What a Workflow Cannot Fix

Most of what is wrong with timesheet approval is structural and fixable: routing, windows, exception rules, recording distance, attribution. That is what the preceding notes are about.

The practical lesson in “What a Workflow Cannot Fix” is that a timesheet only becomes reliable through a process people can operate consistently. For teams exploring remote employee monitoring software, the provider website can add time and project context, provided collection is proportionate, access is limited and every consequential inference receives human review.

A residue is not. Four problems arrive at the timesheet process wearing its clothes, cannot be solved there, and become worse when the process is tightened in response.

For a separate benchmark relevant to “What a Workflow Cannot Fix”, consult the OECD hours-worked data. Use it to test record quality, approvals, retention, employee rights and exception handling against the real workflow rather than treating a software report as self-explanatory evidence.

A workload the organisation will not acknowledge

If people are working substantially more than they are contracted to and the organisation's position is that they are not, no recording change will produce honest hours.

People will keep under-recording, because recording accurately contradicts the official position and the cost of doing that falls on them. The data will keep showing adequate capacity. Each round of improvement produces a more precise record of the fiction.

This is the clearest case of a problem that looks like data quality and is about what the organisation is willing to know. The timesheet process is the messenger, and it is the wrong target.

Approval that nobody wants to be real

Where senior people want the appearance of a control without the consequences of one — the escalations, the delays, the occasional refusal, the occasional awkward conversation — every change described here will be adopted and then quietly neutralised.

The exception queue will be bulk-approved. The query mechanism will go unused. The thresholds will be loosened until nothing flags. None of this requires a decision; it happens through a series of small accommodations, each reasonable.

The test is whether anybody has ever been supported after refusing something inconvenient. If not, the control is decorative by consent and the honest move is to say so.

A manager who discourages recording

One manager who makes it known that recording overtime is unwelcome will produce a team whose data is unusable, and no configuration reaches it.

It is also invisible in the standard measures: their submission rate is excellent, their approval intervals may be fine, their corrections are low. What you see is a team with implausibly regular hours and nothing to flag.

This is a management matter and it has to be handled as one. The timesheet data can supply the evidence — comparison with peer teams, with access records, with turnover — and it cannot supply the intervention.

Work that genuinely cannot be measured in hours

Some work does not decompose into hours against codes in any meaningful way. Research with no defined end, creative work, roles that are mostly judgement exercised in short bursts, people whose value is availability rather than activity.

Forcing a timesheet onto it produces a number that satisfies a process and describes nothing. Worse, it is then used — for capacity, for cost, for comparison — as though it meant something.

Where this is the case, the right answer is often to stop collecting, or to collect at a coarser grain that does not pretend to precision. An organisation that records a researcher's week as one line against one project, honestly, has better data than one recording seven precise entries that were invented on Friday.

Three signs you are in this territory

The measurements stop moving despite changes that should have moved them. A proposal is accepted enthusiastically and then nothing is implemented. Or the same specific exception is requested repeatedly by the same part of the organisation.

Each of these indicates that the constraint is somewhere other than the workflow, and the useful response is to stop improving the workflow and to describe plainly what the data now shows and what it cannot. That description is often the most valuable output of the whole exercise, and it is the one nobody asked for.

The common thread

Each of these is a decision the organisation has made and has not written down. That workload will not be acknowledged. That the control is decorative. That this manager will not be challenged. That this work will be measured in units it does not come in.

A timesheet workflow is a mechanism for recording what happened and getting somebody to affirm it. It is good at that, when it is designed properly, and it cannot carry a decision nobody has made.

Which is the useful thing to know at the end of a project to improve one: when the five numbers stop moving, the remaining problem is probably not in the workflow, and the next conversation is with someone else.