
Sooner or later every growing business faces the question: should we buy a ready-made software product, or build something tailored to us? Vendors of both options have obvious incentives to give you a one-sided answer. The honest answer is that it depends — but it depends on a small number of factors you can assess yourself.
This article gives you a practical framework: what each option really costs over time, where each one shines, a simple scoring model and three realistic scenarios. It also introduces the option that is often best of all: buy, then extend.
Key takeaways
- Buy when the process is standard and not a source of competitive advantage.
- Build when the process is unique, central to how you win, or when combining several tools costs more than one tailored system.
- Compare total cost of ownership over three to five years, not just the first-year price.
- “Buy and extend” — a standard platform plus custom integrations — is frequently the best balance.
The two options, honestly
Off-the-shelf software (SaaS or packaged)
Advantages: available immediately, predictable subscription cost, the vendor handles hosting, security and updates, and you benefit from features requested by thousands of other customers. Documentation, training material and experienced users are easy to find.
Disadvantages: you adapt your process to the software; features you need may never arrive; per-user prices grow with your team; integrations may be limited; your data lives in the vendor’s format; and price increases or product changes are outside your control.
Custom software
Advantages: fits your exact process, can encode what makes you different, integrates precisely with your other systems, has no per-user licence cost, and you own the code and the roadmap.
Disadvantages: higher upfront investment, takes time to build, requires ongoing maintenance and hosting, and its quality depends heavily on who builds it.
Total cost of ownership: compare apples with apples
A common mistake is comparing a monthly subscription with a one-off development price. Instead, estimate the full cost over three to five years for both options:
| Cost element | Off-the-shelf | Custom |
|---|---|---|
| Initial cost | Setup, configuration, data migration | Discovery, design, development, migration |
| Recurring licences | Per user or per tier, usually rising over time | Usually none (open-source stack) |
| Hosting & operations | Included | Hosting, monitoring, backups |
| Maintenance | Included in subscription | Updates, security patches, small changes |
| Integrations | Connectors, middleware, workarounds | Built in, but must be maintained |
| Hidden costs | Manual workarounds, extra tools to fill gaps, training | Knowledge transfer, documentation |
| Exit cost | Data export, migration, retraining | Low if you own code and data |
The most underestimated line is “manual workarounds”. If people spend five hours a week compensating for a tool that does not fit — exporting, re-typing, reconciling — that is roughly 240 hours a year. Put a cost on that time before deciding.
A simple scoring framework
Score your situation from 1 (strongly favours buying) to 5 (strongly favours building) on each question:
- Uniqueness: how different is your process from industry standard practice?
- Strategic value: does this process directly differentiate you from competitors?
- Fit of available products: how much would you need to change your process to fit the best product? (More change = higher score.)
- Integration needs: how many systems must it connect with, and how deeply?
- Scale economics: how many users will it have in three years? (Many users on per-seat pricing = higher score.)
- Time pressure: how soon do you need it? (Urgent = lower score.)
- Ability to maintain: do you have, or can you contract, a reliable team to maintain it? (No = lower score.)
Add the scores. 7–15: buy. 16–25: buy and extend, or build a narrow custom component. 26–35: custom software is likely the better long-term investment. Treat the result as a conversation starter, not an oracle — but it forces the right questions.
The third option: buy, then extend
Many of the best solutions combine both approaches. You use a strong standard product for what it does well — accounting, email, CRM, e-commerce — and add custom components where your business is different:
- A custom integration that synchronises orders between your shop and your warehouse system
- A pricing or quoting engine that feeds the CRM
- A customer portal on top of your existing back-office system
- A reporting layer that combines data from several products
This keeps the custom code small, focused and cheaper to maintain, while you still benefit from the vendor’s investment in the standard parts.
Three scenarios
Scenario A: a 12-person accounting practice needs a CRM
The sales and client-management process is fairly standard, the team is small and they need it next month. Recommendation: buy a mainstream CRM, configure it well and automate a few integrations with the practice management system.
Scenario B: a logistics company with unusual routing rules
Dispatch depends on vehicle types, customer time windows, special handling and driver qualifications. Off-the-shelf tools handle 70% and staff fix the rest in spreadsheets daily. Efficient dispatch is a key competitive advantage. Recommendation: build a custom dispatch tool, integrated with the existing accounting and telematics systems.
Scenario C: an online retailer with complex product bundles
A standard e-commerce platform works well for the storefront and payments, but bundle pricing and stock allocation are unique. Recommendation: buy and extend — keep the platform, build a custom bundling and stock service connected via API.
Questions to ask before you decide
- What exactly is the problem, and what does it cost us today?
- Have we trialled the two or three best products with real data and real users?
- What will this cost over five years, including our own time?
- What happens to our data if the vendor raises prices, is acquired or shuts down?
- If we build, who will maintain it, and is the code in a repository we own?
- Can we start smaller — one module, one integration — and expand later?
Red flags on both sides
When buying: no data export, no API, long lock-in contracts, per-user pricing that makes growth painful, and a roadmap that ignores the features you need.
When building: no discovery phase, fixed price without a clear scope, code owned by the vendor, obscure technologies, no automated tests, no documentation and no plan for maintenance.
Frequently asked questions
Is custom software always more expensive?
Upfront, usually yes. Over three to five years, not necessarily — especially with many users on per-seat pricing or when off-the-shelf tools force costly manual workarounds.
How long does custom software take to build?
Focused tools often take four to ten weeks; larger platforms are delivered in phases, with a first useful release typically within three months.
Can we switch from off-the-shelf to custom later?
Yes, if you keep your data exportable. Many companies start with a standard product and build custom components once they understand their needs better.
What is the biggest risk of custom software?
Dependence on a single developer. Mitigate it with code ownership, documentation, tests and mainstream technology choices.
Get an independent recommendation
We are happy to recommend an off-the-shelf product when it is the right choice — it happens often. Our Discovery & Blueprint evaluates options objectively, prototypes the custom route if it makes sense, and gives you a fixed price to compare. For inspiration on what custom tools can do, see 12 Everyday Business Problems Custom Software Can Solve.