The goal: turn your backlog into a decision
Most operators carry a mental list of things that bug them — the manual invoice process, the scheduling back-and-forth, the report nobody reads but someone still builds every Friday. The list grows because nothing on it ever gets a clear verdict.
This week, give every item on that list a verdict: Automate, Improve, or Ignore. Not someday — this week. The goal isn't to fix everything. It's to stop carrying twenty vague worries and start carrying three decisions you can act on.
Here's how to do it in five days, with a small pilot as your first real move.
Map and score your friction points (Days 1–2)
Open a blank doc and list every recurring task or process that causes friction. Don't filter yet — include the annoying and the merely inefficient:
- Manually copying leads from email into a spreadsheet
- Scheduling client calls by email tag
- A weekly report that takes two hours to assemble
- Payroll calculations done by hand
- Customer follow-ups that fall through the cracks
Aim for 10–15 items. If you're stuck, walk through a normal week hour by hour and note anywhere you thought "there has to be a better way."
Then answer two questions for each one:
- Is the process itself clear and repeatable, or is it messy and inconsistent?
- How much time or risk does it cost per week?
This is the whole trick. A messy process isn't ready for automation — automating a mess just makes the mess move faster. A clear, repetitive, time-consuming task is exactly what automation is for.
Sort your list into three buckets:
- Automate — repetitive, rules-based, clear steps, meaningful time or error cost. Example: copying form submissions into a CRM.
- Improve first — inconsistent, undocumented, or judgment-heavy. Example: order fulfillment where three people do it three different ways.
- Ignore for now — low impact, or something you're only considering because a competitor does it. Example: a monthly report nobody has opened in six months.
Write one line of justification next to each item. If you can't explain why something is Automate rather than Improve, you don't understand the process well enough yet — go watch it happen once before deciding.
Pilot one automation, not five (Days 2–3)
Resist the urge to tackle your whole Automate list at once. Pick the single item that's clearest and highest-friction — the one where you groan every time you do it.
Run a scoped, one-week pilot:
- Define the trigger and the action in one sentence. ("When a new lead fills out our contact form, add them to the CRM and send me a Slack ping.")
- Pick an off-the-shelf tool — Zapier, Make, or Power Automate for connecting apps; a native feature in your CRM or scheduler if one exists.
- Build the simplest version first. Don't chain five steps together — get the core action working, then layer on extras later.
- Run it in parallel with your old method for a few days. Compare outputs before you drop the manual version entirely.
- Note the time saved and any errors caught, even roughly. You'll need this number later to justify expanding.
For a real-world reference on how a small team runs this exact pilot cycle in practice, see this 30-day automation pilot at a nine-person HVAC company — same principle, slightly longer runway.
Fix the process before touching software (Days 3–4)
For everything in your "Improve first" bucket, don't reach for a tool yet. Do this instead:
- Map the current process in five steps or fewer, on paper.
- Ask whether every step needs to exist. Cut anything that's there out of habit rather than necessity.
- Standardize the parts that vary person-to-person. If three team members do it three ways, pick the best way and write it down as a one-page checklist.
- Only after the process is consistent, ask again whether it's now a good automation candidate. Often it is — cleaning up the process is what makes automation possible six weeks later.
A cumbersome onboarding flow with five forms and three approval emails doesn't need a workflow tool first. It needs two forms and one approval. Fix that, then automate the clean version.
Buy before you build, and write down what you're skipping (Days 4–5)
For your Automate items, check whether an existing tool already solves the problem before considering anything custom. For nearly any small-business need — scheduling, invoicing, CRM, live chat — a subscription tool exists and will be running by tomorrow. Building something custom only makes sense when the process is genuinely unique to your business and central to your edge — which is rare. Spend fifteen minutes checking reviews, pricing, and whether the tool connects to what you already use. Pick something reasonable and move on; you can switch tools later if needed.
For everything landing in Ignore, write one sentence explaining why, and a trigger for revisiting it:
- "We're not automating the weekly competitor report — it's not driving decisions. Cutting it entirely instead."
- "Holding off on a mobile app — clients aren't asking, and our web portal covers it. Revisit if requests come up."
This step matters more than it looks. A written Ignore list turns background guilt into a deliberate decision, and it stops you from re-litigating the same idea every month.
Pitfalls to avoid, and what comes after this week
Watch for these common mistakes as you work through the process:
- Automating a broken process. If your scoring revealed inconsistency, fix the process first. Skipping this is the single most common mistake.
- Piloting too big. A pilot that touches five systems isn't a pilot — it's a project. Keep it to one trigger, one action.
- No parallel run. Cutting over to a new automation without comparing it to the old method for a few days means you won't catch a broken edge case until it costs you a customer.
- Skipping the write-up. If you don't record time saved or errors caught during the pilot, you'll have nothing to point to when deciding whether to expand it.
Once your pilot proves out, don't stop at one automation. Document what worked, extend it to adjacent tasks, and give it an owner so it survives you being busy or out sick — that's the difference between a one-off fix and a system, covered in detail in standardizing your automation win. And before you move to the next item on your list, take ten minutes to lock in what you just built so it doesn't quietly revert — see locking in a win for how.
The point of this exercise isn't a perfect system by Friday. It's a backlog with verdicts attached, one working pilot, and a clear next step — which beats a long list of vague worries every time.