Mapping Your Processes

The Input-Action-Output Framework

Every process, no matter how complex, breaks down into three parts: something comes in (input), something happens to it (action), and something comes out (output). An invoice arrives, you verify the numbers, you approve it. A customer asks a question, you find the answer, you send a reply.

When you map a process this way, automation opportunities jump out at you. Any step where the action is predictable and the rules are clear? That's automatable.

The 5-Step Process Audit

Step 1: Pick one process you do at least weekly.

Step 2: Write down every single step, even the ones that feel obvious. "Open email" counts. "Click the link" counts.

Step 3: Mark each step as either decision (requires judgment) or mechanical (same every time).

Step 4: Note where data moves between tools — email to spreadsheet, form to database, chat to task board.

Step 5: Score the process: frequency (daily/weekly/monthly) × time per run × number of mechanical steps. Higher score = higher automation priority.

Mapping a Real Process

What People Get Wrong

The biggest mistake is trying to automate your most complex process first. Start with the boring stuff — the processes so routine you could do them half asleep. Those are the ones where automation delivers immediate, obvious value and teaches you the fundamentals without high stakes.

The second trap: skipping the map entirely and jumping to tools. Tools change. Your understanding of your own processes is forever.

A third common trap: over-engineering the map. You don't need a perfect flowchart with every edge case documented. You need a clear, honest list of steps that shows where the automation opportunities live. Perfection is the enemy of progress — especially in process mapping.

Decision Steps Aren't Always Human-Only

When you mark a step as "decision," don't automatically assume it requires a human. AI has fundamentally changed what counts as a decision step. Many decisions that felt like they needed human judgment are actually pattern-matching tasks that AI handles exceptionally well.

Ask three questions about each decision step:

1. Is there a pattern? If the decision follows a recognizable pattern — "emails about refunds go to billing, emails about bugs go to engineering" — AI can learn that pattern and apply it faster than you can.

2. What's the cost of being wrong? If misclassifying one email means it takes an extra 10 minutes to reach the right person, that's low-stakes. AI can handle it. If approving the wrong expense report could cost $50,000, keep a human in the loop.

3. Can it be verified after the fact? Some decisions can be made automatically and then reviewed in batches. The AI routes support tickets all day; a human reviews the routing decisions in a 15-minute daily audit. This gives you speed AND accuracy.

Mapping Multi-Branch Processes

Real processes aren't always linear. Sometimes step 3 has two possible outcomes, and each leads to a different path. A customer inquiry might be a sales question (route to sales team) or a support issue (route to support queue). Your process map needs to capture these branches.

The Impact-Effort Grid

Once you've mapped several processes, you need to decide which to automate first. The scoring formula from earlier gives you raw numbers, but the impact-effort grid adds strategic thinking:

Simple Tools for Process Mapping

You don't need expensive software to map processes. Start simple:

Pen and paper: Seriously. Sketch the flow with boxes and arrows. It's fast, it forces clarity, and it doesn't let you hide complexity behind fancy formatting. Most great workflows started as napkin sketches.

Markdown lists: Write each step as a numbered list item with its type (mechanical/decision) in brackets. Indent sub-steps. This format is easy to share, version-control with Git, and convert into actual workflow code later.

Flowchart tools: When your process has branches (if X then do Y, else do Z), visual flowcharts help. Free options: Excalidraw, draw.io, Mermaid (text-to-diagram). Use these when the linear list format breaks down.

The tool matters less than the act of mapping. A rough map on a sticky note beats a perfect diagram that never gets made.

Measuring Before You Automate

Before automating a process, time it. Actually time it — with a stopwatch, not an estimate. People are notoriously bad at estimating how long tasks take. You think "oh, that report takes five minutes" but when you actually time it, it's 22 minutes including all the context-switching, tool-opening, and data-hunting.

Time tracking gives you two essential things: an accurate ROI calculation for your automation investment, and a baseline to measure improvement against. After building the workflow, time the new process. The difference is your concrete, measurable win — not a guess, but proof.

Track three runs of the manual process before automating. Average them. Write that number down. It's the "before" in your before-and-after story.

Finding the Automations You Didn't Know You Needed

The most obvious automation candidates are the ones you already think of as "tasks." But some of the highest-value automations are hidden in activities you don't even recognize as processes — context switching between tools, searching for information you've looked up before, reformatting data between copy-pastes.

Keep a "friction journal" for one week. Every time you feel annoyed, slowed down, or bored by something you're doing — write it down. At the end of the week, you'll have a list of automation candidates you never would have identified through a formal process audit.

Common hidden processes that people overlook: searching for the same document templates every week, manually updating multiple systems with the same information, checking dashboards that could send you alerts instead, formatting data for different audiences, and tracking deadlines that could auto-remind. These invisible time sinks often add up to more than the obvious tasks.

The friction journal technique works because it captures opportunities in the moment — when the pain is fresh and the process details are clear. Don't rely on memory alone. Write it down when it happens, and by Friday you'll have a prioritized automation backlog that practically writes itself.

Process mapping as a data structure.

processes = [
    {
        "name": "Weekly client report",
        "frequency_per_week": 5,
        "minutes_per_run": 45,
        "steps": [
            {"action": "Pull data from analytics", "type": "mechanical"},
            {"action": "Copy to report template", "type": "mechanical"},
            {"action": "Write executive summary", "type": "decision"},
            {"action": "Add client-specific notes", "type": "decision"},
            {"action": "Export as PDF", "type": "mechanical"},
            {"action": "Email to client", "type": "mechanical"},
        ]
    },
]

# Score each process: higher = automate first
for p in processes:
    mechanical = sum(1 for s in p["steps"] if s["type"] == "mechanical")
    score = p["frequency_per_week"] * p["minutes_per_run"] * mechanical
    hours_saved = (p["frequency_per_week"] * p["minutes_per_run"] * 52) / 60
    print(f"{p['name']}: score={score}, {hours_saved:.0f} hrs/year saved")
    # → Weekly client report: score=900, 195 hrs/year saved

The scoring formula: frequency × time × mechanical_steps. The weekly client report scores 900 (5 × 45 × 4 mechanical steps). That's 195 hours/year — nearly 5 full work weeks. This is how you prioritize what to automate first.

The 5-Step Process Audit

Lesson 2 Quiz