The 15-Point Automation Audit I Run Before Any Client Engagement

Before I write a single line of a workflow, I run an audit. Not because it's billable — because automating the wrong thing is worse than automating nothing. The goal of this checklist is brutally simple: find the tasks that are frequent, manual, and rule-based, because that's where automation prints time and money. Here's the exact 15-point pass I run, usually in under an hour.

Part 1 — Map where the time actually goes

1. 1.List the daily repeaters. What does the team do every single day, by hand? Copy-pasting, data entry, sending the same reply — these are the obvious wins.
2. 2.Find the copy-paste bridges. Anywhere a human moves data from one app to another (form → spreadsheet → CRM) is a workflow waiting to be built.
3. 3.Spot the 'we always forget' tasks. Follow-ups, reminders, status updates — things that depend on someone remembering are perfect for automation.
4. 4.Count the frequency. A 5-minute task done 40 times a week beats a 2-hour task done once a month. Volume is where the hours hide.
5. 5.Tag the rule-based vs. judgment calls. If you can describe it as 'if this, then that,' it's automatable today. If it needs real human judgment, it's not (yet).

Part 2 — Check the data and the tools

1. 1.Is the data structured? Clean fields in a form or sheet automate easily; messy free-text needs an AI step first.
2. 2.Do the tools have APIs or native connectors? Mainstream tools (Google, Slack, HubSpot, Stripe) are easy. A legacy system with no API changes the plan.
3. 3.Where does the single source of truth live? If the same data exists in three places, fix that first — automating on top of chaos just spreads the chaos faster.
4. 4.What triggers the work? A new email, a form submission, a schedule, a webhook? Every automation needs a clear starting gun.
5. 5.What's the volume ceiling? Knowing peak load decides the platform and the pricing model before anything gets built.

Part 3 — Pressure-test the ROI and the risk

1. 1.Calculate the hours saved. Frequency × time-per-task × people doing it. Put a real number on it — that's the business case.
2. 2.Estimate the error cost. What does a manual mistake cost today (a missed lead, a wrong invoice)? Automation's reliability is often worth more than the time it saves.
3. 3.Define what 'broken' looks like. Every workflow needs error handling and an alert. If it fails silently, it'll fail expensively.
4. 4.Check the human handoff. Where should a person still approve or step in? The best automations augment the team — they don't fully replace judgment.
5. 5.Score it: build now, later, or never. High frequency + clear rules + clean data = build now. Anything failing two of those gets parked.

How to use the score

Run every candidate task through points 11–15 and you'll naturally get a ranked list. I build the top one or two first — the ones with the highest hours-saved and the cleanest rules — ship them, prove the value, then move down the list. Trying to automate everything at once is the fastest way to a fragile system nobody trusts.

You don't need 20 automations. You need the three that quietly give you back ten hours a week — and the discipline to build those first.

The one mistake to avoid

Don't automate a broken process. If a workflow is messy, unclear, or changes every week, automation will faithfully reproduce the mess at machine speed. Fix or simplify the process on paper first — then automate the clean version. A good audit catches this before you've wasted a build on it.