The honest answer: it depends on what exactly you want built, and that range is wider than most pricing pages let on. This article covers the order of magnitude Dutch agencies use in 2026, the characteristics that determine where your project lands within those ranges, and (just as importantly) what it costs to keep the application running for years after launch.
The order of magnitude for building in 2026
Drawn from public pricing pages of Dutch agencies:
| Project type | 2026 indication | Lead time |
|---|---|---|
| Simple tool or automation | € 1,000 to € 15,000 | 3 to 6 weeks |
| Customer portal or dashboard | € 5,000 to € 35,000 | 6 to 10 weeks |
| MVP / first working version | € 3,000 to € 40,000 | 4 to 10 weeks |
| Business software with integrations | € 25,000 to € 75,000 | 2 to 5 months |
| Full platform or SaaS | € 25,000 to € 150,000+ | 4 to 8+ months |
Plus hourly rates that vary significantly:
- Dutch agencies: typically € 90 to € 160 per hour
- Freelancers: € 65 to € 125
- Nearshore (Eastern Europe): € 45 to € 80
- Offshore (Asia): € 20 to € 50
These numbers give you an order of magnitude, not an answer. For your specific question, the figure shifts depending on a handful of characteristics of the project itself.
What determines the build price
1. The workflows and purpose of the application
The biggest variable, and the one most often buried under feature lists in proposals. What needs to happen, when, and for whom? Who does what, with which roles and permissions? Where are the decision points, the exceptions, the manual checks that on second look actually need to be automated? Two applications with identical screens can be a factor of three apart in build cost if one is linear and the other is full of conditional flows, approvals and exception rules.
2. Application type and level of integration
This is where agencies quietly make the most margin, because clients often don't know where their request falls on this scale. Four levels, each with its own workload and pricing profile. A disclaimer up front: the figures below are order-of-magnitude for a bare variant at that level. Anything on top (integrations, roles, exception rules, reporting, integrity requirements) pushes the real price up, sometimes significantly.
Level 1: CRUD or administrative application. Create, read, update, delete. Multiple objects, a decent permission model, a few reports. This is the application type agencies make their margin on: manageable, low-risk, easy to estimate. With the right foundational frameworks for back-end and front-end, we build this at high velocity. Indication: € 2,000 to € 8,000, depending on the number of objects and the depth per object.
Level 2: Data integration. Moving data between systems: orders from Shopify to Moneybird, product data from a PIM into your webshop. In practice roughly factor two on top of a comparable CRUD application, because you're now dealing not only with your own data model, but also with the data model and quirks of an external system. Again dependent on what exactly needs to flow, and how mature the API on the other end is.
Level 3: Workflow orchestration. A process running across multiple systems: dependencies, ordering, error handling, what happens if one system is briefly unreachable. The price depends heavily on how you do this. Using an existing orchestration tool (n8n, LangChain or similar): faster to go live, lower initial investment, but with licensing or usage costs that keep running. Building it yourself: more setup up front, full control, no vendor dependency, and more appropriate for organisations with a genuinely complex or critical operation where standard tools hit their limits. For small and mid-sized businesses the standard solution is almost always more feasible, both in build cost and in total cost of ownership over a few years. The question isn't "which tool is cheaper today", but "which one will still fit how our processes evolve three years from now".
Level 4: Process optimisation. The application built in detail around the client's way of working, with the explicit goal of removing waste, extractive load and inefficiencies from the process and fully aligning the system with how the work actually happens. This level has the most uncertain price range of the four, precisely because the value lies in what isn't yet visible at the start. This is where choosing a strategic partner makes the difference: someone who doesn't just build what you ask for, but thinks with you about what you actually need. The up-front investment is higher, so is the long-term return, usually in time no longer lost to manual juggling, corrections and exceptions.
In practice you'll often see a mix: a CRUD foundation, a couple of data integrations, one orchestrated workflow and, at one strategic point, a piece of real process optimisation. That mix can't be captured in a single number, and that is exactly why a proposal that does capture it in one number should raise questions.
3. Design and user experience
For internal tools, a clean, functional UI can be enough. For customer- or public-facing applications, a serious UX effort adds 15 to 30 percent to the build budget, with roughly 20 percent as a common industry rule of thumb. That isn't optional polish. UX investment typically returns in higher adoption, lower support load, and an application people want to keep using rather than work around.
4. Type of supplier
The hourly rate range feeds directly into the final bill, but the choice runs deeper than rate alone. Roughly four types to choose from:
- Large custom software agencies: broad team, mature processes, extensive track record. Hourly rate at the top end (€ 120 to € 160), with an overhead layer you pay for: PM, account management, a sales organisation. Appropriate for organisations that genuinely need that structure.
- Young, specialised agencies like us. The same solidity and structure as a larger agency (real track record, consistent working method, maintenance and operations in order), but without the overhead layer. Plus the drive of a team still establishing its name: every project counts, every client counts, involvement is personal. In practice a sharper hourly rate at comparable quality, and a client who sits at the table directly with the people doing the work.
- Freelancers: € 65 to € 125 per hour, flexible and personal, but dependent on one individual. No backup if that person drops out or the project grows bigger than one head can carry.
- Nearshore and offshore: € 20 to € 80, lowest rate, but higher communication and time-zone load. Often extra coordination hours needed on your side to reach the same result.
A low hourly rate is not a low final bill if more hours are needed to arrive at the same outcome.
5. The use of AI-assisted development
In 2026, the factor that separates a proposal from 2023 from a proposal today. AI tooling hasn't so much increased speed as it has shifted the composition of a build engagement, away from manual typing, toward architecture, review and integration work. That changes what an hour of development actually produces, which suppliers use that hour efficiently, and why a low hourly rate doesn't automatically mean a low final bill. An agency that uses this tooling effectively tends to deliver faster at comparable quality and can price more sharply without compromising on solidity.
What's always included with us
One factor often presented as a price driver doesn't belong in that list at our agency: corporate compliance baseline, meaning access control, logging, backup strategy, GDPR-aware data storage and sensible error handling. Our baseline applications meet this by default. It isn't extra work, it isn't an optional module. It is part of what "delivered" means at our agency.
What it costs to keep an application running
Build costs are one-off. But an application then needs to keep running for years, and that costs money at two levels often lumped together in proposals: infrastructure (hosting and operations) and application (maintenance and support). Pulling the two apart makes it much easier to see where a proposal is being vague.
Hosting and operations: the infrastructure
Where the application runs, who keeps the servers healthy, who monitors logs, who patches the operating system, where the backups live. Order of magnitude per month for EU hosting:
- From around € 50 for a bare CRUD-oriented application on a basic managed VPS: limited traffic, standard backup, no heavy workflows. The absolute floor for a professionally hosted application.
- € 150 per month as the typical SMB reality for a full web application with data integrations, monitoring and proper backup and security. Multiple Dutch hosting sources cite this as the average for SMB web apps.
- € 300 or more per month for large, complex applications with heavy workflows, ongoing change programmes, higher availability requirements or stricter compliance demands.
These figures cover: hosting (server, database, backup, SSL), monitoring, security patches on the infrastructure, log storage. They do not cover: changes to the application itself.
Maintenance and support: the application
Separately from the infrastructure, the application level keeps running. An API version change, an integration that adds a new field, a small process adjustment, a dependency that needs a security update, a user bumping into a question. Budget for 15 to 25 percent of the build investment per year.
How much of this goes into support depends largely on two decisions made at build time:
- How self-evident the workflow and user experience are. An application whose logic is immediately clear generates few support questions. An application that works cleverly but confuses users asks for explanation indefinitely.
- How good the end-user documentation is. Building documentation costs time once, during the build, but reduces support load across the whole lifetime. Whether you want that documentation built is your call as the client, and that call directly affects both the build proposal and the support cost afterwards.
An application with clear UX and proper documentation sits at the bottom of that 15 to 25 percent, one without at the top.
Total cost of ownership
An honest cost comparison between suppliers doesn't just look at the build, it looks at the total cost of ownership (TCO) over three to five years. The simplified baseline formula that usually suffices for SMB:
TCO = Build cost + (Hosting/operations × 12 × years) + (Maintenance/support per year × years)
For completeness: classic TCO (as originally introduced by Gartner) also includes items like user training, productivity loss during downtime, and inefficient use. Gartner itself notes that organisations focused only on sticker price miss an average of 15 to 20 percent in "invisible" costs. For most SMB applications, build, hosting/operations and maintenance/support are the three items you want clearly on paper. The others matter, but are harder to quantify up front.
A simple three-year example for a € 20,000 build:
| Item | Per month / year | Over 3 years |
|---|---|---|
| Build (one-off) | € 20,000 | € 20,000 |
| Hosting/operations | € 100 per month | € 3,600 |
| Maintenance/support | 20% of build = € 4,000 per year | € 12,000 |
| Total | € 35,600 |
An application that's € 5,000 cheaper at build time, but € 150 per month more expensive in hosting and five percentage points higher in annual maintenance, is actually more expensive over three years. Those are the numbers that belong on the table when you compare two proposals, not just the build figure.
In closing
The question "what does a web application cost?" isn't the right one. The right question is "what does a web application cost that still fits us three years from now, including maintenance, operations and continued development?" Suppliers who give an honest answer pull build, infrastructure and application maintenance apart in their proposal. Suppliers who don't, tend to hide where the bill lands later.
We prefer to do it the other way around: the total picture first, the components transparently underneath, and a maintenance arrangement that fits how critical the application is for you. Whether it turns out to be € 5,000 or € 150,000, the questions to ask are the same.