How to use ChatGPT effectively in a small business

The customer email is still sitting there. You opened ChatGPT, typed a few words the way you would type them into Google, and got a smooth paragraph that could belong to any business. If that is the mush on your screen, this is how to use ChatGPT effectively for the admin a small business actually has: the booking change, the quote follow-up, the note that has to sound like you.

The short answer: tell it what the business is, show it one piece of your own writing, let it ask you questions, then edit the draft in the same chat. Read every fact against the original before it goes anywhere.

The settings facts below are from OpenAI’s help pages, checked on 16 October 2026. If a menu on your screen has moved, follow the help page.

Why does ChatGPT give you a mushy answer?

Search rewards a short question. ChatGPT does not look up your shop, your prices, or the way you talk to customers, so a short question comes back polite and useless.

The same tool writes a tighter note once it has what a new person at the desk would need: who you are, who you are writing to, and what done looks like.

If you have not picked the job yet, How to start using AI in your small business is the page for the first weekly task. This page is for when you already have the task and the answer came back mushy.

What should you tell ChatGPT before the task?

Start a new chat. Give it a short brief before you paste the customer’s words.

I run a small [salon / clinic / shop / office] in [town].
Customers write about bookings, prices, and timing.
I write short emails. I do not use jargon.
I do not offer discounts unless I say so in this chat.
Here is the job: draft a reply I can send today.

Then paste the message with the name and phone number taken out. Say what the reply must do. “Tell them Thursday at 2 still works, and ask them to reply if that time is wrong” is a job. “Write a professional response” is not.

How do you show ChatGPT your own writing?

Paste one email or text you were happy to send.

Match the length and tone of this email I already sent.
Do not copy it line for line.
Example:
Hi, Thursday at 2 still works. I have you down for a cut.
Reply if you need a different time.
Thanks, Sam

Take the real customer’s name out of the example. One real sample beats a list of adjectives like “friendly but professional.” If the sample has a price or a date, say whether the new reply may repeat it. Otherwise the draft will borrow last month’s number.

What happens when you ask ChatGPT to ask you questions?

Ask for the questions before the draft. Owners who skip this spend the time fixing a guess.

Before you write anything, ask me up to five questions
you need answered so the reply is accurate.
Wait for my answers.

Answer in the same chat. “We can do Thursday, not Wednesday. The price on the card is $80. We do not waive the fee.” Then:

Using my answers, write the reply.
Keep it under 80 words.
If a fact was not in my answers, write [CHECK THIS] instead of guessing.

The questions show you what the tool does not know.

How do you fix a draft without starting over?

Stay in the same chat. A new chat forgets the brief.

Make it shorter.
Take out the apology.
We do not say "certainly" or "absolutely."
Keep the Thursday time.
Do not add a discount.

Read the next version out loud. Save the brief that worked in a note on your phone.

How do you check the facts before you send it?

You know the booking, the price, and what you promised. The draft does not. Before it leaves:

• Every time, date, price, and address matches what you already know.
• It does not offer a discount, a refund, or a deadline you did not approve.
• Any [CHECK THIS] is replaced or cut.
• It sounds like the sample you pasted, at a length you would actually send.

What should you not paste into ChatGPT?

While you are learning, leave out anything you would not put on a postcard. Use “the customer” instead of a name.

OpenAI’s page on how data is used to improve model performance, checked on 16 October 2026, says that for Free, Plus, and Pro, OpenAI may use your content to train its models. You can turn that off. On the web: open your account menu, select Settings, select Data controls, select Improve the model for everyone, turn it off, and select Done. The same page says you can choose “Do not train on my content” in the Privacy Portal instead. After you opt out, new conversations are not used to improve the models.

That page also says a thumbs up or thumbs down on a response can allow the whole conversation to be used for training, even after you opt out, on Free, Plus, and Pro.

It says OpenAI does not, by default, use inputs or outputs from ChatGPT Business, ChatGPT Enterprise, or ChatGPT Edu to improve its models. On those plans, response feedback does not permit training unless the organization has a separate agreement that allows it.

OpenAI’s data controls page, checked the same day, says turning the training setting off does not delete saved chats. Temporary chat, while it stays temporary, does not appear in your history, does not update memories, and is not used to improve models. Temporary chats may still be retained for up to 30 days for safety purposes. That is not a drawer for a password or a card number.

Canadian prices, and which plan fits one person versus a small team, are in ChatGPT for small business: which plan, and what it costs in CAD. What never goes in the box is in Is ChatGPT safe for my customers’ information?.

Questions owners ask

How do I use ChatGPT effectively if I am not technical?

Brief it like a new person. Say what the business is, paste one sample of your writing with names removed, let it ask a few questions, then ask for a short draft. Check every number against the original before you send it.

Why does ChatGPT give me a generic answer?

A short search-style question leaves out who you are and how you write, so the reply is a polite paragraph that could fit any business. Add a brief and a sample of your writing, then ask again.

Should I paste a customer email into ChatGPT?

Remove names and contact details while you are learning. OpenAI’s help page, checked 16 October 2026, says Free, Plus, and Pro content may be used to train models unless you turn off Improve the model for everyone. ChatGPT Business, Enterprise, and Edu are not used that way by default.

Does turning off model training delete my chats?

No. OpenAI’s data controls page, checked 16 October 2026, says turning off Improve the model for everyone does not delete or hide saved chats. Delete a chat, or the account, separately if you want the words gone.

Which ChatGPT plan should a small business buy?

Canadian dollar prices are on the plan page linked from this article. For training, OpenAI says Free, Plus, and Pro may be used to improve models unless you opt out. ChatGPT Business, Enterprise, and Edu are not used to improve models by default.

What if you want this done at your desk?

If you want to try the brief on a real email, with someone beside you, that is a good use of the free hour. Handoff comes to your place. You leave with one way of asking that you can reuse the next morning. There are no dumb questions.

What the hour covers is on the local AI help page.

Book your free hour.

Sources

• How your data is used to improve model performance, OpenAI Help Center, checked 16 October 2026. Free, Plus, and Pro may train models unless you opt out. Thumbs up or thumbs down can still allow that. Business, Enterprise, and Edu are not used to improve models by default.
• Data controls in ChatGPT, OpenAI Help Center, checked 16 October 2026. Where to turn off Improve the model for everyone, that this does not delete chats, and temporary chat retention of up to 30 days for safety purposes.