The conventional wisdom is to buy off-the-shelf software and build only when nothing on the market does what you need. For most of the last decade, that was broadly right. SaaS tools improved, integration ecosystems matured, and the cost of maintaining custom software was genuinely high.
That calculus is shifting. AI is changing what custom software can do and how quickly it can be built. The question isn't "build or buy" anymore — it's knowing which problems fall into which category.
The Case for Off-the-Shelf (Still Strong for Most Things)
Buying wins when:
The problem is generic. Payroll, email, CRM, project management — these problems are well-understood, the market has solved them well, and your differentiation doesn't come from doing them differently. Use Xero, HubSpot, Linear.
Your requirements fit 80% of the product. The danger zone is buying something that's 80% right and spending years bending it to the remaining 20%. If that last 20% is genuinely important, the customisation cost often exceeds building from scratch.
You need it tomorrow. Off-the-shelf software is live in days. Custom software takes weeks to months. If speed is paramount, buy and optimise later.
The vendor's roadmap aligns with yours. If a well-funded SaaS company is building exactly what you need next, waiting might be the right call.
When Custom Software Wins
Custom software earns its cost when it becomes a competitive advantage — when the way you do something is a meaningful part of why customers choose you.
Your process is genuinely different. Most businesses use CRM the same way. Some don't. If your sales motion, service delivery, or operations don't fit standard categories, off-the-shelf tools create friction that compounds over time.
Integration is the product. Many of the most valuable custom systems exist not to replace an off-the-shelf tool, but to connect five of them in a way no vendor has bothered to build. A custom middle layer that unifies your CRM, ERP, and support desk — and adds business logic that spans all three — is often transformational.
You're sitting on data that creates value. If your business accumulates proprietary data that could power better decisions, recommendations, or automation, that data is an asset. Custom software lets you exploit it. Generic SaaS locks it inside someone else's schema.
Compliance or security requirements rule out SaaS. Some industries and some data types can't live in third-party infrastructure. Custom software, deployed in your own environment, is sometimes not optional.
AI Changes the Economics
Two things have changed recently that shift the build/buy balance toward building more often:
Custom software is faster to build. AI-assisted development — used properly by experienced engineers — genuinely accelerates delivery. Projects that took three months now take six weeks. This doesn't eliminate the maintenance cost argument against custom, but it reduces the upfront investment significantly.
AI enables a class of product that didn't exist before. If your custom software incorporates AI — an intelligent workflow, a domain-specific assistant, an automated decision engine — you're building something that no off-the-shelf tool offers. The question shifts from "can we afford to build this?" to "can we afford not to?"
The Hidden Cost of the Wrong Decision in Either Direction
The hidden cost of buying when you should build: Years of workarounds. Shadow spreadsheets that track the things the SaaS tool can't. Manual steps that exist because the system can't handle your edge cases. Staff who've internalised a broken process so thoroughly they don't see it anymore. These costs are real but diffuse — nobody puts them on a spreadsheet.
The hidden cost of building when you should buy: Maintenance. Every custom system is a long-term commitment. It needs updating when your infrastructure changes, patching when security vulnerabilities emerge, refactoring when the team grows. If you build something and then don't invest in maintaining it, it becomes technical debt at the centre of your operations.
A Framework for the Decision
Before deciding, answer these questions honestly:
1. Does this problem touch our competitive advantage? If yes — the way you price, serve customers, or operate differently from competitors — lean toward building. If no, buy.
2. How well do the best off-the-shelf options actually fit? Not "can we make them work" but "do they genuinely fit, without workarounds". If the answer is "mostly, but..." — score this toward building.
3. Do we have the appetite to maintain it? Custom software needs an owner. Not just at build time — ongoing. If you don't have a technical resource who will own this for three years, factor that into your cost model.
4. What's our data strategy? If you're building a data asset, custom software often pays for itself through the analytics, AI, and automation it enables downstream.
5. What's the cost of being wrong? For some decisions, a six-month bet on the wrong approach is survivable. For others, it's existential. The higher the stakes, the more you should invest in getting the architecture right before writing code.
The Honest Answer
Most businesses should buy most things and build selectively. The question is whether your selective choices are the right ones — the systems where your differentiation actually lives.
That clarity is usually the most valuable output of a good software strategy conversation. Not the code that follows — the decision about what to build in the first place.
Framz helps technical leaders make these calls well — and builds the software that follows. Get in touch if you're weighing a build or buy decision.