Support functions: where the lost hours hide

In most SMEs and mid-sized companies, support functions (finance, HR, procurement, IT, general services) account for a modest share of headcount but a large share of time that nobody measures. That time does not vanish in one visible failure. It is diluted into tasks of two, five or ten minutes, repeated thousands of times a year, that no one has ever added up.

This is what makes the subject hard to address. A transformation project targets an identifiable process. The lost hours, however, are rarely in the nominal process. They are at its edges.

Why these hours stay invisible

Three reasons explain it.

The first is organizational. An invoice, a hire or a purchase order passes through three to five people. Each sees their own share as reasonable. No one sees the cost of the full chain.

The second lies in the nature of the work. Support tasks are fragmented, interrupted, and often done on top of something else. They appear in no dashboard.

The third is cultural. In a mid-sized company, resourcefulness is a virtue. The reconciliation spreadsheet someone built six years ago is seen as an asset, until the day that person leaves.

Four places where hours hide

1. Re-keying and reconciliations

The same data is entered in the accounting system, then in a tracking file, then in an HR or project tool. Each entry is brief, but each creates a risk of discrepancy, and therefore a later reconciliation that costs more than the entry itself. In Switzerland, multiple entities, currencies and document languages amplify the effect. US companies with several ERPs after acquisitions will recognize the pattern.

2. Exceptions

The nominal flow is usually well rehearsed. It is the cases outside the frame that consume time: a missing purchase order, a new supplier, an amount that differs from the order, an atypical employment contract. A minority of cases often absorbs half of the time, or more.

3. Approvals and chasing

An approval workflow protects the company, but it also creates waiting, and then reminders by email or chat. The time spent chasing often exceeds the time spent deciding. Reminders multiply when the approver does not know what is expected, or by when.

4. Searching for information and manual reporting

"Where is the latest version?", "Which figure is right?", "Who has the contract?" Add the monthly reports rebuilt by hand from three exports. This work creates no new value, yet it remains necessary as long as the sources are not reliable.

A worked example

Take an industrial company with 300 employees, 40 of them in support functions. The assumptions below are illustrative and should be replaced by your own measurements. We use 1,700 productive hours per full-time equivalent.

Supplier invoices. Assume 6,000 invoices a year. Eighty percent (4,800) are compliant and take 8 minutes on average. Twenty percent (1,200) are exceptions, taking 35 minutes including follow-up. That gives roughly 640 hours for the nominal flow and 700 hours for exceptions, close to 1,340 hours in total. Exceptions are a fifth of volume but more than half of the time.

Monthly close. Assume 15 people each spend 4 hours a month on reconciliations and on correcting data entered elsewhere. That is 720 hours a year.

Searching and chasing. Assume 3 hours a week per person over 45 weeks, for the 40 people in support. That is 5,400 hours.

The total approaches 7,500 hours, about four full-time equivalents out of forty, or in the order of 10% of capacity. This is not a pool one recovers in full. Part of it is normal work. But if a third to a half of it is reachable, the stake is in the order of one to two positions: enough to justify a targeted effort, not enough to justify a large programme.

What this means for leadership

Measure little, but measure properly

There is no need to map the whole organization. Pick two or three flows (invoices, hiring, month-end close) and measure for a few weeks the volume, time per case, share of exceptions and main causes. That is usually enough to show where the substance lies.

Treat the causes of exceptions before the tools

An exception almost always has an upstream cause: an unclear purchasing rule, an incomplete supplier master file, a form that does not enforce the right field. Fixing the cause removes the case. Automating the exception only makes it faster. Our work on automating invoice processing illustrates this: the gain comes first from data quality and rules, and only then from automated document reading.

Place AI where it helps

AI is useful for reading heterogeneous documents, sorting requests, drafting replies or matching approximate labels. It is much less useful for compensating for a poorly designed approval chain. A simple rule: if the problem is a bottleneck in reading or writing, AI can help; if it is a problem of decision or accountability, the organization is what needs work. The same logic applies outside the back office, as our case on field technician routing shows: the gain comes from the assignment rule before the algorithm.

Decide who owns the flow

A flow without an owner degrades. Naming an end-to-end owner, with one simple indicator (lead time, exception rate), changes behavior more than a new tool does. It is also a question of adoption: without the involvement of the people who do the work, gains erode within months, which we discuss in our Organization & adoption approach.

Keep compliance in view

Support functions handle personal data, especially in HR and payroll. Since the revised Swiss Federal Act on Data Protection (nLPD) came into force on 1 September 2023, transparency about processing and control of subcontractors, including AI tools hosted abroad, must be addressed at design stage. European and US groups with Swiss subsidiaries will find a familiar logic here, with its own requirements.

A realistic sequence

In our experience, a sober sequence works: measure over a few weeks, fix the causes of exceptions and simplify the workflows, then deploy a tool or AI on the remaining bottleneck, and finally track two or three indicators monthly so that the gain holds. This path is part of our Operations & processes approach, which runs from strategy to deployment and then to steering results.

If you want to know where, in your own company, the hours that escape measurement today are located, our diagnostic can produce a first quantified reading within a few weeks, with clear priorities and no commitment to a heavy programme.