What a Working Setup Looks Like

Rather than a summary, a description of the arrangement these notes point toward.

The selection

A written rule list, maintained outside the product, with the product as an implementation of it.

Requirements including explicit exclusions: no screen capture, no activity scoring, no automatic idle deduction, no silent manager edits.

A trial configured by you, on your own hard cases, with criteria written first.

References at your size, using your module, spoken to by an administrator.

A three-year total cost, not a per-seat figure.

The data

Times stored to the minute, unrounded, with rounding applied at calculation.

Start and end as instants, with zones handled explicitly.

Created-at separate and immutable.

A complete edit trail, exportable.

No hard deletion.

Rule versions recorded against calculated values.

The integrations

One system owns each kind of reference data; the timesheet system owns entries and nothing else.

A written boundary with payroll: which system applies which rule, with one answer to "why was this person paid this".

A reconciliation every period: sent, received, paid.

Integration runs that log attempts, successes and failures, and alert on zero records.

Service accounts, scoped and rotatable.

The implementation

Rule discovery before evaluation.

Three parallel periods, compared line by line, signed off by payroll in writing.

History either migrated with its audit trail intact, or archived readable — never a degraded copy called the record.

Ten-minute training, because the configuration allows it.

Four to eight weeks of stabilisation capacity.

The operation

A named owner with allocated time, and a deputy who can follow the runbook.

A period-close checklist with six checks and a defined response to each.

Four data quality measures, reported quarterly with the analysis.

A twenty-minute check after every vendor release, including the capture features.

A quarterly permissions review, reported including clean results.

Retention per category and jurisdiction, automated, verified, including backups.

The exit position

An export tested annually, not at renewal.

Rules documented independently.

Integrations on your side of the boundary.

Analysis from an extract rather than inside the product's report builder.

Switching cost measured once a year, which turns a renewal from an announcement into a negotiation.

What it produces

Payslips that are right, records that are defensible, and a renewal you can walk into with a position.

None of this requires an unusual budget, and all of it requires deciding what you need before deciding what to buy.

What it costs to run

Ongoing, once live, and smaller than most expect.

A named owner with allocated time, and a deputy.

A period-close checklist run every period.

A quarterly data quality check and permissions review.

A twenty-minute check after each vendor release.

An annual export test, retention verification and rule list update.

None of this is a large commitment, and all of it separates a system producing correct payslips from one producing plausible ones.

Reviewing it annually

Six questions that describe whether the arrangement still works.

Is the rule list current and maintained outside the product?

Has the export been tested this year?

Has a permissions review been done and reported?

Did the last upgrade check find anything, and was it recorded?

Is period close taking longer than a year ago?

Would you buy this product again?

The last one produces the renewal decision and nobody asks it unless the process does.

Fit the system to creative delivery

A healthy setup records enough detail for costing without interrupting the work itself. That balance is central when evaluating time tracking for designers.