How do you run an automation audit for a small business?

An automation audit is a short, structured review of how work actually moves through your business: which tasks repeat, which apps they touch, which automations already exist, and where time and errors pile up. For a small business it takes a week of light note-taking plus one or two hours of sorting, and it ends with a ranked shortlist of what to automate, what to fix and what to leave alone. AI tools can help you sort the notes, but nothing replaces writing down what your team really does.

What an automation audit is (and is not)

An audit is not buying a tool or building anything. It answers four questions:

1. What do we do repeatedly? The tasks, how often and how long they take.
2. Which apps hold the data? Your forms, inbox, CRM, spreadsheets, accounting and booking tools.
3. What is already automated? Including the forgotten Zap someone set up two years ago.
4. What should change first? A ranked list, with reasons.

People often skip straight to the fourth question. The first three are what make the answer trustworthy.

Step 1: log a normal week

For five working days, everyone who does admin keeps a simple running list. One line per task:

• What the task was ("copied web enquiry into the CRM").
• Roughly how long it took.
• Which apps were involved.
• What triggered it (an email, a form, a time of day, a phone call).

Keep it rough. A note on a phone or a shared sheet is enough. The point is to capture the dull, invisible jobs people no longer notice, because those are usually the best candidates.

Step 2: list your apps and accounts

Make a second list of every tool the business uses, with three details for each:

• Who owns the account (a person or a shared business login).
• What data lives there (customers, invoices, bookings).
• Whether it connects to anything else (built-in integrations, Zapier, Make, n8n, scripts).

This is also a security check. Every connection between apps is a route into your data. The NCSC's small organisations guide to cyber security recommends strong passwords and two-step verification on important accounts; your audit is a good moment to confirm that each one has them, and that no key account belongs to someone who has left.

Step 3: inventory the automations you already have

Most small businesses have more automation than they think: an email rule, a booking reminder, a Zapier task, a spreadsheet script. For each one, note:

• What it does and what triggers it.
• Who set it up and who gets told if it fails.
• When it last ran successfully (check the run history in the platform).
• Whether anyone still needs it.

You will often find automations that fail silently, duplicate each other, or run under a former employee's personal account. Fixing or switching off those is a quick, safe win before building anything new.

Step 4: score each repeated task

Group the week's log into distinct tasks, then score each one:

1. Hours a month: minutes per task times how often it happens.
2. Rule-based? Could you write the rule on one card? Yes, partly or no.
3. Risk if it goes wrong: low (an internal note), medium (a customer email) or high (money, legal or decisions about people).
4. Stable? Has it been done the same way for at least a few weeks?

Tasks that are high on hours, clearly rule-based, low risk and stable go to the top. Anything that makes significant decisions about people on its own needs extra care: the ICO's guidance on automated decision-making explains the rules, so keep a person in the loop for those.

Step 5: sort into automate, fix, or leave

Every task ends up in one of three columns:

• Automate: frequent, rule-based, stable and low to medium risk.
• Fix the process first: the task is messy, changes every week or depends on one person's memory. Tidy it up, then reassess next quarter.
• Leave human: judgement calls, sensitive customer conversations, money going out.

Take only the top one or two "automate" items forward. Then write each one up as a plan (trigger, steps, data and error handling) before anyone opens a building tool; our guide on what an automation blueprint should include covers that part.

Can AI do the audit for you?

Partly. AI is useful for tidying a messy task log, grouping similar tasks and spotting obvious candidates. It cannot see the jobs nobody wrote down, know which customer emails need a careful human touch, or check who owns your accounts. Treat AI suggestions as a first draft, and keep the final ranking with the people who do the work.

A worked example

A three-person plumbing business logs one week. The log shows about 45 web enquiries, each copied into a spreadsheet by hand (5 minutes each), 30 booking reminders sent manually, and invoices typed into accounting software. The app list reveals that the booking tool can already send reminders, but the setting was never switched on. The automation inventory finds one old Zapier task under a former apprentice's account, failing every day for two months.

The result: switch on the booking tool's built-in reminders (no new build), move the Zapier account to a shared business login and delete the broken task, then plan one new automation for enquiries into the spreadsheet with an instant acknowledgement email. Invoices stay manual because the owner wants to check every one.

Key takeaways

• An audit maps tasks, apps and existing automations before choosing anything new.
• Log one normal week; the dull, invisible jobs are often the best candidates.
• Inventory current automations: many are broken, duplicated or orphaned.
• Score tasks by hours, rules, risk and stability, then sort into automate, fix or leave.
• AI can tidy the notes, but people make the final call.

FAQ

How long does an automation audit take for a small business?

About a week of light logging plus an hour or two to sort and score. Larger teams may need two weeks to see less frequent tasks.

Do I need a consultant for an automation audit?

Not usually. The steps above work for most small businesses. Outside help is worth it when processes cross many apps or nobody has time to log the week.

What if the audit finds nothing worth automating?

That is a useful result. It often means the bigger gain is fixing a process or switching on a feature your existing tools already have.

How often should I repeat the audit?

Once or twice a year, or whenever you add a major tool or change how you sell or deliver.

Once you have picked a task, Automation Blueprint turns a plain-language description of it into a buildable plan: a workflow diagram, a recommended platform with reasons, numbered build steps, the data between steps, error handling, a testing checklist and estimates of build hours and monthly tool cost. It is a planning tool, not an audit or a build service, so it works best after the audit above.

More Automation Blueprint articles