Case Studies: Start Here

The gap between knowing the rules and applying them

You've learned the fundamentals. The Door Rule, the anatomy of a good prompt, the common tips that separate a vague request from a precise one. All of that was foundation work. Theory without application doesn't change how you operate day to day.

This section is where theory becomes practice.

The case studies ahead take you through real prompting scenarios, start to finish. Each one targets a specific type of task that fractional leaders and operators handle routinely. Summarizing a 30-page report. Simplifying a technical document for a non-technical client. Improving a rough draft into a client-ready deliverable. Structuring output as Markdown, XML, or JSON so it plugs directly into your workflow tools.

These aren't hypothetical exercises. They're the same tasks that eat your Tuesday afternoons and Friday mornings across every client engagement you manage.

What each case study covers

Every case study follows the same structure so you can move between them without relearning a format. Here is what to expect inside each one.

Overview of the prompting type. A short explanation of what the technique is, when to deploy it, and what makes it different from the others.

Anatomy of a good prompt, repeated. The framework from the fundamentals section appears in every single case study. This is intentional. Repetition builds the instinct. You'll see the same mental checklist applied against different tasks until the pattern becomes second nature.

Specific tips for that technique. Each type of prompting has its own failure modes and shortcuts. Summarization has different pitfalls than XML formatting. These sections cover what to watch for.

A bad prompt and a good prompt, side by side. You'll see the starting point and the end goal before any explanation happens.

A step-by-step build from bad to good. This is the core of every case study. It walks you through the thought process of iterating on a weak prompt until it produces reliable output. Each step addresses a specific gap: missing context, vague instructions, absent formatting constraints.

Practice challenges. Optional prompts you can try yourself to apply what the case study taught.

Here is what a typical before-and-after looks like across these case studies:

BAD PROMPT:
Summarize this report.

GOOD PROMPT:
You are a fractional CFO preparing a client-ready summary for a
board presentation.

Summarize the attached Q3 financial report into 5 bullet points.
Each bullet should be one sentence. Focus on: revenue growth,
burn rate, and runway. Use neutral, confident language. Flag any
metric that changed more than 15% from Q2.

The gap between those two prompts is the gap between a generic paragraph and a deliverable you can send to a client without editing.

The six case studies and what they solve

You don't need to go through these in order. Pick the ones closest to what you're working on this week, then come back for the rest as those tasks show up in your engagements.

The first three are content transformation tasks. You have something, and you need the model to reshape it. The last three are output formatting tasks. You need the model to deliver its response in a structure that fits your tooling.

Both categories rely on the same foundational principles you've already learned. The difference is in the specific constraints and techniques you apply.

> Which ones to prioritize. If you spend most of your time preparing deliverables for clients, start with Summarization and Improvement. If you're building automations or feeding AI output into other tools, start with JSON and XML. Markdown sits in between and is useful for almost everyone.

How to get the most out of this section

Reading through a case study will teach you something. Actually building prompts alongside it will teach you ten times more.

Follow along with your own content. When a case study walks through summarizing a report, open one of your own reports and try the same steps against it. When it shows you how to improve a rough feature request, pull up a real ticket from one of your client engagements and run the process yourself.

Focus on the iteration, not the final prompt. The finished prompt at the end of each case study is not the point. The point is the step-by-step reasoning that got there. Watching how each addition to the prompt solves a specific problem trains you to diagnose your own prompts when they produce weak output.

Look for the common threads across case studies. After you've gone through two or three, you'll notice the same patterns showing up. Role assignment matters every time. Specificity in the task instruction matters every time. Output format constraints matter every time. These repeated elements are the actual skill. The individual techniques are variations on top of that foundation.

> The fastest way to improve. Pick one case study that matches something on your task list this week. Work through it with real content from an actual client engagement. Apply the principles from the step-by-step section to your own prompt. Compare the model's output before and after. That single exercise will build more prompting intuition than reading all six case studies passively.

Common patterns you'll see repeated

Every case study reinforces the same core framework, applied differently depending on the task. After a few, you'll start recognizing these patterns before we point them out.

Role assignment changes the output quality dramatically. Telling a model to act as a "fractional CMO reviewing a content calendar" produces different results than leaving the role blank. Each case study demonstrates this with a specific role matched to the task.

Constraints prevent generic responses. Word limits, bullet counts, tone requirements, audience specifications. These act as guardrails that keep the model's output tight and usable. Without them, the model defaults to verbose, generic text that requires heavy editing.

Providing the input content inline gets better results than referring to it vaguely. "Summarize this report" loses to "Summarize the following report" followed by the actual text. The case studies show you exactly where and how to include source material in your prompts.

The step-by-step build pattern works for any prompt, not only the six techniques covered here. Once you internalize the process of identifying what's missing from a prompt and adding it systematically, you can apply that same approach to classification, data extraction, tone matching, or any other prompting task that comes up in your work.

The case studies are the training ground. Your client engagements are where the skill compounds.