What to automate first: a one-week time audit for a small business

Most small businesses that ask what to automate first already have an answer in mind, and it's often the wrong one. It tends to be the most visible job, or the one a supplier demonstrated well, rather than the one that quietly eats the most hours. A week of honest measurement fixes that. It costs nothing, and it gives you a number to judge every proposal against.

Log one ordinary week

Pick a normal week, not the one before a deadline or a holiday. Ask everyone who does admin to note each task they repeat. Against each one, they write down three things: what started it, roughly how long it took, and where the result went. A shared spreadsheet works. So does a notebook by the keyboard.

The format matters far less than writing it down at the time. People remember the irritating jobs and forget the frequent ones, so a list made from memory on Friday afternoon will be wrong in a predictable way.

By the end of the week you'll have a messy list. Group the entries that are the same job ("logged a repair from an email", "logged a repair from a text"). Then add up the time each group took across the whole team. Multiply that by the working weeks in your year. That figure is the honest size of each problem, in your business, measured by your people.

Rank each task on three questions

For every group on the list, answer three questions.

1. How often does it happen? Daily jobs beat monthly ones, even when the monthly job takes longer. An automation pays back on every run, and a daily job shows you within a week whether it works.
2. Could a careful person write down the rules? If a new starter could do the job from a one-page note, software probably can too. If the job depends on knowing a client for ten years, it can't.
3. What does a mistake cost? A misfiled email costs a minute. A payment to the wrong bank account costs a great deal more. Costly jobs can still be automated, but they need a person approving the final step, and that changes the design (there's more on that in where human approval belongs).

Mark each task high, medium or low against each question. The ones that come out frequent, rule-bound and cheap to get wrong go to the top. You now have a ranked list drawn from your own week, which tells you more than any case study from a supplier.

Why the inbox is usually near the top

Email is where most work arrives in a small business, and handling it follows rules people rarely write down. Someone reads each message, works out what it is and who it's for, forwards it, replies to the sender and files any attachment. That happens many times a day, the rules are clear once you ask, and a misrouted email gets forwarded again at little cost.

Take a twelve-person lettings agency as an illustration. Repair requests land in a shared mailbox. For each one, somebody finds the property from the address and forwards the message to the right property manager. They reply to the tenant to say it's logged, then add a line to the repairs spreadsheet. An automation can do all four of those steps in seconds and keep a record of each. The property manager still decides which contractor to send and how urgent the job is, because that's where a wrong call costs money and goodwill.

None of this needs new software. A business on Microsoft 365 or Google Workspace already pays for the mailbox, the files and the sign-in that an automation like this would use.

Leave these alone for now

Some jobs score badly however much time they take. Put them aside until the easy wins are working.

• Rare jobs, such as an annual insurance renewal. Building and looking after an automation costs more than doing it by hand once a year.
• Processes that are still changing. If the way you price work is changing this year, settle it first and automate the settled version.
• Work that is the relationship: the call to a client who is behind on payment, the reply to a complaint, the conversation about renewing a contract.
• Jobs nobody can explain the same way twice. If the people who do the work disagree on how it's done, agree the steps on paper first. Automating a muddle only spreads it faster.

Turn the top task into a brief

Write one short paragraph for each of your top two or three tasks. Say what arrives, what has to happen to it, what the finished result looks like, who checks it, and how many hours a week your log showed it taking. That paragraph is your brief.

Give it to whoever builds the automation, whether that's someone on your team or a consultant, and ask two things. How will they show it's working after three months? And what will it cost to keep running each month? If the answer to either can't be tied back to the hours in your log, ask again or keep looking. Measure the same task for a week after it goes live, and you'll know whether it was worth doing.

Questions people ask

How long should a time audit take?

One ordinary working week is enough for most small businesses. Each person spends a few minutes a day writing notes, and one person spends an hour on Friday adding them up. A longer audit rarely changes the order of the list, so it mostly delays the decision.

Can an AI tool tell us what to automate?

It can help total and group a messy log, which saves an hour of spreadsheet work. The judgement about what a mistake costs, and which jobs carry a relationship with a client, belongs to people who know the business.

What if the biggest job is also the riskiest?

Automate the preparation and keep a person on the decision. With supplier invoices, for example, software can read and match each one, and a person approves the payment. You still save most of the time, and the risky step stays with someone accountable.

Do we need to buy new software first?

Usually not. Many of the jobs at the top of a typical list can be built inside Microsoft 365 or Google Workspace, which most small businesses already pay for. Start from the tasks, then choose the tools, never the other way round.