The question arises sooner or later in almost every SME: there is a process that does not run smoothly, and two possible answers. Either you buy a ready-made off-the-shelf product, or you have something built that fits you exactly. The honest answer is: it depends, and not on the price, but on the nature of the process. Anyone who looks only at the acquisition cost here often gets the decision the wrong way round.
Off-the-shelf software and custom software: the real difference
Off-the-shelf software is a finished product used by many companies at the same time. You buy a licence or take out a subscription and immediately get something proven. The cost is spread across many customers, which is why entry is cheap. The catch: the software decides how you work. You adapt your processes to the product, not the other way round.
Custom software is built for your specific process. You determine the functions, the software belongs to you, and no one else works with the same solution. That costs more at the outset and takes longer. In return, it maps exactly what you do, and grows with you.
The central difference therefore lies not in the technology, but in the question of who adapts: you to the software, or the software to you.
Upfront costs deceive: calculate over years
When you buy, off-the-shelf software almost always looks cheaper. Sign the licence, get going, done. But the relevant figure is not the acquisition price, but the total cost over the period of use, known in technical terms as the Total Cost of Ownership (TCO). The sensible window for this is three to five years, not twelve months.
Over that distance the picture shifts. With off-the-shelf software, licences, implementation, training, ongoing support, interfaces, updates and adjustments add up. With custom software, the costs come at the front; after that you mainly pay for maintenance and targeted extensions. Which model is cheaper depends on how long you use the solution and how far it moves to the centre of your business.
One point is regularly underestimated here: subscription prices rise. Software-as-a-Service has for years been getting more expensive significantly faster than general inflation. The SaaS Inflation Index from the procurement provider Vertice records an annual price increase of around eleven to twelve per cent for the start of 2025, while general inflation in the G7 states stood at around 2.7 per cent, so roughly four to five times higher. What is a good deal today can be an expensive subscription in three years, one that is hard to get out of again.
The hidden costs of off-the-shelf software
The licence is the visible price. The expensive items are rarely on the quote:
- Interfaces: Many providers advertise “flexible integrations”. In practice, every connection to another system costs money, often not just once but annually for maintenance. Three or four systems that are meant to talk to each other cleanly can cost more than the licences themselves.
- Workarounds: When the software can do 85 per cent, your people make do with the rest using Excel lists, double entry and manual intermediate steps. These workarounds appear on no invoice, but cost working time every day and produce errors.
- Licence model and lock-in: Per-user prices grow with your team. And the more deeply a product is built into your daily routine, the more expensive the exit becomes. High switching costs are precisely the lever with which providers push through price increases.
- Data location: Where is your data held? The revised Federal Act on Data Protection (revFADP) has been in force since 1 September 2023. Where data is moved abroad, for instance because a cloud provider’s servers are located there, the statutory requirements must be met. Wilful breaches of certain obligations can be penalised with fines of up to CHF 250,000, which are directed primarily against the responsible natural person and not against the company (legal fact under the revFADP). This is not an argument against off-the-shelf software, but a reason to look closely at the provider.
These items are not a fundamental argument against off-the-shelf software. They simply belong in the calculation before you sign.
When off-the-shelf software is the right choice
For a large part of your software, buying is the sensible answer. The rule of thumb: where your process is standard for the industry and yields no competitive advantage, you buy.
Off-the-shelf software is right when:
- the process looks the same everywhere: bookkeeping, payroll, email, document storage, calendar;
- a proven product covers 80 to 90 per cent of your requirement and the rest is a manageable compromise;
- you need to start quickly and have no time for a development project;
- there are many providers and therefore real competition, which keeps prices and quality in check.
No one should have a bespoke bookkeeping program built. This process is governed by law and is roughly the same for everyone. Writing your own code here costs money and brings no advantage.
When your own process justifies your own code
The decision tips as soon as software touches what distinguishes you from the competition. This is precisely where the decisive test lies:
Would a ready-made product for this process force you to share your advantage with every competitor who uses the same product? If so, that is a strong argument for custom development.
Custom software is worthwhile when:
- the process carries your business model and is not just a secondary function;
- your way of working is so distinctive that no standard product maps it cleanly, and you have no wish to make yourself smaller to fit;
- you have to connect many systems and the standard product forces you to live permanently with workarounds;
- you need full control over data, functions and further development.
The reason is not enthusiasm for technology, but economics. When a process is your core business, a solution that maps it exactly pays off, rather than forcing a generic middle path. Software then changes from a cost item into an asset that belongs to you.
The third way: standard plus agent
In practice, the question is rarely “buy everything or build everything”. The pragmatic path for most SMEs is a combination: you keep proven off-the-shelf software for standard tasks and supplement it precisely where it slows you down.
Instead of replacing an expensive system entirely, you put an AI agent or a small piece of custom development at the sore points. The agent takes on the bridges between systems, the manual double entry, the reading of documents, the follow-ups. The existing standard tool stays, the agent connects, supplements and automates what would otherwise be done by hand.
This is not a promise of the future. At Vollmer Labs we apply exactly this pattern in our own projects. For a US accountancy firm (CPA) we run bookkeeping agents that take on the recurring routine work around the leading off-the-shelf accounting software, while that software remains the system of record. With ballistic.club and the RFQ tool for LED signage behind rfqbuddy.com we build our own platforms, where the individual process was the core and a standard product would not have fitted. For kitchen studios we have brought to market, with jeffri.ch, an AI-assisted workflow tool that bundles order compilation, supplier confirmation and installation planning. To be honest: the building blocks are running, but not every industry already has a finished live operation, and we do not claim otherwise.
The advantage of the hybrid model is plain. You only pay for custom development at the points where it measurably delivers something, and you keep the stability and the low entry price of the standard product. For more on where custom development pays off, see Custom Software Development; how an agent slots into existing systems is shown on the page about AI Agents.
The decision in four questions
Before you buy or have something built, answer these four questions honestly:
- Is the process standard for the industry or unique? Standard speaks for buying, unique for building.
- Does the process carry your business model or is it incidental? Core business justifies bespoke code, incidental functions rarely.
- What does the solution cost over three to five years, including interfaces, training, subscription price increases and switching costs? The acquisition price is not what decides.
- Could the problem be solved by keeping off-the-shelf software and only supplementing the sore point with an agent? Often the cheapest answer.
Anyone who answers these four questions cleanly needs no ideological stance of “do everything yourself” or “off-the-shelf only”. The right answer is usually nuanced, and it saves money at precisely the point where an undifferentiated gut feeling costs it.