Direct answer: Buy off-the-shelf ERP software for standard back-office functions, customize it for workflows that differ from the norm but aren't your competitive edge, and only build custom software for the small slice of your operation that's genuinely unique to how you compete.
The framework
For every function under consideration, inventory, procurement, HR, order management, ask three questions in order:
- Is this function standard across our industry, or genuinely unique to how we operate?
- If unique, is that uniqueness a real competitive advantage, or just an accident of history ("we've always done it this way")?
- What's the cost of forcing this into an off-the-shelf tool's model vs. the cost of maintaining custom software forever?
Most functions fail question one, they're standard. Buy those. A smaller set pass question one but fail question two, customize a platform's existing extensibility rather than building from scratch. Only the rare function that passes all three deserves custom-built software.
Why teams get this backwards
The instinct to build custom is often about control, not strategy. Off-the-shelf tools feel constraining because they force you to adapt your process to their model, but that constraint is frequently a gift: it exposes which of your "unique" processes were actually just inherited quirks nobody had questioned.
What "customize" actually means
Between buy and build sits a huge, underused middle ground: configuring and extending an existing platform (most modern ERPs, NetSuite, Odoo, Microsoft Dynamics, have serious extensibility) to handle your specific exceptions, without owning the entire codebase. This gets you 80% of custom-build's flexibility at a fraction of the long-term maintenance cost.
A real-world pattern
In enterprise workflow projects I've scoped, the recurring shape is: standard accounting and inventory bought off-the-shelf, a customized approval-and-exceptions layer built on top of the platform's extension framework, and one genuinely custom system for whatever makes that specific business's core operation different from its competitors (a proprietary allocation algorithm, an unusual multi-party settlement process). Getting that split right up front is what keeps ERP projects from becoming multi-year, never-finished efforts.
FAQ
Is it ever right to build a full custom ERP from scratch? Rarely, and almost never as a first move, usually only once a company has hit hard limits with configured off-the-shelf tools after genuinely trying, and has an operation large enough to justify the ongoing maintenance cost of owned software.
How long does ERP customization typically take vs. custom build? Customization on an existing platform is typically measured in weeks to a few months per workflow; full custom builds are typically 6–18 months for a comparable scope, plus ongoing maintenance the vendor would otherwise absorb.
What's the biggest risk of over-customizing an off-the-shelf ERP? Customizing so heavily that platform upgrades break your configuration, at that point you're paying licensing fees for a platform while carrying most of the maintenance burden of custom software.
Working through a buy-vs-build decision right now? Book a free strategy session or grab the ERP Requirements Gathering Template.