Skip to content
Tutorials Center
My learning

Topics

Lesson 8 of 36 ยท Automation Concepts

What to Automate: Process Mapping, Scoring and ROI

Decide which tasks deserve automation: map the manual process, find triggers and decisions, score candidates by frequency, effort and risk, and calculate ROI with runnable Python.

The hardest part of automation is not the tool; it is picking the right job and understanding it well enough. Automating a bad or unclear process just produces errors faster. This lesson shows a lightweight method: map the process, find the decisions, score the candidates, then build small.

Zapier cannot run on this site. The Python blocks are stand-ins: calculators and scorers you can edit with your own numbers.

Step 1: Write down the manual process

Pick one job and describe it as steps a new colleague could follow:

  1. A customer fills in the contact form.
  2. Someone reads the message.
  3. They copy the details into the CRM.
  4. They decide who should answer (sales or support).
  5. They send a chat message to that person.
  6. They reply to the customer with a standard acknowledgement.

Now annotate each step: is it manual copying, a decision, a notification, or creative work? Copying and notifying are easy to automate. Decisions can be automated when the rule is clear. Creative or sensitive work usually should not be.

flowchart TD
  A["Form submitted"] --> B["Copy into CRM"]
  B --> C{"Sales or support?"}
  C -->|"keywords: price, buy"| D["Notify sales"]
  C -->|"otherwise"| E["Notify support"]
  D --> F["Send acknowledgement"]
  E --> F

Mapping on paper first reveals the trigger, the decision rule and the actions before you open any tool.

Step 2: Find the trigger, data and rules

For the process above:

  • Trigger: new form response.
  • Data needed: name, email, message.
  • Rule: if the message mentions pricing or buying, route to sales; else support.
  • Actions: create CRM record, post to the right channel, send an email.

If you cannot state the rule clearly, stop. Clarify with the people involved first. Unwritten rules ("it depends") are the main cause of automations that nobody trusts.

Step 3: Score the candidates

Good candidates are frequent, repetitive, rule-based, low risk if wrong and have apps with connectors. A simple scoring model rates each on a scale of 1 to 5 and combines them.

Python Stand-in: score automation candidates

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The low-rated "reply to angry customers" is exactly the kind of job a human should keep.

Step 4: Estimate ROI

Return on investment compares time saved with time spent. Include everything: building, testing, monitoring and fixing. Also consider error reduction and speed benefits, which can be worth more than minutes saved.

Python Stand-in: ROI and break-even

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Currency is whatever you choose. Rare jobs often lose because platform and maintenance costs outweigh the minutes saved.

Step 5: Build small and iterate

Do not build the full ideal flow at once. Build the thinnest slice that delivers value (form to CRM), run it for a week, check Zap History, then add routing, alerts and clean-up. Each extra step is a new place to fail.

Questions before you automate

  • What is the trigger, and is it reliable?
  • What does success look like? How will you notice a silent failure?
  • What happens if it runs twice, or never?
  • Who owns this Zap when its builder leaves?
  • Does it touch personal or sensitive data? Do you have the right to automate it?
  • Is there a manual fallback?

When not to automate

  • The process changes every week.
  • The rule is vague, or the outcome affects people's money, health or employment without human review.
  • The task takes two minutes a month.
  • A simple native feature of the app (a built-in rule or integration) already does it.
  • You do not understand the process yet. Do it by hand a few times first.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. What is the first step before building an automation?

    Mapping the manual process reveals the trigger, data, rules and actions.
  2. 2. Which job is the WORST candidate?

    Sensitive, judgment-heavy work with unclear rules should stay with humans.
  3. 3. Which cost should an ROI estimate include besides build time?

    Ongoing maintenance and platform fees often decide whether a job pays back.
  4. 4. A good build strategy is to:

    Small slices deliver value early and limit the places where things can fail.

Summary

  • Map the manual process first; identify trigger, data, rules and actions.
  • Score candidates by frequency, repetition, clear rules, safety and connectors.
  • Calculate ROI including maintenance and platform costs.
  • Build a thin slice, watch it, then extend.
  • Keep humans on sensitive, vague or rarely repeated work.

Next: Choosing Triggers and Apps