Reach the right customers: how small businesses and SMEs can use AI to pick who to target this sale
2026-11-15
The way to reach the right customers this sale is to start with the people who have already bought from you, then use AI to find the pattern in them. Most small businesses and SMEs do the reverse. They pay to shout at strangers and email the same discount to everyone.
This is an opinion, and the reasoning is below. The method that puts it to work is the numbered list in the third section, and you can run it this week.
The crowd is big, but it isn't yours
Black Friday and Cyber Monday are huge in Australia. Australia Post reported record parcel volumes over last year's sales period, up 6.3% on the year before.
It is tempting to read that as "everyone is shopping, so I just need to be visible". Our view is that this is the wrong lesson for small-medium enterprises. Most of those buyers are heading for big retailers with big ad budgets, and a sole trader or small shop will lose a bidding war for their attention every time. The better question is not how to reach more people. It is how to reach the right ones.
To reach the right customers, start with your own records
A tradie has past jobs and quotes. A cafe has loyalty sign-ups. A small online shop has an order history. A bookkeeper or lawyer has a client list. That is better targeting data than any ad platform will sell you, because it describes real people who paid you.
AI is good at one thing here: reading a messy list and telling you what is in it. You don't need a data team. You need an export and a clear question.
Most small businesses, tradies and contractors we talk about this with are sitting on a list they have never properly read. The first job this sale is to read it.
A method you can run this week
Call it the Best 20 test. It takes an afternoon.
- Export your records. Pull your last 12 to 24 months of customers or jobs into a spreadsheet. Include date, what they bought, how much they spent and their suburb or postcode.
- Swap names for a customer number. Give each customer a code such as C001, and keep the sheet matching codes to names separate, on your own computer only. Then delete names, phone numbers and emails from the copy you will paste. The pattern doesn't need them, but the AI does need a code to spot repeat buyers and total each person's spend. Check the tool's privacy settings too, and leave out anything sensitive, such as health information.
- Ask for the top 20%. Paste the data into ChatGPT or Claude. Ask: "Which customer numbers are my top 20% by total spend? What do they have in common? Where are they based?"
- Ask for the lapsed. Ask: "Which customers bought twice or more, but not in the last six months?" These are your most likely targets. Before you message them, remember that email and SMS in Australia need consent and an unsubscribe option under the Spam Act.
- Ask what they bought first. Ask: "What did my best customers buy the first time?" That is your sale hero. Lead with it.
- Write it down. End with one sentence: "This sale, I am targeting [group] in [places] with [product or offer]."
If the AI's answer looks odd, check it. These tools can miscount. Sort the sheet yourself and confirm the top five customer numbers match. It takes two minutes and it is worth it.
Where they are: how tradies, contractors and SMEs pick suburbs and channels
"Where are they" has two answers. One is geography. The other is channel.
For geography, ask the AI to rank your suburbs or postcodes by spend. As an example of a typical tradie, you might find that most of your best work comes from a handful of suburbs. Then a letterbox drop, a local Facebook group post or a suburb-targeted ad beats a state-wide push.
For channel, look at how past customers came to you. Did they reply to an email, book through Instagram, or ring after a referral? Pick one or two channels, not six. Sole traders especially cannot run six channels well in sales week, and half-done campaigns everywhere are worse than two done properly.
Only after this should you think about lookalike audiences on Meta or Google. Our opinion is that they work best when built from a clean list of your best customers rather than everyone who ever bought a discounted mug. But they involve uploading customer details to the platform, so make sure your privacy notice and consent cover it first.
Where we would be sceptical
AI will not tell you who will buy. It tells you who has bought. That is a guide, not a forecast.
Be wary of any tool that promises to find your perfect customers for a login and a monthly fee. If you are weighing one up, run the checks in our piece on AI software sales and the annual discount trap first. You may find a spreadsheet and a chat tool do the job.
For the bigger picture of sales week, including offers, content and admin, see the BFCM playbook for small businesses and SMEs.
What to bring to a first conversation
If you want help working out who to aim at, that is the kind of thinking we do under AI strategy at MILabs. Bring one thing: an export of your last year of customers or jobs, with names and contact details swapped for customer numbers. We can look at it together and talk about what a sensible target looks like for your business. Discovery is free, no obligation from the first chat. Every enquiry gets a reply within 24 hours.
Sources
Frequently asked questions
- Can a sole trader use AI to find the right customers?
- Yes. Export your past customers or jobs into a spreadsheet, swap names for customer numbers, and ask an AI tool such as ChatGPT or Claude to show what your best customers have in common and where they are.
- Should small businesses target new or existing customers on Black Friday?
- In our opinion, start with existing customers. You already have their details and they already trust you. Spend on new audiences only after you have worked through your own list.
- Is it safe to put customer data into an AI tool?
- Only if you check the tool's privacy settings first. Remove names, phone numbers and emails, use customer numbers instead, and keep health or other sensitive information out of consumer AI tools altogether.
