Mapping Your Decision Landscape

The 147 Invisible Decisions

A marketing director I worked with was asked to list the decisions she makes in a typical week. She thought about it carefully and said “maybe 15-20 major ones.” When she actually tracked them for a week, the number was 147.

Most of those decisions were invisible—quick judgments she made without consciously recognizing them as decisions. Scan an email, decide it’s not urgent. Glance at a report, decide to follow up later. Hear about a problem, decide whether to escalate. Thirty seconds here, fifteen seconds there, dozens of times each day.

She had no idea.

This is the decision blindness problem. You can’t improve decisions you can’t see. And if you can’t see your decisions, you can’t figure out where AI would actually help. You end up like most failed AI implementations—deploying technology that addresses decisions you rarely make while ignoring the ones consuming your actual time.

This chapter teaches you to map your decision landscape. It’s the essential first step before applying AI to your work, and it’s the step almost everyone skips.

Part 1 of this book gave you the mental model—treat AI like an intern, not an oracle. But knowing how to work with AI doesn’t tell you where to apply it. That’s what Part 2 delivers: a systematic approach to finding the right places to deploy AI decision support. And it starts with seeing clearly.

Why Decisions Disappear

Experienced professionals make most decisions automatically. That’s what expertise is—pattern recognition so fast it feels like instinct rather than deliberation. A senior support manager knows which tickets need escalation before consciously analyzing them. A veteran salesperson senses when a deal is going sideways before the metrics show it.

This autopilot is a feature, not a bug. It’s why experienced people are more productive than novices. But autopilot means invisible. When a decision doesn’t feel like a decision, you don’t count it.

There are three types of decision blindness:

Routine blindness. Decisions you’ve made so many times they don’t register. The email triage decisions. The “is this worth my time?” assessments. The constant micro-prioritization happening throughout every workday. You’ve made these decisions thousands of times, so they feel automatic—but each one still requires information and judgment.

Bundled blindness. Multiple decisions disguised as a single task. “Prepare for the client meeting” sounds like one thing, but it’s actually a dozen decisions: What topics to cover? Which data to pull? Who needs to attend? What format to use? How much detail to include? When a task contains multiple hidden decisions, you undercount.

Reactive blindness. Decisions hidden in responses to others. When someone asks a question and you answer, you made a decision—what information to share, how much detail to provide, what tone to use. Reactive work feels like responding, not deciding. But every response required judgment.

These three types of blindness explain why most people dramatically underestimate their decision volume. And this matters for AI because: if you can’t see a decision, you can’t identify the information gap. You can’t define a clear task. You can’t know when AI support would help.

The mapping exercise makes the invisible visible.

The Decision Inventory

The most rigorous approach is a one-week tracking exercise. Keep a running list—notebook, app, voice memo, whatever works. Log every decision, no matter how small. Don’t filter or categorize yet. Just capture.

Days 1-3: Capture everything. Every time you make a choice, note it. “Decided to answer this email now vs. later.” “Decided this report needs more detail.” “Decided to involve Sarah in the conversation.” The volume will surprise you.

Days 4-5: Categorize. Sort your decisions into three buckets:

Information decisions: “What do I need to know?” These include research, analysis, synthesis—any decision that requires gathering or processing information before acting.
Prioritization decisions: “What should I work on next?” These include sequencing, resource allocation, time management—any decision about what gets attention when.
Action decisions: “How should I handle this?” These include execution choices once you’ve decided what to do—the specific approach, format, tone, or method.

Days 6-7: Identify patterns. Which decisions repeat daily or weekly? Which consume the most time? Which cause the most stress or uncertainty? The patterns reveal your decision signature—the recurring choices that define how you spend your cognitive energy.

The Quick Alternative

A week of tracking isn’t realistic for everyone. Here’s a 10-minute version that captures most of the value:

The 10-Decision Sprint:

List 10 decisions you made today (or will make tomorrow)
For each one, note: What information did you need? How long did gathering it take?
Star the three decisions where better or faster information would have helped most

Those three starred decisions are your starting candidates for AI support. They’re high-information decisions that you make regularly—exactly the profile where AI creates value.

The Four Quadrants

Once you can see your decisions, you need to map them. The most useful framework plots decisions on two axes:

Axis 1: Frequency. How often does this decision occur? Daily/weekly (high frequency) or monthly/quarterly/once (low frequency)?

Axis 2: Information Intensity. How much information does the decision require? High intensity means gathering, synthesizing, and analyzing data from multiple sources. Low intensity means the criteria are straightforward—you just need to make the call.

This creates four quadrants:

	High Frequency	Low Frequency
High Information Intensity	THE SWEET SPOT	Worth investment if high stakes
Low Information Intensity	Automate or template	Don’t overthink

The Sweet Spot: High Frequency + High Information Intensity

These decisions happen often enough to justify setup time, and they involve information problems AI can actually solve. Examples:

Reviewing customer feedback for patterns (daily, information-heavy)
Qualifying leads based on multiple data points (daily, information-heavy)
Prioritizing support tickets by urgency and complexity (daily, information-heavy)
Preparing for client calls with relevant context (several times weekly, information-heavy)

When AI helps with these decisions, the time savings compound. Twenty minutes saved on a daily decision is 80+ hours per year.

The Trap: Low Frequency + Low Information Intensity

Building AI workflows for rare, simple decisions wastes effort. If you make a decision once a quarter and the criteria are clear, just make the decision. Don’t over-engineer it.

The Middle Quadrants

The other two quadrants require judgment:

High frequency, low information intensity: These decisions happen often but don’t require much data gathering. Consider simple automation or templates instead of AI. A decision like “which email template to use for follow-up” might just need a quick reference guide, not an AI system.

Low frequency, high information intensity: These decisions are rare but require significant research when they occur. Strategic planning decisions, major vendor selections, annual budgeting—they’re important but don’t happen often enough to justify dedicated AI workflows. Use AI ad-hoc for research support when these decisions arise, but don’t build permanent infrastructure around them unless the stakes justify it.

The sweet spot decisions deserve your setup investment. The others can wait.

How This Looks in Practice

The Operations Director’s Discovery

Kenji manages operations for a property management company. His team had invested $45,000 in an AI tool that promised to “optimize operations.” Six months later, nobody used it.

Frustrated, Kenji did the decision inventory exercise. He predicted 20-25 major decisions per week. The actual count: 147.

The breakdown revealed something he hadn’t expected:

Decision Category	Count	% of Total
Tenant escalation triage	41	28%
Resource allocation	33	22%
Vendor communication	28	19%
Budget micro-decisions	21	14%
Strategic/planning	11	8%
Other	13	9%

Nearly half his decisions were triage and allocation—quick judgments about who should handle what, and how urgently. These were invisible because they happened in 30-second bursts: glance at email, decide priority, assign or handle.

Meanwhile, the $45,000 AI tool was designed for strategic planning decisions that happened maybe 11 times per week. It was optimized for the 8% while ignoring the 50%.

Once Kenji saw this mismatch, he redesigned his approach. For tenant escalation triage—his highest-volume, highest-information decision—he built a simple AI summary that pulled context from four systems automatically: tenant history, property records, service agreements, technician availability.

Triage time dropped from 8 hours per week to 2.5 hours. Response time to tenants improved 40%. The AI investment finally worked—because it was aimed at the right decisions.

The key insight: Kenji’s expensive AI tool wasn’t bad technology. It was mismatched technology. The vendor had designed it for the decisions that sounded important (strategic planning, predictive maintenance) rather than the decisions that consumed actual time (operational triage). Without the decision map, Kenji would have concluded “AI doesn’t work for operations.” With it, he realized “AI wasn’t aimed at the right decisions.”

The Sales Rep’s Sprint

Aisha is an account executive who dismissed AI as “for people who don’t know what they’re doing.” She consistently hit quota but worked 55-hour weeks.

Skeptical but curious, she tried the 10-Decision Sprint. Her 10 decisions from the previous day:

Which prospect to call first
Follow up on stalled deal vs. new pipeline
Talking points for discovery call
Whether to involve solutions engineer
Responding to pricing objection
Which case study to send
Whether to discount to save a deal
What to include in forecast update
Prep for technical buyer demo
Push for CFO meeting now or wait

For each, she tracked the information needed and time to gather it. The pattern was clear: decisions 3, 5, 6, and 9 all required extensive information gathering—30-45 minutes each of searching through CRM, email, Slack, and shared drives.

Aisha started with call prep only. A simple prompt that summarized relevant company news, past interactions, likely objections, and case studies. Instead of 35 minutes of searching, she got a starting brief in 2 minutes that she could verify and adjust.

Call prep time dropped from 35 to 12 minutes. She used the saved time for one extra call per day. After eight weeks, her quota attainment jumped from 108% to 127%—not because AI made better decisions, but because she had more time to make decisions herself.

The distinction Aisha discovered was critical: the decisions stayed human. The information gathering became AI-assisted. She still decided which prospects to prioritize, how to handle objections, when to push for meetings. AI just got her the context faster.

Her initial dismissal—“AI is for people who don’t know what they’re doing”—turned out to be half right. AI wasn’t replacing her expertise. It was removing the tedious information gathering that consumed time she could spend using that expertise. The decision map made this distinction visible.

Connecting the Map to Chapter 3

Remember the three information problems from Chapter 3?

Volume: Too much information to process manually
Velocity: Information changes faster than you can keep up
Variety: Information scattered across formats and sources

Your decision map reveals which decisions suffer from which problems.

Kenji’s triage decisions had a variety problem—information lived in four different systems. Aisha’s call prep had both volume (lots of relevant information existed) and variety (scattered across CRM, email, LinkedIn, company news) problems.

When you look at your high-frequency, high-information decisions, ask: Is this a volume problem? Velocity? Variety? The answer shapes what kind of AI support will help.

For volume problems: AI excels at summarization and pattern detection. If you’re drowning in customer feedback, support tickets, or research documents, AI can surface what matters without forcing you to read everything.

For velocity problems: AI shines at monitoring and alerting. If markets shift, competitors move, or customer behavior changes faster than you can track manually, AI can watch for signals and surface what’s changed since your last review.

For variety problems: AI works well for integration and consolidation. If the information you need lives in multiple systems, formats, or sources, AI can pull it together so you’re not hunting across four different tools to make one decision.

The decision map tells you where to focus. Connecting it to Chapter 3’s framework tells you how AI will help. Kenji’s variety problem needed consolidation. Aisha’s volume problem needed synthesis. Same mapping exercise, different AI solutions—because the information problems were different.

Common Objections

“This seems like a lot of work for minimal payoff.”

The mapping takes a few hours. Most failed AI implementations took months and thousands of dollars before failing. This investment prevents that. Kenji’s $45,000 AI tool sat unused for six months before he did this exercise—a few hours of mapping would have saved six months of frustration.

“I already know where I need help.”

Maybe. But most people discover decisions they’d forgotten about. The structured approach often reveals high-value opportunities hiding in plain sight. Kenji thought his pain point was strategic planning. Turns out it was 30-second triage decisions happening 41 times per week.

“My work is too unpredictable to map.”

Even unpredictable work has patterns. You may not know which fire drill is coming, but “handle unexpected urgent request” is a decision type you can map. The trigger varies; the decision structure often doesn’t.

“I don’t have time to track decisions for a week.”

Use the 10-Decision Sprint. Ten minutes now beats months of misdirected AI effort later. If you truly can’t spare ten minutes, that’s probably a sign your decision landscape needs mapping more than most.

“What if my decisions don’t fit neatly into categories?”

They won’t. Real work is messy. Some decisions will span categories, some will resist classification, some will seem trivial until you notice how often they occur. The point isn’t perfect categorization—it’s visibility. A rough map beats no map. You can refine as you learn.

“Won’t the act of tracking change my behavior?”

Probably, but that’s not necessarily bad. If tracking makes you more conscious of how you spend cognitive energy, that awareness itself has value. The goal is understanding your patterns, not conducting a double-blind study. Be honest about what you’re deciding, and the map will be useful.

Your Monday Morning Action Item

This week, do the 10-Decision Sprint:

List 10 decisions you made today (or will make tomorrow)
For each one, note:
- What information did you need?
- How long did getting that information take?
- How confident were you in the decision?
Star the three decisions where better or faster information would have helped most

Those three starred decisions are your starting candidates for AI support. Bring them to Chapter 5’s scoring framework.

The goal isn’t perfection—it’s visibility. You can’t score opportunities you can’t see. You can’t improve decisions you don’t know you’re making.

Once you can see your decision landscape, you’re ready to evaluate where AI actually fits. That’s what Chapter 5 delivers: a framework for scoring the opportunities your map revealed.