In short
Custom software pays off when your process is so specific that standard software only covers it with workarounds, or when the software itself is your edge in the market. A good project starts with a requirements phase, builds a prototype first and then a first usable version (MVP), and grows in short iterations after that. Cost and duration depend on scope, integrations, roles and permissions, data migration, data protection and maintenance after launch. Before you sign, settle in writing who gets the rights to the code, where the software runs and what the handover includes.
When does custom software pay off?
The basic question of whether to buy or build is covered in the comparison Standard software or custom development. In short: for accounting, email or a ticket system, an off-the-shelf product is almost always enough. Your own software pays off where your process differs from other companies' and that difference makes or saves money.
Between buying and building there are middle paths worth checking before you ask for a quote:
| Path | Fits when | Limit |
|---|---|---|
| Adapting standard software | The product covers most of it and has an API or extensions | You stay tied to the vendor's prices, features and roadmap |
| Low-code or automation | It's about forms, approvals or data moving between existing tools | Complex logic, many users or custom interfaces quickly hit the platform's limits |
| Custom software | The process is your core, the data should stay with you, several systems have to work together | You carry development and maintenance, including after launch |
For the middle path, see Self-hosting n8n: if the job is mainly moving data between tools you already have, an automation is often enough instead of an application of your own.
A good test before any request: write down in three sentences what the software should do differently from what you use today. If that's hard, the problem isn't clear enough yet for a development project.
How does a software project run?
Every provider names the phases a little differently. The core is always the same: understand first, then build small, then expand.
Clarify requirements (discovery)
Together you describe who uses the software, which tasks it handles and which systems it connects to. The result is a written document with goals, user groups, workflows and a list of features sorted by priority. This phase is the basis for any quote that is more than a guess.
Prototype
A clickable draft of the most important screens shows whether the workflow works for users before a line of production code exists. Show it to the people who will work with it, not just management. Changes are cheapest here.
MVP (first usable version)
MVP stands for minimum viable product: the smallest feature set that lets real users complete their most important task. Everything else waits. An MVP is not a throwaway prototype; the code should be built so it can grow.
Iterations
After that, the software grows in short rounds: plan, build, show, work in the feedback. In the widely used Scrum framework these rounds are called sprints, and according to the Scrum Guide (retrieved November 2026) they last one month or less, each ending in a usable result. So you regularly see what has been built and can change priorities.
Launch
Before launch come tests with real data, the import of existing data, training and a plan for when something goes wrong. Many teams first launch to a small group and then open it to everyone.
Operation and maintenance
After launch the software isn't finished. Security updates for frameworks and libraries, backups, monitoring, bug fixes and new requests keep coming. Settle beforehand who handles them and how they are billed.
What drives cost and duration?
A serious price only exists once it's clear what will be built. These factors shape it most:
| Driver | Why it matters | What you can clarify beforehand |
|---|---|---|
| Scope | Every feature has to be built, tested and maintained | Which features does the MVP really need, which can wait? |
| Integrations | ERP, CRM, payment providers or legacy systems have their own rules, limits and failure cases | Which systems, is there a documented API, who has access to a test environment? |
| Roles and permissions | Who may see and change what? Every role multiplies the cases that need testing | Which user groups exist, and where do their permissions differ? |
| Data migration | Old data is rarely clean: duplicates, missing fields, free text instead of structure | How many records, from which sources, who decides on conflicts? |
| Data protection and compliance | Personal data calls for protection from the design stage, plus logs, deletion rules and contracts | Which data is processed, are there industry-specific rules? |
| Design | A custom design system costs more than a proven component library | Is there a corporate design, how important is the interface for success? |
| Maintenance after launch | Updates, hosting, monitoring and further development continue every year | Who runs the software, with what response times? |
With data protection, timing matters: in Art. 25, the GDPR requires data protection by design already "at the time of the determination of the means for processing", that is, during planning, as well as privacy-friendly defaults (retrieved 21 November 2026). Retrofitting it costs more than building it in from the start.
To the question of what custom software costs compared with standard software, there is no honest one-size answer. Standard software is paid continuously as a licence or subscription, often per user. Custom software costs mostly up front in development and then in maintenance and operation, but without a per-user licence. So never compare a quote with a subscription price; compare the total cost over several years, including the time your team spends on workarounds today.
Fixed price or time and material?
The two common billing models split the risk differently. A short overview is in the comparison Fixed price or hourly rate; here are the pros and cons in detail.
| Fixed price | Time and material | |
|---|---|---|
| Billing | One price for a scope described up front | By the hours or days actually worked |
| Advantage | You know the budget in advance; the provider carries the risk of it taking longer | Changes are possible at any time without renegotiating the contract |
| Disadvantage | Needs a precise description; changes need a written amendment | The budget is harder to predict; you need a good overview of the effort |
| Fits when | Goal and scope are clear, for example after a requirements phase | Much is still open or the product is developed continuously |
A mix often makes sense: the requirements phase and the MVP at a fixed price, further development afterwards in clearly bounded packages. With fixed-price quotes, make sure the scope is described concretely enough that both sides mean the same thing by "done". With time and material, a budget ceiling and regular reports on the hours used help.
Werkvertrag or Dienstvertrag?
Under German law, two contract types sit behind these billing models. Under a Werkvertrag (contract for work, § 631 BGB) the provider owes the "production of the promised work", that is, a result. This comes with acceptance under § 640 BGB: you check and confirm that the work meets the contract; by law, you may not refuse acceptance because of minor defects. Under a Dienstvertrag (service contract, § 611 BGB) the provider owes the "promised services", that is, the work itself rather than a specific result (all retrieved 21 November 2026).
How a specific contract is classified, and what that means for acceptance and defects, is a legal question. This section is not legal advice: have contracts for larger projects reviewed by a lawyer specialising in IT law.
Who owns the code?
This question is often asked only once there's a dispute. Under German copyright law, software is protected by § 69a UrhG, including its design material. For employed developers, § 69b UrhG provides that the employer exercises the economic rights to the program. That rule applies to employees and civil servants; with an external provider, the contract determines which rights of use you get.
If the contract isn't specific, the scope of the rights follows the shared purpose of the contract under § 31 (5) UrhG (retrieved 21 November 2026). That doesn't automatically cover everything you plan later, such as switching to another team. So have it put in writing:
- whether you get exclusive or non-exclusive rights of use, and whether you may modify the code, pass it on and have another team develop it further
- that you receive the complete source code, with access to the repository, not just the running application
- which third-party libraries and open-source components are used, and under which licences
- which parts the provider brings from their own reused building blocks, and what rights you get to those
Again: this is not legal advice, but the questions the contract should answer.
How do you choose the right provider?
Whether you hire a studio, a freelancer or a larger team: the answers to these questions say more than any portfolio.
Questions to ask
- How does the requirements phase work, and what do I get in writing afterwards?
- Who specifically works on my project, and will that person stay until the end?
- How often do I see working interim versions?
- How do you test, and how are bugs fixed after launch?
- Where is the software hosted, and can I later run it with another provider or myself?
- What does the handover include: source code, documentation, access, deployment instructions?
- What does maintenance after launch cost, and what does it include?
Red flags
- A binding price without any questions about your workflows
- No statement on who gets the rights to the code, or the code stays with the provider
- Hosting only on the provider's own platform, with no way out
- No interim versions until the "big presentation" at the end
- No answer to how updates and security vulnerabilities are handled after launch
Hosting and data protection
If the provider runs the software for you and processes personal data while doing so, they are usually a processor. In that case Art. 28 GDPR requires a data processing agreement that, among other things, provides for the data to be deleted or returned at the end, at your choice (retrieved 21 November 2026). Ask for this agreement and for the server location.
Documentation and handover
Good documentation explains how the software is structured, how to run it locally, how it's deployed and where credentials are kept. I recommend asking for the handover as a separate item in the quote from the start. That makes it part of the project rather than a favour at the end, and another team can take over the software if you change providers.
What to prepare before the first call
The better prepared you are, the more precise the quote.
- The problem in your own words: what doesn't work today, how much time or money it costs, and how you'll know the new software solves it
- The users: who works with it, how many people, with which different permissions
- Today's workflows: screenshots, spreadsheets, forms that should be replaced
- The system landscape: which tools need to be connected and who has access to them
- The data: what has to be migrated, in which format, how clean it is
- Requirements: data protection, industry rules, accessibility, internal IT policies
- A contact person who can make decisions and has time during the project
- A budget range: not an exact figure, but an order of magnitude, so the provider can suggest a fitting scope
The last point feels awkward, but it saves time on both sides: a provider can propose a lean and an extensive solution for the same problem, and without a range it's unclear which one you mean.
Frequently asked questions
How long does it take to have custom software built?
It depends on the same drivers as the cost: scope, integrations, roles, data migration and requirements. A serious answer only exists after the requirements phase. The fastest way to a usable result is a clearly bounded MVP.
What does custom software cost compared with standard software?
Standard software costs ongoing licence or subscription fees; custom software costs mostly the development up front, then maintenance and operation. Compare the total cost over several years, including the working time that workarounds with today's solution cost.
Can I switch providers later?
Yes, if you have the complete source code, the necessary rights of use, the documentation and all access. That's exactly why these points belong in the contract before the project starts.
[ Anolilab — Blog ]
More hands-on writing on AI, web development and infrastructure.
