How to measure what manual procurement coordination actually costs you

Why the cost stays invisible

Procurement reporting typically covers savings, spend under management, supplier counts and cycle time at a high level. None of these show how much human effort was consumed moving work forward.

The cost is real but distributed: five minutes interpreting a request, ten chasing a quotation, fifteen rebuilding a comparison. Individually trivial, collectively the reason the team has no capacity.

Measure one: manual touches per workflow

Count every occasion a person has to act to move one instance of a workflow forward, excluding decisions the organisation deliberately wants a person to make.

Interpreting, chasing, forwarding, consolidating and rekeying all count. Approving does not. The distinction between coordination and judgement is the whole point of the measure.

Measure two: cycle time by stage

Total cycle time hides where the delay is. Broken down by stage, it usually shows that most elapsed time is waiting rather than working — waiting for a supplier, an approver or someone with capacity.

Waiting time is the portion that agent coordination addresses directly, which makes this the measure most likely to move first.

Measure three: first-time-complete rate

What proportion of requests arrive with everything needed to proceed? A low rate generates clarification loops, and each loop adds both effort and elapsed days.

This measure is also a good early indicator, because guided intake tends to improve it quickly and visibly.

Measure four: exception and escalation rate

How often does a workflow leave the standard path, and what happens when it does? A high rate is not necessarily bad — it may mean the rules do not match operating reality.

Reviewed regularly, exception patterns become the most reliable guide to which policies need refining.

Resist the benchmark temptation

It is tempting to compare these figures against industry benchmarks. Most published procurement benchmarks are drawn from inconsistent definitions and self-selected samples, and applying them to a business case usually weakens it.

Your own baseline, measured consistently before and after, is more defensible than any external number — and it is the only comparison an internal audience will actually trust.

Why the cost is invisible

Coordination cost does not appear in any budget line. It is distributed across the working days of people whose salaries are already accounted for, which means nobody ever sees an invoice for it.

That invisibility is why it persists. A cost with no line item has no owner, no benchmark and no pressure to reduce it — and it grows with entity count, supplier count and transaction volume without anyone deciding it should.

How to count it without a consultant

The measurement is straightforward and can be done in an afternoon with one experienced buyer.

• Choose one workflow and one recent, ordinary example of it — not the worst case and not the cleanest.
• List every point at which a person read, decided, chased, rebuilt or retyped something.
• Mark each touch as judgement or coordination. Judgement is work you want people doing.
• Estimate minutes per coordination touch, conservatively, and multiply by frequency.
• Multiply by the number of people performing that workflow.
• Convert at loaded hourly cost. Present the coordination figure separately from the judgement figure.

The second-order costs

Direct effort is the easiest figure to produce and the smallest part of the answer. Three consequential costs are larger and harder to see.

The first is opportunity: categories that reach renewal untested because no week was available to run an event. The second is leakage: terms negotiated and then not applied, because the person raising the order could not see them. The third is exposure: an audit finding that arrives eventually, and arrives with a remediation programme attached.

Making the number defensible

A figure built from your own baseline survives challenge in a way a vendor benchmark never does. Three properties make it defensible.

The inputs are yours and can be questioned by anyone in the room. The arithmetic is visible rather than embedded in a model. And the assumptions are conservative enough that the sceptical finance director in the meeting can revise them downward and still find the conclusion holds.