Skip to content
Join 7,000+ leaders following Alastair's work on LinkedIn.

Free skill

Case Study Writer

Draft case studies from interviews that centre the client's actual outcomes. Every figure and quote comes from your source material.

New to skills? A skill is a saved set of instructions your AI assistant reads whenever the job comes up, so you set it up once and it works the same way every time. AI Skills for Teams explains how they work.

Case studies centre the client

A case study is a story about what the client achieved. Many drafts drift into product features, technical details or marketing claims the client never made. This skill keeps the focus on the client's outcomes instead.

It asks you questions one at a time - the challenge they faced, the numbers that prove the solution worked, a quote in their voice - and builds a draft from those real details. The moment a claim needs proof you haven't supplied, it flags it. No invented results, no rounded figures, no scope stretched wider than the facts support.

Why this matters

Case studies sell because they're specific. "Our AI solution improves efficiency" is a claim. "Processing time dropped from 18 hours per week to 2 hours" is evidence. That difference is the whole reason to write case studies.

But specificity is fragile. A detail gets rounded in the draft. A qualifier drops out - "on client work" becomes just "work". A timeframe drifts - "had used in the previous year" becomes "now uses". One by one, your proof disappears and you're left with a persuasive-sounding story nobody can verify.

This skill catches that. It extracts every number, quote, name and outcome from what you tell it, keeps them intact through the draft, and marks anything missing so you can see what still needs to be filled in or checked with the client.

A worked example

Here's what you give the skill and what comes back.

Interview notes: Westmere Financial, a four-person accounting firm. Challenge: invoicing took 3 hours per client per month - data entry, cross-checking, sending. Solution: I taught them a structured prompt for invoice generation. Results: down to 30 minutes per client per month. That's 2.5 hours saved per client per month, 10 hours/month across all clients = 120 hours/year saved. They haven't quoted a cost per hour yet. Client said: "It's freed us up to do what we're actually trained for."

Draft excerpt from the skill: Westmere Financial spent 3 hours every month on invoicing for each client - repetitive data entry and cross-checking that pulled the team away from advisory work. By structuring their invoice workflow with a well-designed AI prompt, they cut the time to 30 minutes per client per month. "It's freed us up to do what we're actually trained for," one partner said. The annual time saving across their whole client base is 120 hours, worth approximately [€X at their billing rate - to be confirmed].

Every figure is exact. The quote is word-for-word. The one gap - the cost basis - is flagged clearly so you know what to ask the client before the draft is final.

The interview

The skill asks seven questions one at a time: client name and industry, the challenge they faced, business impact, your approach, measurable results, a client quote, and the key lesson. It skips any it can answer from what you've already told it, and adapts as you add detail later.

Proof and gaps

Claims from the client are: numbers they've reported, their exact words or paraphrases, their assessment of outcomes, and metrics they track. You provide context (why the problem mattered), explanation (how your solution fit) and next steps separately.

If you say "they saved €50,000" but don't supply the calculation, the draft shows [calculation needed - confirm with client]. The bracket tells you what's not solid enough to publish yet.

When to use it

You've got client interview notes and want them shaped into a case study. Or a client sent feedback you want to turn into a proper story. Or a project's finished and you have metrics - before/after numbers, timescales - and you need a narrative to explain what changed.

What comes next

Take the draft back to the client for review - they need to sign off that the facts, figures and quotes are accurate since their name's attached. Once they've approved it, it's ready to publish. The skill handles the draft and proof-checking; publishing and distribution are yours.

What you still do

The skill doesn't invent results, round figures or estimate outcomes you don't have. But it also doesn't verify whether the client's claim is true. If they tell you they saved €200,000, the skill uses that figure - verification is your responsibility. Similarly, your interview notes may contain confidential information (client names, financial details, internal email). Check what your AI account does with uploaded files before you share them.

How to install it

Download it. The zip below holds the skill as a single folder. Upload it wherever your AI assistant takes skills. In Claude, Skills are on the free plan. In ChatGPT, they're on Business, Enterprise, Healthcare and Edu plans only.

Or copy it. Paste the whole block below into any AI assistant. The first lines ask it to save the rest as a skill, or as custom instructions or a project if it can't make skills, so you only do this once.

Read it first. The skill is one plain-text file with no scripts, so you can read every line before you install it. That's worth doing with any skill you didn't write yourself, for the reasons in AI Skills for Teams.

Then give it what you have. Interview notes, client quotes, metrics, a testimonial - whatever you've gathered. The skill will ask for the pieces it needs and produce a draft from the real details you supply.

More free skills

Before the draft goes to the client, the Source Checker checks every figure and quote in it against your interview notes. When a page or a proposal only has room for a shorter version, the Text Shortener cuts it to fit without dropping a number. The full set is on the Resources page.

The skill

Download the zip and upload it wherever your assistant takes skills. Or copy the whole block below into any AI assistant: the first lines ask it to save the rest as a skill, so you only do this once.

The skill
Please install the skill below for me.

Save everything between the BEGIN SKILL and END SKILL lines as a new skill called "case-study-writer", exactly as written. If you can't create skills, save it as custom instructions or as a project instead. When it's saved, tell me in one line how to use it.

===== BEGIN SKILL =====
---
name: case-study-writer
description: Draft case studies from interviews. Keeps client claims accurate, flags missing proof, avoids invented results. Use whenever you need to shape an interview into a case study, write a client success story, or turn a testimonial into a structured narrative.
---

# Case Study Writer from HumanSpark.ai

Draft case studies from interviews that centre what the client actually achieved. This skill structures the story, asks you for the details it needs, and flags any proof that's still missing.

## How it works

Case studies persuade through specificity. A case study claiming "the client saw big improvements" reads as marketing copy, whereas one saying "response time dropped from 12 hours to 2 minutes" reads as fact - that precision is what matters.

This skill works in two stages. First, it asks you questions one at a time to gather the key details: the challenge, the solution, the measurable results and the client's own words. Then it drafts the case study and flags any claim that needs proof you haven't supplied yet, so you know at a glance what still needs verifying.

## What it requires

Every number, every result, every quote and every outcome must come from you. The skill will not invent figures, estimate improvements or create client testimonials. It marks any missing proof with a bracketed placeholder like `[result needed]` or `[percentage: reduction in X]`, so you can see at a glance what still needs to be filled in or verified with the client.

If a claim matters in your story, you need the evidence. That evidence comes from you: from an interview, a message, a report, an email, metrics the client tracks, or a conversation where they told you the result. Until you supply it, the draft carries a placeholder.

## The structure

A case study follows this shape, and the skill works through each part:

1. **Title and headline** - captures the key benefit in one line.
2. **Executive summary** - one or two paragraphs showing the problem, the approach and the outcome.
3. **The challenge** - what was the client struggling with? Make it relatable.
4. **The solution** - what did you do? Explain in business terms, not technical detail.
5. **The results** - what changed? Measurable outcomes come first (the numbers), then qualitative wins (feedback, testimonials).
6. **Key takeaways** - what does this teach? Why did it work?
7. **Next steps** - what should the reader do?

## The interview

The skill asks you these questions, one at a time:

- What's the client's name and industry?
- What was the challenge they faced?
- What was the business impact of that challenge (cost, time, morale)?
- What approach did you take to solve it?
- What were the measurable results (numbers, timescales, metrics)?
- Do you have a quote from the client about the experience or the outcome?
- What's the most important lesson or takeaway from this project?
- What action do you want readers to take after reading this?

If you've already answered a question in the material you've given the skill, it'll skip that one. If you need to come back and add more detail later, you can. The draft updates each time.

## After the draft

Once you have a draft, get the client's approval before it goes live. A case study is a public statement in the client's name, and they need to sign off on the facts, the quotes and the results. Use their feedback to refine the draft.

The skill will not publish anything. Its job is to get you from an interview to a draft that's ready to check with the client.

## What it won't do

It won't claim credit for results you haven't documented. It won't round a figure, drop a qualifier or widen a scope without flagging it. It won't invent a result to fill a gap. If you say "the client saved money" but don't supply the amount, the draft says `[amount saved needed]`.

And it won't check whether the client's claim is true. If they tell you "we saved €200,000", the draft will use that figure. Verifying it is your job.

## Worked example

Here's what happens with real interview notes.

**Your input:**

Client name: Bellingham & Co (financial services)
Challenge: Manual invoice processing was taking 18 hours per week across their accounts team.
Solution: We built them a structured prompt template + trained the team on how to use it.
Results: Processing dropped to 2 hours per week. That's 16 hours/week saved = 832 hours/year. They haven't told me a cost per hour yet.
Quote: "It's transformed how we work. Before, one team member had to drop everything to process batch invoices. Now it's done in an afternoon."
Takeaway: AI doesn't have to be complex. A simple, well-structured prompt can automate routine work.

**The draft excerpt:**

Bellingham & Co processes invoices manually, a routine but labour-intensive task. Before, the team spent 18 hours each week on data entry and verification. One accountant was often pulled away from client work entirely. Now, with a structured AI prompt workflow, they complete the same work in 2 hours per week. That's 832 hours of freed-up capacity per year.

In their words: "It's transformed how we work. Before, one team member had to drop everything to process batch invoices. Now it's done in an afternoon."

The total annual savings is [cost per hour × 832 hours] - something you'd need to confirm with Bellingham & Co.

Notice: the numbers are precise, the quote is exact, the missing piece is flagged clearly, and nothing is assumed.

## Before you start

Your sources may be confidential - client meetings, internal emails, project reports. Check your AI account's privacy policy before you upload anything, especially if the client is named. The Golden Rule here is: if you wouldn't want it in a public database, don't upload it.

---

*Written with the Case Study Writer, a free skill from [HumanSpark.ai](https://humanspark.ai/case-study-writer/).*
===== END SKILL =====

Let's explore what AI can do for you and your team

  • Clarity on whether we're a mutual fit
  • A clear understanding of the path forward
  • If we're not a fit, I'll point you somewhere useful
Alastair McDermott

25 mins · Free · No obligation

Book a Focus Call