the scope in each vendor's head is different. Who charges less may be removing things (custom design, automated tests, documentation, training) that the expensive one is including. Asking for detailed deliverables in the proposal solves it — when deliverables are in the same spreadsheet, price becomes comparable.

Can I pay in installments?

Yes — the standard is splitting in milestones (downpayment + monthly deliveries or per sprint). Vendors offering structure like "30% down / 40% during / 30% on final acceptance" have risk alignment with you. Those who want 100% upfront, avoid. Those who accept 0% down, be equally suspicious (either desperate for cash, or will abandon mid-project).

How long does a USD 50k ERP take?

Real range: 5 to 9 months of active development, plus 1 to 2 months of acceptance, training and post-go-live adjustments. Who promises 3 months for something this size either has an off-the-shelf product disguised as "custom", or will delay and charge more for it later.

Is it worth starting with cheap MVP and evolving?

Almost always yes. USD 10k MVP in 2 months gives you real product and user validation before spending USD 60k on the complete version. Most of Reiko's large systems started small, validated hypothesis and grew in 2-3 month cycles. Investing everything in the complete version before validating is like building a tower without foundation.