“Build or buy” is a false binary. Most business software decisions are about how much of the process you own, how much you configure and who operates it. A commercial product with a custom integration is neither pure buy nor pure build. A low-code workflow that your operations team maintains is a third category altogether.
This comparison sets out five delivery models, the criteria that should decide between them and a way to score the choice without comparing a subscription price to an unfinished development estimate.
The five models in practice
- Commercial SaaS product. A vendor owns the code, hosts the service and sets the roadmap. You configure, integrate and adopt.
- Low-code or platform configuration. A product provides the engine and you assemble the workflow, forms, permissions and reports within its limits.
- Custom build. You commission software for a defined process, own the repository and carry the maintenance burden.
- Hybrid extension. A product covers the core, and a thin custom layer handles the integration, workflow or report the product does not support.
- Managed service. A provider operates the process and supplies the software as part of an outcome-based service.
Criteria for a fair comparison
- Process fit: how closely the model matches how the work actually runs, including exceptions.
- Time to first value: how long before a real user completes real work.
- Total cost: licence, implementation, integration, data migration, training, support, internal administration and exit.
- Differentiation: whether the process is a source of advantage or a commodity.
- Control: who decides the roadmap, the data model and the change priority.
- Security and compliance: access, audit, residency, retention and vendor assurance.
- Integration: the number and fragility of connections to existing systems.
- Internal capability: the people you can hire, retain and spare to own the model.
- Lock-in and exit: portability of data, configuration and knowledge.
- Maintenance: who patches, monitors, upgrades and supports the system at 2am.
Comparison table
| Model | Best at | Weakest at | Ownership burden | Exit difficulty |
|---|---|---|---|---|
| Commercial SaaS | Standard processes, fast start, predictable operations | Differentiation and unusual workflows | Low | Medium, depends on data export |
| Low-code configuration | Specific but stable workflows with internal administration | Complex logic, high volume and deep integration | Medium | Medium to high, configuration is sticky |
| Custom build | Exact fit, differentiated rules and complex integration | Speed, cost certainty and long-term maintenance | High | Low if you own the code and documentation |
| Hybrid extension | Keeping a good product while fixing one gap | Upgrade conflicts and split ownership | Medium | Medium, the extension becomes a dependency |
| Managed service | Outcome delivery without internal operating capacity | Transparency, control and margin over time | Low to medium | High if knowledge and data sit with the provider |
Where each model breaks
SaaS breaks when the business keeps a workaround for every second order. The licence renews, the workaround becomes a permanent side process and nobody can say which system holds the truth. It also breaks when the vendor's roadmap moves away from your needs and the data cannot leave cleanly.
Low-code breaks when the workflow outgrows the platform's logic or performance envelope. The team that built the forms moves on, the configuration is undocumented and every change becomes a rediscovery exercise. Low-code is still software; it needs version control, testing and an owner.
Custom build breaks when it is treated as a project rather than a product. The launch is celebrated, the product owner moves to another role and the backlog becomes a list of complaints. The code may be excellent and still fail because nobody funded its second year.
Hybrid breaks at upgrade time. The product changes a data model or an API and the extension stops working. It also breaks when the extension quietly becomes the most important part of the system while remaining outside the vendor's support boundary.
Managed service breaks when the contract measures activity rather than outcome. Tickets are answered, reports are produced, but the business cannot see the cost drivers or take the process back in-house without losing the knowledge.
Weight the criteria by strategic role
Not every process deserves the same analysis. A payroll engine and a customer loyalty programme have different strategic value, different risk and different tolerance for customization.
Use a simple weighted score. For a commodity process, put more weight on time to value, total cost, security and exit. For a differentiating process, put more weight on fit, control, integration and the ability to change. Score each model from zero to three, multiply by the weight and add. Keep the evidence for each score: a reference call, a tested integration, a written quote, a maintenance estimate.
The weighted result is a conversation starter, not an answer. If the highest score depends on an assumption nobody has tested, the assumption is the real decision.
Choose the model by context
- SaaS when the process is common, speed matters, the vendor is financially credible and the data can be exported.
- Low-code when the workflow is specific but stable, the volume is moderate and an internal administrator can own it.
- Custom build when the process is a real differentiator, integration is complex and a product would force a workaround on every transaction.
- Hybrid when a product covers the core well and a narrow extension solves the remaining gap without duplicating the product.
- Managed service when you lack operating capacity, the outcome can be defined and the contract includes data, knowledge and transition rights.
Build a like-for-like business case
Compare the same scope over the same period. Include subscription or licence, implementation, integration, migration, training, support, internal administration, change requests, infrastructure, security review and exit. For a custom build, include the cost of a product owner, testing and the second-year roadmap. For SaaS, include the cost of the workarounds the product does not cover. For a managed service, include the cost of contract management and transition.
Then run a sensitivity check: what happens if volume doubles, the vendor raises prices, the key administrator leaves or the process changes? A model that only works under one set of assumptions is a bet, not a business case.
Red flags in the decision paper
- “We are unique” without evidence that the process differs from peers.
- A vendor demo treated as a requirements workshop.
- A build estimate with no maintenance, support or product ownership line.
- A subscription comparison that ignores integration and internal administration.
- No named owner for the system after launch.
- No exit plan, data export test or transition clause.
- Low-code adopted without version control, testing or documentation.
- A managed service contract with no outcome definition.
If you need help framing the decision or building the custom option properly, start with our Custom Software & Integrations practice. Our pricing page explains the scoping of discovery, build and support, including estimates. For the cost of changing direction later, read how cloud vendor lock-in changes as a business grows.
Sources
- UK Government Service Manual, Choosing technology: an introduction, accessed 17 November 2026.
- ISO/IEC 25010: product quality model for software evaluation, accessed 17 November 2026.
- NIST, SP 800-64: Security Considerations in the System Development Life Cycle, accessed 17 November 2026.
- Gartner, Worldwide IT spending forecast, accessed 17 November 2026.
Responses (0)
No comments yet. Add the first one below.
Leave a response
Your email address will not be published. Required fields are marked *.