What Changes as You Grow

Most timesheet pain is an arrangement that was correct at a previous size and was never revisited.

What breaks, and roughly when

Manual account creation. Fine at ten joiners a year, a permanent tax at ten a month. The fix is provisioning from the identity system.

One global rule set. Works until the second jurisdiction or the second collective agreement.

A single shared project list. Manageable at forty codes, unusable at four hundred without filtering.

A spreadsheet reconciliation. Fine for one pay group, impossible for six.

One administrator with no deputy. Works until they take leave during period close.

Approval by every manager individually. Fine at ten managers, a bottleneck at a hundred, and the point at which exception-based approval stops being optional.

File-based payroll export. Tolerable monthly, painful weekly, and a source of error at scale.

The pattern

Each fails through accumulation rather than through an event.

Nobody notices the day the project list crossed a hundred codes. It is noticed a year later when the catch-all rate is a third of all entries.

Which is the argument for the quarterly data quality check: the value is catching a crossing while it is still small.

The triggers to watch

Catch-all category rate rising — the list has outgrown its structure.

Recording promptness falling — friction has increased somewhere.

Corrections per period rising — a rule or an integration is wrong.

Period close taking longer each month — data quality, and it is the earliest of these signals.

Approval rejection rate near zero — approval has become a click.

Any of these crossing a threshold is the signal that the current arrangement has expired, and acting on it is much cheaper than waiting for complaints.

What to put in place early

Cheap when small, expensive to retrofit.

Provisioning from the identity system, even at low volume.

A rule list maintained outside the product.

A project list with a closure process, not only a creation process.

An automated reconciliation, which is a day of work at any size.

A named deputy administrator.

A tested export, from the first year.

The size-related purchase mistake

Buying for today's headcount when a planned acquisition doubles it, and discovering the tier boundary at renewal.

Buying enterprise capability for fifty people, which is the opposite error and produces a system nobody can configure.

Size for three years, not for one, and ask where the pricing tier boundaries fall relative to that.

The earliest signal

Of all the triggers, one arrives first.

Period close taking longer each month.

It reflects data quality before any other measure moves, because the close is where every upstream problem surfaces as manual work.

Track the hours it takes, which nobody does.

A rising line is the signal to look upstream — at the rules, the reference data, the integrations — rather than to accept it as the cost of growth.

Optimisation needs stable definitions

Growing teams often look for workforce optimization software. Before relying on its outputs, keep role, project and capacity definitions consistent enough to compare one period with the next.