The real cost of stitching SaaS tools together.

A small business of fifty people is a perfect target for SaaS pricing. Each tool, on its own, looks cheap. £8 per user per month for a leave tracker. £12 for a timesheet tool. £6 for an expense app. £15 for a survey platform. £10 for a learning portal. Each one is a sensible decision in isolation.

Stack eight of those. £61 per user per month, give or take. Fifty users. £3,050 per month. £36,600 per year. Climbing every year as headcount grows and per-tool prices nudge up. That is a senior engineer’s salary, paid every year, to subscribe to software you do not own and cannot change.

The obvious cost is the visible cost. The deeper costs are the ones nobody puts on a spreadsheet.

First, integration debt. Each tool has its own login, its own export format, its own admin console, its own approval workflow. People copy-paste between them. Spreadsheets are the integration layer. Every quarter someone tries to reconcile the data, fails, and gives up.

Second, ownership debt. When a workflow lives across five tools, nobody owns it. If something breaks, the answer is “it is the tool’s fault”. There is no person who can sit down and fix the workflow end-to-end, because the workflow does not exist as a single thing.

Third, lock-in debt. Each tool holds a piece of your data. Migrating off any one of them is a project. Migrating off all of them is a multi-month effort that nobody has time for. The longer you stay, the deeper the debt gets.

I am not anti-SaaS. There are tools where renting beats owning every time.

Anything where you genuinely need network effects: payroll providers (HMRC integrations, RTI submissions), payment processors, email deliverability infrastructure. Building these yourself is a forever project; renting is sane.

Anything truly commodity: cloud hosting, source control, transactional email. The market has commoditised these to the point where rolling your own is silly.

Anything mission-critical with regulated data: GDPR-heavy verticals, financial compliance. The buying decision is about the SOC 2 report, not about the feature set.

The opposite case is when the workflow is core to your business and the SaaS tools only kind of fit. This is most internal HR-adjacent stuff: leave, attendance, expenses, timesheets, asset registers, document storage, simple analytics.

The argument for custom is rarely about the licence fees. It is about owning the workflow, removing the integration tax, and being able to change the system when the business changes. A small custom internal platform pays back its build cost in about a year on a 30+ person business — and after that it is pure operating leverage.

The other argument is data. When you own the system, your data lives in your Postgres, in your accounts, on your terms. When something needs an analytics dashboard, you write a query. When somebody asks for a new column, you add a column. There is no roadmap to wait for.

For each tool you currently rent, ask three questions. Could a small, focused custom tool do this part as well or better? How much would the workflow improve if it was integrated with everything else you already own? Are you renting because the tool is genuinely best-in-class, or because nobody has had time to think about it?

If the answers point towards custom, the next question is who builds it. Spoiler: it can be a single solo engineer, in months, not years. That is most of what I do.