Off Prompt

AI Tools for Small Business

Marketing

Using AI to turn your end-of-project notes into a simple case study you can use on your website or in proposals without a copywriter

How to write a case study small business owners can actually publish — turn project notes into a polished case study using AI in 30 minutes.

Owen Grant 9 min read
Using AI to turn your end-of-project notes into a simple case study you can use on your website or in proposals without a copywriter

You just wrapped a solid project. The client's happy, the results were real, and somewhere in your inbox there's a thank-you email that basically writes itself — and yet, six months later, your website still has zero case studies on it. This post shows you how to write a case study as a small business owner, without a writer, using the messy notes, emails, and scribbles from any finished project. With AI handling the heavy lifting, the whole thing takes about 30 minutes. You don't need to be a writer, and you don't need to hire one.

What you need before you start

Claude{:target="_blank"} — an AI chat tool made by Anthropic that's particularly good at turning rough, unstructured notes into clean, well-organized writing. There's a free tier that works fine for this task; the paid plan (Claude Pro) runs about $20/month if you want faster responses and higher limits.

Time required: About 30 minutes the first time, closer to 15 once you've done it once or twice.

Skill level: If you can copy and paste text into a chat window, you can do this.

Gather your raw ingredients before you write a single word

This is the step most people skip, which is why their case studies never happen. You don't need perfect notes. You need enough to answer four questions: What was the client's problem? What did you do? What changed? What did the client say about it?

  1. Open a blank document (Google Docs, Notepad, anything) and call it "Case Study Raw Notes."
  2. Go through your project files and copy-paste anything relevant into that document. Don't edit it. Just dump it in. You're looking for: the original client email describing their problem, any proposal or scope of work you sent, invoice line items, status update notes or Slack messages, the final delivery email, and any client feedback or thank-you messages.
  3. Add any numbers you remember that might not be in the files — turnaround times, before/after figures, specific deliverables. Even rough ones. "Roughly 40% faster" is better than nothing.
  4. Paste in the client's words wherever you have them. A single quote from an email is gold. You don't need a formal testimonial.

That messy pile of text is your raw material. It's enough. Don't let the roughness of it stop you — the AI's job is to make sense of it, not yours.

How to write a client case study from project notes without a writer

Now you open Claude{:target="_blank"} (or ChatGPT{:target="_blank"} if that's what you have — GPT-4o also handles file uploads, so you can drop a Word doc or PDF straight into the chat instead of copy-pasting). Start a new conversation.

  1. Type a brief setup message to give the AI context before you paste anything. Something like: "I'm going to give you some rough project notes and I need you to write a case study from them. Don't start writing yet — just confirm you're ready and I'll paste the notes."
  2. Wait for confirmation, then paste your entire raw notes document into the chat.
  3. Send the main prompt. This is the one that does the real work. Here's exactly what to use:

Before you paste this, know what it's doing: it's giving the AI a clear format to follow, a voice to write in, and a hard rule not to invent anything. That last part matters a lot — you don't want your published case study containing numbers the AI made up.

Write a case study based only on the notes I've provided. Do not add any facts, figures, or results that are not present in the notes. If something is unclear, leave a [FILL IN] placeholder rather than guessing.

Use this structure:

  • Headline: One sentence that names the client type and the result (anonymize the client name as "a [industry] business" unless I've indicated otherwise)
  • Summary: 2 sentences. The problem and the outcome, plain and direct.
  • The Problem: 2–3 sentences describing what the client was dealing with before we worked together.
  • What We Did: 3–4 sentences describing the work. Focus on the decisions and approach, not just the deliverables.
  • The Results: 3–5 bullet points. Use the specific numbers from my notes. If a number appears, use it — don't replace it with vague language like "improved significantly."
  • Client Quote: Pull the most compelling sentence from any client feedback in my notes. Format it as a pull quote.

Tone: Confident and professional, but not salesy. Write as a small service business talking to a potential client in the same industry. No buzzwords. Plain language.

Target length: 350–400 words for the main body.

After you send this, you'll get a structured draft back in a minute or two. It won't be perfect — but it'll be 80% of the way there, and editing is always faster than writing from scratch.

  1. Read it through once and note anything that feels off, vague, or missing. Then follow up in the same chat: "In the Results section, the third bullet says 'reduced turnaround time' — the notes mention it went from 12 days to 4. Please update that bullet with the specific numbers."
  2. Ask for a longer version if you need one for proposals: "Now rewrite this at 800 words, expanding the Problem and What We Did sections with more detail. Keep the Results bullets tight."

How to handle the one thing AI gets wrong

The most common failure here is vague results language. The AI drafts something like "the client saw improved efficiency and faster delivery times" when your notes clearly say "cut delivery time from 3 weeks to 5 days." It's not making things up — it's just defaulting to safe, general language when it's unsure how to frame something.

The symptom: Your results section reads like a brochure instead of a before/after comparison.

Why it happens: The AI hedges when numbers appear alongside qualifications ("roughly," "about," "we think it was around") — it sometimes strips the number rather than include an approximate.

The fix: Go back to your raw notes, find the specific figure, and say directly in the chat: "The notes say delivery went from 3 weeks to 5 days. Rewrite the second results bullet using that exact comparison." Claude will apply it immediately and cleanly.

This is worth the 90 seconds it takes. Specific numbers are what make a case study convincing — according to Proposify's State of Proposals report{:target="_blank"}, case studies included in proposals increase win rates by an estimated 15–20%. That bump disappears if your results sound generic.

Get client sign-off without the awkward ask

Here's the part that stops most people: asking the client for permission to use their project as a case study. It feels like a big ask, especially if you're not sure they'll say yes.

Flip the sequence. Don't ask first. Draft it first, then send the draft for review.

  1. Finish your AI draft and do a quick edit pass yourself.
  2. Send it to the client with a note like: "Hi [name] — I put together a short case study based on our project together. I'd love to share it on my website if you're happy with how it reads. I've kept your company name out of it for now — happy to add it in if you'd prefer, or leave it anonymous. Let me know if anything needs changing."

That's it. You're giving them something finished to react to, not a blank page to fill. Response rates go up dramatically because reviewing is easy; writing is hard. Most clients are quietly flattered.

How to write a simple case study for your small business website — your 30-minute end-of-project routine

The goal is to make this automatic — something you do at the end of every project before you close the file.

  1. Spend 10 minutes pulling raw notes into your "Case Study Raw Notes" doc
  2. Spend 5 minutes running the prompt sequence in Claude
  3. Spend 10 minutes editing the draft for accuracy and voice
  4. Spend 5 minutes sending the approval email to the client

That's it. The case study sits in a "pending approval" folder until the client replies. Some will take a week, some will take a day. Either way, you've done the hard part.

Over a year of doing this, even once a quarter, you'll have four solid case studies on your website. That alone puts you ahead of the majority of small service businesses — fewer than 30% have even one published{:target="_blank"}, despite case studies being among the most effective content for influencing buying decisions, according to the Content Marketing Institute's B2B research{:target="_blank"}.

Once you have your first one done, the second one takes half as long. Start with the project that had the clearest result — the one where you know exactly what changed. That's your easiest win.

What to do next

Once you've got a case study drafted and approved, the natural next step is getting it in front of the right people. If you want to use it in your proposals without rebuilding it every time, we've written a walkthrough on building a reusable proposal template with AI that pairs well with this.


FAQ

How do I write a case study for my small business without a writer? Gather your raw project notes — emails, scope documents, client feedback, any numbers you have — into a single document. Then use an AI tool like Claude or ChatGPT with a structured prompt to turn those notes into a draft. The prompt in this post will get you 80% of the way there in under 10 minutes. Your job is to check the draft for accuracy, not to write from scratch.

What format should a simple case study follow for a small business website? A headline naming the client type and the result, a two-sentence summary, a short problem section, a what-we-did section, a results section with specific numbers or before/after comparisons, and a client quote if you have one. Keep the whole thing under 400 words for web use — readers scan, they don't read. The prompt in this post follows exactly this structure.

What if my project didn't have clear numbers or measurable results? You don't always need percentages. "Delivered a full brand refresh in 3 weeks instead of the 8-week timeline the client had budgeted for" is a result. "Took over a process the client was spending 6 hours a week on" is a result. Look for time, quantity, or before/after comparisons, even informal ones from emails. Concrete is better than impressive.

Do I have to name the client in the case study? No. Anonymous case studies work fine — and for some clients, they're actually preferable. "A regional HVAC contractor in the southeast" tells potential clients everything they need to know about whether the situation applies to them. Add the client name only if they've explicitly said yes.

Is it okay to publish AI-written content on my website? Yes. Google's guidance on this is clear: AI-assisted content is fine provided it's accurate, edited by a human, and genuinely useful to the reader. Your job is to review the draft for accuracy before you publish — especially the results figures. Don't publish anything the AI wrote that you haven't personally verified against your actual project notes.

What if the client doesn't respond to the approval email? Follow up once after about a week. If you still don't hear back, you have two options: publish an anonymized version that doesn't name them or their company (no approval needed if no one can identify them), or park it and move on to the next project. Don't let one non-responsive client become a reason to skip the whole process.

Was this useful? ·