Day 1: Pick a task, map it, and measure it
Most automation projects die in planning. Someone builds a spreadsheet of "opportunities," debates tools for three weeks, and nothing ships.
Skip that. This week, automate exactly one repetitive task — start to finish — using an off-the-shelf tool. Not a strategy. A working automation you can point to and say "that used to take me an hour a day, now it takes zero."
Pick a task that's rules-based and boring (30 minutes). Don't pick your hardest problem. Pick your most repetitive one. Good candidates:
- Copying new form submissions into a spreadsheet or CRM
- Sending a reminder email when an invoice is 7 days overdue
- Posting a Slack message when a new order comes in
- Moving files from an inbox attachment into a shared drive folder
The test: could you write the rule in one sentence with no "it depends"? "When a new row appears in this sheet, send this email" — yes. "Follow up with the client depending on how the call went" — no, that needs judgment, and it's not ready for automation yet.
If you're not sure which task qualifies, run a quick gut check against a readiness framework before you commit — see Is Your Process Ready to Automate? A Readiness Check. It'll save you from automating something that isn't actually stable yet.
Write down how it works today (30 minutes). Before you touch any software, write the current process as it actually happens — not how it's supposed to happen. Five bullet points is enough:
- New lead fills out the contact form on the website.
- You get an email notification.
- You manually copy name, email, and phone into your CRM.
- You send a "thanks, we got it" email by hand.
- You add a task to follow up in 2 days.
This step matters more than people expect. If the process is actually messier than you think — three different people handle it three different ways — you've found a process problem, not an automation problem. Fix that first. For a fuller walkthrough of this mapping step, see Map It Before You Automate It: A One-Week Process Guide.
Measure the "before" (15 minutes). Time the task, or estimate honestly from a week of doing it. How many minutes per occurrence? How many times a week? Multiply it out.
Example: 5 minutes per lead, 20 leads a week = 100 minutes a week, or about 7 hours a month.
Write this number down somewhere you'll see it again in 30 days. Without a baseline, you won't know if the automation actually helped — you'll just have a vague feeling. For a short framework on capturing this properly, see Baseline or Bust: How to Measure Automation Before You Touch It.
Day 2: Choose your tool and build in a sandbox
For 95% of small-business tasks, you don't need to build anything. Pick a no-code automation platform:
- Zapier — largest app library, easiest for beginners, pricier at scale.
- Make (formerly Integromat) — more flexible logic, cheaper at volume, slightly steeper learning curve.
- Power Automate — a natural fit if you're already in Microsoft 365.
For a first automation, default to whichever one already connects to the tools you use daily. Check that both your trigger app (the form, inbox, or sheet) and your action app (the CRM, Slack, email tool) are on the supported list before you commit. Most are.
Create a free trial account today. Don't overthink the choice — you can always migrate later; a working automation on an imperfect tool beats a perfect tool that's still in the "considering" phase.
Then build the automation, but don't point it at real data yet:
- Set the trigger (e.g., "New row in Google Sheet" or "New form response").
- Set the action (e.g., "Create contact in CRM" and "Send email via Gmail").
- Use a test entry — a fake lead, a dummy invoice — to run it once manually.
- Check every field lands where it should. Names in the name field, not the notes field. Dates formatted correctly.
- Add error handling if the tool supports it: a notification to you (email or Slack) if a step fails, so a silent failure doesn't quietly break your pipeline for two weeks.
This is where most first-timers rush and get burned. Test with fake data three or four times before it ever touches something real.
Days 3–5: Run in parallel, then cut over
Turn the automation on, but keep doing the manual version alongside it for 2–3 days. Compare the outputs. Did the automation catch every lead? Did formatting stay clean? Did anything get duplicated?
This parallel run is your safety net. It's tedious, but it's much cheaper than discovering a bug after the manual backup is gone.
Once you've confirmed a few clean runs, stop doing the task manually. Write a one-page note: what the automation does, which tool it lives in, who owns it, and what to check if it breaks. This isn't bureaucracy — it's insurance against the automation living only in your head. If you're out sick and a lead comes in, someone else needs to know why nothing showed up in the CRM.
Common pitfalls to avoid
- Automating a process nobody agrees on. If three people do the task three different ways, automating "a way" just locks in inconsistency at higher speed. Fix the process first.
- Skipping the baseline. Without a "before" number, you can't prove the win later — and you'll need that proof to justify the next automation.
- No error notifications. A silent failure is worse than no automation at all, because you stop checking manually and assume it's working.
- Trying to automate judgment calls. If the task requires reading tone, weighing exceptions, or making a call, it's not ready. Automate the parts around it — data entry, reminders — and leave the judgment to a human.
- Perfectionism on tool choice. Spending two weeks comparing Zapier vs. Make for a task that takes 10 minutes to build in either is a bad trade. Pick one and move.
Check the result in 30 days
Come back to your baseline number. Did the task's time cost actually drop? Did errors go down? Write down the real result, not a vague impression — a concrete "saved 6 hours this month, zero missed follow-ups" beats "seems better." For a structured way to check this, see Did Your Automation Actually Work? How to Measure It.
If it worked, you now have a template. Pick the next repetitive, rules-based task on your list and run the same five-day process again. Momentum compounds — each automation you ship makes the next one faster to build and easier to trust.