Blog
Custom build, off-the-shelf package or low-code: a decision guide in five questions
· 5 min read
Your planning runs on Excel, your quotes on email, and your customer data lives in three different places. Clearly, something has to change. Then the conversation begins: do you buy an off-the-shelf package, have something custom-built, or go for low-code? Every supplier has its own answer, and it usually happens to be whatever that supplier sells.
We build with OutSystems ourselves, so we are not neutral. That is why what follows is not a sales pitch but five questions you can answer yourself, even if you never work with us. The answer is "off-the-shelf package" more often than you would expect from a software builder.
Question 1: Does an existing package do most of what you need?
Start here, because this is the quickest question to answer. Write down what your process needs to be able to do, and check it against two or three off-the-shelf packages in your sector. Do not look at the demo; look at your own trickiest case: the exception that comes back every month.
Does a package fit without major rework? Take the package. You get maintenance, updates and other companies running into the same problems. Accounting, an HR system or a standard CRM is not something you build yourself, not even with low-code.
Does the package fit for the most part but not entirely? Then the question is whether you can adjust the rest through settings, or whether you have to rework your process to fit the package. The latter is a choice you need to make deliberately, not an afterthought.
Question 2: Is this process something that sets you apart, or something everyone does?
Everyone books invoices. How you certify dealers, schedule customers or assess applications may be exactly what makes your business different. For ordinary processes, you want to be as standard as possible. For the process that earns you money or wins you customers, you want to stay in control yourself.
A real-world example is on our projects page. MEA Beveiliging processed every installation notification from its dealers by hand through email, which caused chaos and delays. The process of that company and its dealers was so specific that a dedicated platform was the logical route: a fully automated dealer and certification platform, built on OutSystems, in which dealers manage their own installations and certificates are generated instantly. You can read more about that pattern in From Excel and email to an application.
Is your process simply "the way everyone does it"? Then custom software is rarely worth it.
Question 3: How often does the process change, and who adjusts it?
A process that stays the same for five years can safely go into a package. A process that changes every quarter, because your market is moving or you are growing, calls for something you can adjust quickly.
This is where low-code performs well. On its own platform page, OutSystems describes visual, model-driven development while keeping control over the architecture. The idea is that you make a change in the model and have to rewrite less code. But it is still work for a developer. Do you want an employee to adjust screens and rules themselves without anyone else involved? Then you are more likely looking for a no-code tool, or a package with good configuration options. We explained that choice in What is low-code, and when does OutSystems suit your SME?
Question 4: How deeply is this embedded in the rest of your systems?
Almost every application has to talk to something else: your accounting, your ERP, a payment service, a customer portal. The more integrations, the more important it is that your build method supports them well.
The OutSystems documentation says on this: when you want to fetch or edit data in another system that offers REST APIs, you can use those APIs in your app. For authentication, there is built-in support for basic authentication and the OAuth 2.0 client credentials flow. Other forms of security can be handled through customisation.
There is also a way out for what the platform cannot do out of the box. The ODC documentation states that you can build your own logic in C# when the standard features do not cover everything, and then use it in your apps. It does mention a limitation: the actions and structures from such an external library are read-only and cannot be changed in the visual environment. So if you need something special, you can go beyond the visual model, but you pay for it in maintainability.
In practical terms: write down which systems you need to integrate with and whether they have an API. If a system has no API, that is often the real risk in the project, whichever route you choose. Prove such an integration first, not in the final week before go-live.
Question 5: Who will maintain it in three years, and how do you get out again?
This is the question almost nobody asks up front. Every route has a price that only becomes visible after go-live.
An off-the-shelf package depends on the supplier, who decides on updates and the direction of the product. Fully custom software depends on the builder or on an in-house team, because someone has to understand the code. Low-code depends on the platform, and on someone who knows the platform.
So ask three questions up front, whichever route you take:
- Who will adjust this once the builder is gone?
- Can we get our data and our rules out of it in a usable form?
- What happens if the supplier changes course?
If you cannot get a concrete answer to those questions, you have a risk, whatever the brochure says. We will be honest: with low-code, the platform is a dependency you need to enter into deliberately. For small, one-off problems, it is not worth it.
How to choose
Go through the five questions and see where you end up:
- The package fits and the process is standard: buy the package.
- The process is unique, changes regularly and has many integrations: low-code is usually a good candidate.
- The process is unique and the software is itself your product, or you have your own development team: fully custom software can make sense.
- A small, one-off problem: making smart use of what you already have is often enough.
Is the outcome unclear? Then that is the answer: you do not know enough yet. Start with a clickable mockup with realistic data. It quickly shows whether the direction is right, and it costs less than a wrong choice. You can read why that matters even more with AI in Mockup first, build later.
What does this mean for your business?
- Write down the process you want to replace, including the five exceptions you run into every month.
- Compare it against two or three off-the-shelf packages and test your trickiest case, not their demo.
- Make a list of the systems you need to integrate with and check whether they have an API.
- Answer the maintenance question in writing before you choose a route: who will adjust this in three years' time?
Are you stuck, or would you like a second opinion on a proposal or a package choice that is already on the table? We are happy to help, even if the answer is "buy that package". Feel free to get in touch.