Map the Workflow Before You Automate

Start with an example

A counseling practice wants to “automate intake.” Before building anything, the office manager spends 15 minutes mapping how intake actually works today. She discovers:

• Three people handle intake three different ways.
• Insurance questions are sometimes asked by phone and sometimes by email.
• Completed intake forms are saved in two different places.
• Nobody is sure who is supposed to call people back when the form is incomplete.

If they had automated first, they would have automated confusion, and probably lost some new clients along the way. The map took 15 minutes and saved weeks of rework.

The 5 parts of any workflow

Every workflow, from a pizza order to a hospital admission, has the same five parts:

Part	Question	Example
Trigger	What starts it?	A new contact form is submitted
Steps	What happens, in order?	Read form → check service → reply → add to tracker
Decisions	Where does someone choose between paths?	“Is this a service we offer?”
Output	What’s the end result?	Client gets a reply; lead is in the tracker
Data	What information moves, and where?	Name, email, message → inbox, spreadsheet

Common triggers

Most automations start with one of these:

• A form is submitted (contact form, sign-up, survey)
• An email arrives (in a specific inbox, or with a certain subject)
• A time arrives (every Monday at 8 am, the first of the month)
• A record changes (a new row in a spreadsheet, a status set to “Paid”)
• Someone clicks a button (in an app or a chat tool)

You’ll learn more about triggers in the next lesson.

Step 1: Map today, honestly

Write the process as it actually happens now, not how it’s supposed to happen. The best source is the person who does it every day. Sit with them and ask:

• “What’s the first thing you do when this comes in?”
• “Then what?”
• “Where do you look that up?”
• “What do you do when something’s missing?”
• “What do you do differently when it’s busy?”

Write down every tool, every copy-and-paste, and every “then I check…” The small, boring steps are often where automation saves the most time.

Step 2: Label every step

Go through your map and label each step:

• A — Automate: Rule-based, no judgment needed. Example: send a confirmation email, add a row to a sheet.
• D — AI drafts: Needs reading, writing, or summarizing, and a person approves. Example: draft a reply, summarize a request.
• H — Human only: Needs judgment, involves sensitive information, or carries high stakes. Example: pricing exceptions, clinical decisions, complaints about staff.

Step 3: Follow the data

For every step, ask three questions:

1. What information is involved? Names, emails, addresses, payment details, health information?
2. Which tools will touch it? Form tool, spreadsheet, AI model, email?
3. Are those tools approved for that kind of data?

This is where privacy problems get caught early, before a client’s health details end up in a tool that isn’t approved for them. A useful rule: if you don’t need a piece of information, don’t collect it.

Step 4: Look for waste

Mapping often reveals steps that shouldn’t exist at all:

• The same information typed into two places
• Approvals nobody actually reads
• Emails sent “just in case”
• Steps that exist because of a tool you no longer use

Remove waste before you automate. A shorter process is cheaper and easier to automate, and less likely to break.

A worked example: a complete map

Here’s a finished map for a small design studio’s website inquiries:

Trigger: Website contact form submitted

1. Save submission to tracker sheet — A
2. Send “We got your message” email — A
3. Summarize the request in 2 lines — D
4. Draft a personalized reply — D
5. Review and send the reply — H
6. Decide if it’s a good-fit project — H

Decisions: Is this a service we offer? Is the budget realistic?

Output: Client gets a fast, personal reply within a few hours; lead is tracked.

Data: Name, email, project description, budget range. Stored in form tool and Google Sheet. No sensitive data requested.

While mapping, the studio noticed it had been asking for a phone number it never used. They removed that field, which means less data to protect and a shorter form for visitors.

Think deeper: see the work before you change it

Operations experts have a rule: map how work really happens before you try to improve it. They call the current process the “as-is” map and the improved version the “to-be” map. Skipping the as-is map is one of the most common reasons improvement projects fail. Teams end up fixing problems that don’t exist and missing the ones that do.

A method called Lean, which started in car factories, gives names to the waste you’ll spot on a map:

• Waiting: approvals sitting in someone’s inbox
• Rework: fixing mistakes caused by unclear instructions
• Overprocessing: collecting information nobody uses
• Handoffs: places where work passes between people and details get lost

Each one is a chance to simplify before you automate. Automating a messy process just makes the mess happen faster.

Professionals often draw maps with swimlanes: one row per person, team, or app. Swimlanes make handoffs obvious. For AI workflows, add a lane for the AI step so everyone can see exactly where AI content enters and where a human reviews it.

Do the same for data. Data minimization means collect only what you need, keep it only as long as you need it, and limit who can see it. Many privacy laws now require this.

Questions to sit with:

• Think of a process at your school, job, or home that frustrates people. Which type of waste is causing it?
• Why do you think people so often try to fix a process before fully understanding how it works today?

Practice: map your task

1. Choose your top-scoring task from Lesson 1.
2. Interview whoever does it, even if it’s you. Write every step as it actually happens.
3. Fill in the five parts: Trigger, Steps, Decisions, Output, Data.
4. Label every step A, D, or H.
5. Trace the data: what information, which tools, and whether they’re approved.
6. Circle one step you could remove or simplify before automating.

Next, you’ll learn how automation tools think, in triggers, actions, and data, so you can turn your map into something a tool can run.