How to cost a manual process before you automate it

The cost of a manual process is the loaded cost of the hours it consumes each month, plus the external help you pay to support it, plus the value of the work it loses. Put that figure together before automating anything, because the price of the work as it stands today is the number every build, tool or hire should be judged against. A business that skips this step ends up judging an automation quote against a feeling.

How do you count the hours?

Track the workflow for a fortnight rather than asking people to estimate, because estimates run low: the five minutes here and there rarely make it into anyone's guess. Count every person who touches the work, including the owner or fee-earner who checks it at the end, and include the interruptions, because a ten-minute task dropped into the middle of focused work often costs twenty minutes of actual output.

Then choose an hourly cost. Use a loaded cost rather than bare salary: employer National Insurance, pension and overhead sit on top of wages, and a common rule of thumb adds a quarter to a third to the salary figure. For general admin, the UK median hourly wage is a sensible floor. For owners and fee-earners, use the real internal cost or the blended billable value of the hour, because that is what the admin is displacing.

What does a worked calculation look like?

Take an illustrative example with round numbers. Suppose a three-person admin team spends 30 hours a month on enquiry intake, at a loaded hourly cost of £22. That is £660 a month. Suppose the owner spends another five hours a month checking and chasing the same work, and the owner's hour is worth £75 in billable or business-building terms. That adds £375, taking the direct labour cost to £1,035 a month.

Now add the external help. Suppose the firm pays an answering service £150 a month to catch the calls the team misses. Then add the leads: suppose ten enquiries a month arrive after hours or while everyone is on jobs and go cold before anyone responds, the firm converts roughly one enquiry in ten, and an average job is worth £900. That is one lost job a month, worth £900.

The total is about £2,085 a month, or roughly £25,000 a year, for a workflow most owners would have guessed cost a few hundred pounds. Every figure above is illustrative, and that is the point: run the same arithmetic with your own numbers, and the shape of the answer is usually similar, because the missed work and the senior time are the parts nobody counts.

What do the hours miss?

Missed or slowly answered leads are usually the largest hidden line, and in many businesses they are worth more than the admin hours themselves. Add the cost of errors: a re-keyed invoice with a wrong amount, a booking entered against the wrong date, and the time spent unwinding each one after a customer notices.

Add delay. Work that waits in an inbox has a cost even when it is eventually done, whether that is a quote going out three days late or a month-end closing a week after it could have. And add key-person risk: a workflow that lives in one person's head carries a cost that appears the week that person is on holiday.

How should the figure be used?

The monthly manual cost reframes the automation decision. With the figure in hand, the question becomes whether a quoted build is a sensible price to remove a cost you can now see and verify, and payback can be stated in months rather than adjectives.

It also disciplines the supplier conversation. A supplier who asks for your manual baseline before quoting is planning to be measured against it. A supplier who quotes without asking is selling technology rather than a result.

Measure the build against the baseline after go-live, using the same arithmetic you used before it. A system that fails to pay back on the hours and the leads it recovers is the wrong build, and the baseline is what makes that visible early enough to act.