Fichaloto

Blog Operations

What a fragmented stack really costs you

9 min read

Add up the licences and it does not look like much. The cost sits somewhere else, and it never appears on an invoice.

The typical stack, and how it got there

A services firm between twenty and eighty people usually ends up with something like this: timesheets in one application, leave requests in another, payroll in a third, project management in a fourth, and on top of all of it a spreadsheet or a small dashboard that somebody assembles by hand from four separate exports.

None of those decisions was wrong on its own. Each tool arrived to solve a real problem, at a time when the others did not exist or did not reach far enough. The problem is not any one of them. It is that the same person's data now lives in four places and none of them is authoritative.

The tell is simple: somebody on your team downloads a CSV every week. The moment that CSV exists, so does the problem.

The four cost lines

Licences are the small line and the only one anyone looks at. The other three are where the money goes.

  • Duplicated licences. You pay for the same person's record in four systems. Irritating, but it is countable money and it is bounded.
  • Transfer hours. Download, clean, reconcile, paste. Two to six hours a month from someone who knows how — and that is almost always someone expensive, because they are the only person who understands all four formats.
  • Integration. Either you pay for it or your team pays in time. And it is not a one-off: every time either end changes version or schema, somebody has to go back in.
  • Decisions made late. The most expensive line and the one nobody books. An engagement eating its margin gets spotted at month-end rather than in week three, because the data that would have shown it was split across two systems nobody had reconciled yet.

Work out the number for your firm

This does not need a study. Half an hour and four questions gets you a figure good enough to decide on:

  • How many hours a month does someone spend moving data between tools? Multiply by their fully loaded cost per hour, not their salary.
  • How many days pass between something happening — an entry, an absence, an expense — and it appearing in the report where someone looks at it?
  • How many times last year did you find a margin problem when it was already too late to act? Put a rough number on each one.
  • What have you paid over the last two years to connect or reconnect two systems?

Integrate or consolidate

The answer is not always consolidate, and that is worth saying plainly. Integration wins when one of the two systems is non-negotiable — payroll handled by an external provider, a corporate ERP — when the data schema is stable, and when volume justifies the maintenance.

Consolidation wins when several systems need the same master record of people: who works here, on which team, at what cost, allocated to what, away on which days. That record replicated across four systems is what generates the manual work, because there is no way to keep it in step without a human doing it.

That is the deciding question: how many of your systems need to know who is who and what they are doing. One or two, integrate. Four, and you are maintaining four copies of the same truth.

Frequently asked questions

What about payroll? Does that have to come inside too?

Not necessarily, and often it should not — payroll tends to sit with a provider who carries their own liability and their own software. What does need to come out cleanly and without manual work is what payroll consumes: hours, absence, monthly variables. That is the contact point, and it should be an export rather than a count.

Is it not riskier to put everything with one vendor?

A fair concern. What reduces it is not spreading data across tools, which makes the problem worse, but being able to get it out: complete export in an open format, a read API, and no part of the history that exists only inside the product. Ask for that before you consolidate anything.

How long does migration take?

Loading the data is the fast part if you go in the right order: people, teams, clients, projects, allocations. Each level depends on the one before it, and skipping the order means doing it twice. The slow part is people stopping recording things where they used to.

All of this is built into Fichaloto. See how it works.

Keep reading