WHEN THREE DEPARTMENTS REPORT THREE DIFFERENT HEADCOUNTS

HR, payroll and finance usually report three different headcounts, and each is correct for its own purpose. A one page bridge ends the argument permanently.

Emmanuel Duncan · 7 September 2026 · 6 min read

Ask a chief executive how many people work in the organisation and you will often get a pause. The pause is usually well founded. HR has one figure from the establishment list, payroll has another from the last run, and finance has a third from the budget. The three are close enough to look like the same number and far enough apart to make anyone reading them uneasy.

This is one of the most common things I am called in to sort out, and it is rarely caused by bad record keeping. It is caused by three departments answering three different questions with the same word.

The Three Questions Hiding Inside One Word

HR counts positions and people attached to them. Payroll counts people who received money in a given period. Finance counts the cost of labour expressed as full time equivalents against a budget line.

Each is correct for its own purpose. The differences follow directly.

Payroll includes people who left mid month and were paid for part of it. HR has already removed them. Payroll includes casual and contract staff paid through the main run. HR often keeps those on a separate list. Finance converts two half time staff into one FTE, so a department with fourteen people can legitimately report twelve. Someone on unpaid leave appears in HR's count, disappears from payroll, and sits in finance depending on how the budget treats them. Someone seconded from one directorate to another is counted twice or not at all, depending on who was asked.

None of that is an error. All of it produces a different number.

Start by Writing the Definitions Down

Before reconciling anything, get the three definitions on one page, in the words the people who produce them would use.

I ask four questions of each figure. As at what moment is this counted, a point in time or an average over a period. Who is in scope, and specifically what happens to casuals, contractors, national service personnel, interns and secondees. Are people counted as heads or as FTE. And what triggers someone leaving the count, the last working day, the last paid day, or the date the exit was processed.

Most reconciliation work is finished by the time those twelve answers are written down. The remaining gap is small and specific, which is the point.

Build the Bridge, Not the Argument

Once the definitions are clear, the reconciliation is a single table that walks from one figure to another and names every step.

Start with the payroll count for the month, because it is the hardest to dispute. Someone was paid or they were not. Then add and subtract, one line per category, until you reach the HR figure. Leavers paid part month, minus. Casuals on the main run, minus. New starters not yet on payroll, plus. Unpaid leave, plus. Each line carries a count and a one line reason.

Then run the same bridge from headcount to FTE, with the part time conversions and the vacancy treatment shown as separate lines.

That table is the deliverable. It ends the argument permanently, because the next time the figures diverge, the question stops being which department is wrong and becomes which line of the bridge moved.

The Categories That Cause Most of the Trouble

Four groups account for the bulk of the difference in the organisations I work with.

Casual and daily rated staff. They are frequently paid outside the main run, sometimes from a departmental budget rather than a central one, and they are often invisible in HR records entirely. In labour intensive operations they can be a fifth of the working population. Decide whether they belong in the headcount, then apply that decision everywhere.

National service personnel and interns. Not employees in the legal sense, present in the building every day, and doing work the organisation depends on. Excluding them from headcount is defensible. Excluding them from workload and span of control analysis is not, so keep them as a named line rather than dropping them.

Vacancies. Establishment lists carry approved positions whether or not anyone occupies them. A directorate reporting its establishment rather than its occupancy will look larger than it is. Report filled, vacant and approved as three separate columns and the ambiguity disappears.

Secondments and shared roles. Agree a single rule, usually that a person is counted where their cost sits, and record the other location as a note.

The Frozen Snapshot

Even with clean definitions, headcount moves daily, and two people running the same query a week apart will get different answers. That alone can restart the argument.

The fix is to freeze a snapshot on a fixed date each month, store it as a dated file that nobody edits afterwards, and report from the snapshot rather than from the live system. Month end is the obvious date because payroll is already anchored there.

The stored snapshots then become the trend line. Pulling twelve of them gives you a headcount history that is genuinely comparable month to month, which a live system query can never provide because it silently rewrites history every time a backdated correction is entered.

What It Costs You to Leave This Unresolved

The obvious cost is credibility. A leadership team that has seen three headcounts will discount the fourth, and that scepticism transfers to every other people figure you present.

The less obvious cost is that nothing built on headcount can be trusted either. Turnover rate has headcount in the denominator. Cost per employee, revenue per employee, span of control, absence rate, training coverage, all of them do. An unresolved headcount definition quietly corrupts every ratio in the people report, and it does so invisibly, because each ratio still looks like a plausible number.

I have seen a turnover figure move by three percentage points purely on the choice of denominator, with no change in the number of people who actually left.

Where to Start This Month

Ask the three departments to send their current figure with no discussion beforehand. Put the three numbers on one page with the gap written out.

Then hold one meeting with the four questions above and write the answers down. Build the bridge table from payroll upward. Circulate it as the agreed basis and put a review date on it for the next financial year.

Then pick the official figure. One number, one owner, one definition, published on one date each month. Everyone else can keep their own count for their own purpose, provided the bridge shows how it reaches the official one.

The organisation does not need three departments to stop counting differently. It needs to be able to explain, in one page, why they do.

Emmanuel Duncan is the founder of Aletheia Lunara Consult, a professional education and consulting practice working with organisations across Ghana and beyond on people analytics, workforce research and organisational capability.