Re-keying: the most expensive habit in operations

A small team running on a stack of five SaaS tools has a problem they rarely admit: the tools are cheap individually and expensive collectively. Each one is £30 per user per month. Five of them is £150 per user per month. For a team of fifteen that is £27,000 a year, before integration costs.

The argument for SaaS is real. Off-the-shelf is faster to start, lower commitment, well supported, regularly updated. None of that is wrong. But there is a point at which the stack costs more than the bespoke alternative, and most teams never do the maths to find it.

The integration tax

The first hidden cost. Five tools that do not natively talk to each other create:

• Two or three Zapier-style automation accounts, with retry storms when an API changes.
• One person who knows how the integrations work and is now a single point of failure.
• Re-keying of data between tools, partial automation, partial manual.
• Monthly audit reconciliation between tool A and tool B because the totals do not quite match.

The integration tax in a typical small ops team is around 15-20% of one full-time equivalent. At a fully-loaded staff cost of £45,000, that is £7,000 to £9,000 per year on top of the licence fees.

The data-fragmentation cost

The deeper cost is harder to see, but it dominates the others.

When the customer record lives in HubSpot, the contract lives in DocuSign, the invoice lives in Xero, the support ticket lives in Zendesk and the project lives in Asana, every question that crosses two of those tools requires a person to assemble an answer from multiple sources.

The questions that cross two or more tools, in a typical week:

• What is the lifetime value of customer X? (CRM + Xero)
• How long did the last project for customer Y take, end to end? (HubSpot + DocuSign + Asana + Xero)
• Who has open tickets and what is their support spend? (Zendesk + Xero)
• Which contracts expire in the next 90 days, and what is their renewal status? (DocuSign + HubSpot)

Each of these is a 30 to 90 minute job for someone senior. Multiply by ten such questions a week and the data-fragmentation cost is roughly two days per week of senior time. That is £25,000 to £40,000 per year of opportunity cost.

The escape velocity problem

The third cost arrives when you want to leave a tool. Each SaaS vendor makes export possible but unfriendly. Some hold relationships in opaque internal IDs. Some throttle the export API. Some provide CSVs that lose the relationship structure.

The team going through a SaaS migration is paying somewhere between £5,000 and £30,000 in consultant fees and internal time, every time, and the migrations never quite reach 100% data fidelity.

When custom wins

The cross-over point depends on the team size, the breadth of tools and the criticality of the data. As a rough heuristic, custom starts to make sense when:

• The team is over ten people.
• You run more than four core SaaS tools.
• You spend more than £1,500 per user per year on tooling.
• You have at least one person whose job includes integration babysitting.
• The work has a structure no off-the-shelf tool quite fits, so every tool has been bent into shape with custom fields.

Below those thresholds, the SaaS stack is genuinely the right answer. Above them, the maths starts to flip.

What custom actually costs

The unhelpful headline number people quote for “custom software” is £200,000. That is the price of an enterprise-style six-month build with a London consultancy. It is not the price of what most teams need.

The realistic cost of a custom operations platform for a fifteen-person team, built by a small senior studio:

• Two-week paid diagnostic, fixed price.
• Six to eight weeks of build to a production v1.
• Ongoing flat monthly to host, support and evolve, typically £2-4k per month for a system at this scale.

Total in year one is around £30-50k including the build. Annualised over five years (which is the shortest timeline anyone should think about for a system like this), it is somewhere in the £15-20k per year range.

Compared to a five-tool SaaS stack at £27k a year plus the integration tax and data-fragmentation cost, the breakeven for custom is usually 18 to 30 months.

The decision

The decision is not “SaaS or custom” in the abstract. It is “SaaS for what, and custom for what”. Most well-run teams end up with a hybrid: an off-the-shelf tool for accounting (because Xero is genuinely better than anything you would build) and a custom platform for the work that defines the organisation.

The custom system is the one that holds your operating model. Everything else can be SaaS.

If you are paying for five SaaS tools and rebuilding the same dashboard in three different places, we can do the maths with you. Our diagnostic is two weeks of paid work, ending in a written recommendation. Sometimes it says “keep your SaaS, you don’t need us”; sometimes it says “you would save half the cost in a year”.