How to work out what a manual process is really costing you

Every business has jobs that are done by hand because they've always been done by hand. Somebody rekeys the same figures into a second system. Somebody builds the same report every Monday. Somebody chases the same approval every month.

Nobody costs them, because individually they're small. That's exactly why they survive.

Here's how to put a number on one. It takes twenty minutes and it's usually uncomfortable.

The basic sum

Pick one process. Just one, and a specific one — not "invoicing" but "copying delivery notes into the finance system".

Then:

How long does it take, each time? Ask the person who does it, not their manager. Managers underestimate by roughly half, consistently, because they last did it three years ago.

How often? Per day, per week, per month.

What does an hour of that person cost? Not their salary divided by hours. Add employer's NI, pension, holiday cover, the desk they sit at. A rough loaded rate is fine.

Multiply:

25 minutes × twice a day × 220 working days = 183 hours a year At £28/hour loaded = £5,124 a year

That's the number people expect. It's also the smallest part of the real cost.

The three costs nobody counts

1. Errors, and what they cost downstream

Manual rekeying has an error rate. It's small — but the cost of a single error isn't the thirty seconds to fix it. It's the wrong delivery, the credit note, the phone call, the customer who now checks everything you send.

Ask a different question: "When did this last go wrong, and what happened?" You'll get a story. That story has a cost, and it's rarely small.

2. The wait, not the work

This is the one that's usually biggest, and it's invisible in the sum above.

A task takes four minutes. It sits in someone's inbox for three days. The customer waited three days, not four minutes.

The work is cheap. The waiting is what your customers actually experience, and what they compare you against. Measure elapsed time end to end, not hands-on time — the gap between them is often the whole problem.

3. The person-shaped risk

If one person does it, and only they know how, you have a single point of failure with a holiday allowance.

Ask: "What happens to this if they're off for two weeks?" If the answer is a wince, you've found something more urgent than the hourly cost.

The bit that decides whether to fix it

Now the question that stops you wasting money:

Is this expensive because it's manual, or because it's badly designed?

They need completely different answers.

If the process makes sense and it's just laborious, automation helps. If the process itself is daft — a step that exists because of a system you replaced in 2019, an approval nobody reads, data being copied because two systems were never joined — then automating it makes a bad process faster.

That's the most common and most expensive mistake we see. Somebody buys software to speed up a process nobody ever mapped, and two years later everyone's working around the new system too.

A quick test: ask why each step exists. If three answers in a row are "I don't know, it's always been like that", you're looking at design, not effort.

What to do with the number

Once you've got it, one of three things is true:

Under a few thousand a year, and it works. Leave it. Not everything worth measuring is worth fixing, and something else on your list is worth more.

Expensive, and the process is sound. This is the clean case for automating. You know the number, so you know what it's worth spending.

Expensive, and the process doesn't make sense. Fix the process first. Sometimes the answer is deleting a step, and that costs nothing at all.

Do one this week

Pick the process people complain about most — complaints are a good signal — and run it through the sum. Time it honestly, count the waiting, ask what happens when the person's off.

Most people are surprised twice: by how big the number is, and by how much of it turns out to be waiting rather than working.

---

If you'd rather someone else did the looking, that's what we do — a fixed-price review that follows one process end to end and tells you what it's costing and what to do about it. If we don't find anything worth more than the fee, we'll say so.