DEV Community

Otto Brennan
Otto Brennan

Posted on

I Helped 3 Small Business Owners Use ChatGPT. Here's What Actually Worked (And What Didn't)

Most guides on using AI are written by developers, for developers.

But the people who arguably need these tools most — small business owners — aren't going to wade through a Medium post about prompt engineering or a GitHub repo full of example notebooks.

They need to know: can I use this thing without breaking anything?

Over the past few months, I've been helping a few non-technical small business owners figure out how to actually use ChatGPT in their day-to-day work. Not to build anything. Not to automate their backend. Just to do the stuff that eats up their time — writing, responding to customers, figuring out what to post.

Here's what I learned.


The Three People I Helped

Maria runs a small café. She's on her phone constantly but hasn't posted on Instagram in three weeks because "I just don't know what to say." She's also drowning in customer emails.

David is a realtor. He writes a lot — listing descriptions, follow-up emails, newsletters. He's been doing this for 18 years and he's good at it, but it's slow.

Keiko owns a residential cleaning company. Her biggest pain point: new employee onboarding. She was re-explaining the same procedures over and over because nothing was written down.

Three different people. Three different problems. Same tool.


What Worked: The Surprisingly Mundane Stuff

The first thing I noticed: none of them were impressed by what ChatGPT could do.

They were impressed that it just did it.

Maria typed: "Write me 5 Instagram captions for a cozy café on a rainy day. We're in Austin. Friendly tone." She got five captions in about four seconds. She posted one, got thirty-something likes (good for her), and was in my DMs saying "this is magic."

David asked it to write a listing description for a three-bedroom house. The first draft was generic. He typed "make it more warm and family-focused, this neighborhood is known for great schools." Second draft was almost exactly what he'd have written, but in a third of the time.

Keiko described her morning cleaning checklist verbally — just rambling into a voice-to-text tool on her phone — and then pasted the transcript into ChatGPT and asked it to "make this into a step-by-step checklist a new employee can follow." She now has five SOPs she didn't have before.

None of this is technically impressive. But it saved hours.


What Didn't Work: Vague Prompts Get Vague Results

The #1 problem I saw across all three of them: they asked ChatGPT to do things they hadn't fully thought through themselves.

Maria: "Write me a caption." → Generic result, she doesn't like it, gives up.

David: "Write something about the Austin housing market." → He gets a wall of general facts, none of it useful.

The fix is almost embarrassingly simple: be specific. The more context you give, the better the output.

❌ "Write me a caption"
✅ "Write a warm Instagram caption for my café promoting our new lavender latte.
    We're Austin-based, our vibe is cozy and community-focused, under 100 words."
Enter fullscreen mode Exit fullscreen mode

Once they got the hang of this, the outputs improved dramatically. I started calling it the "fill-in-the-blanks" approach — you're basically completing a template, not having a conversation with an oracle.


The Mental Model That Actually Helped Them

Here's the frame that made everything click for all three of them:

ChatGPT is an intern who knows everything but has no context.

It's brilliant but knows nothing about your specific business, your tone, your customers, or what you're going for. Your job is to give it that context. Its job is to use it.

Once they started thinking about it this way:

  • They stopped hoping it would "just know" what they needed
  • They started giving more context upfront
  • They stopped accepting the first draft as the final answer

That last one is big. The first output is almost never what you'd actually use. Type "shorter" or "make it sound less formal" and you'll often get something much better in one extra step.


The Prompt Categories That Actually Get Used

Not all use cases are equal. Here's what they actually kept using after the novelty wore off:

Daily/weekly: Social media captions. Fast, repeatable, immediately useful. Maria now does a weekly batch — she spends 20 minutes on Sunday generating captions for the week.

A few times a month: Emails. Follow-ups, complaint responses, promotions. David uses it for listing announcements and market update emails. He still edits them, but starts from a draft instead of a blank page.

Occasionally: Documents and SOPs. Keiko used it for the onboarding stuff. High one-time value, doesn't need to revisit it often.

Rarely: Long-form content. None of them found it great for things they cared about getting exactly right — like a heartfelt neighborhood spotlight or a personal story in a newsletter. They use it as a starting point at best.


What I'd Tell a Small Business Owner Getting Started

  1. Start with something you hate doing. Not something exciting — something you dread. The email you've been putting off. The Instagram post you can't think of anything to say for. That's where you'll get the most relief.

  2. Give it more context than you think it needs. Your business name, your city, your tone, your customer, your goal. All of it.

  3. Treat it like a first draft generator. Read it, edit it, send it. Not: read it, reject it because it's not perfect, go back to writing everything yourself.

  4. Build a small library of prompts that work. Once you find a prompt that consistently gets you a good result, save it somewhere. That's your reusable template.

  5. Don't try to learn everything at once. Pick one use case — captions, or emails, or documents — and get good at that first.


A Note for Developers Reading This

If you work with clients, have family who runs a small business, or are building tools for non-technical users: this gap is real.

The "AI for business" space is still mostly aimed at people comfortable with technology. The people who could benefit most — the restaurant owner, the realtor, the cleaning company — are left with YouTube tutorials that assume too much.

If you're looking for a product to build or a consulting niche to own: there's a lot of value to be created in the translation layer between "ChatGPT exists" and "here's how to use it for your specific business."


If you want a done-for-you starting point, I put together a pack of 50 ChatGPT prompts specifically for small business owners — social media, email, operations, marketing. No tech skills required. You can find it here — I update it as I find new patterns that actually work in practice.

What's working (or not working) for non-technical people you know? Drop it in the comments.

Top comments (0)