How to power your BFCM sales with AI: a playbook for small businesses and SMEs
2026-11-13
If you want to power your BFCM sales with AI, use it on the boring jobs around the sale, not on the sale itself. Small businesses and SMEs that do this get through the weekend calmer, answer faster and have something left over for the follow-up.
That is my opinion, and the rest of this post is the reasoning. It is the opening post in our BFCM 2026 series, so each section below is a part of the playbook that later posts will build on.
Why BFCM is a staffing problem before it is a marketing problem
The crowd is real. Australia Post reported that its last Black Friday and Cyber Monday period set a record, with more than 5.8 million parcels in a single day and about $1.5 billion spent online. It also reported 3.1 million households taking part.
A small-medium enterprise does not get a bigger team for that weekend. You get the same two or three people, a fuller inbox and more orders to pack. Most sales go badly because of slow replies, stock errors and tired staff. A weak offer is rarely the cause.
So the first question is not "what should we post?" It is "which job will break first?" Write that down. Everything below is organised around it.
Playbook part 1: Take the inbox off a person's desk
This is where AI pays back fastest. Sole traders, tradies, clinics and shops all get the same pattern: the same ten questions asked a hundred times. Where is my order? Do you ship to WA? Can I change my booking? Is this still on sale?
Set up a simple sort. Use a tool such as ChatGPT, Claude or the AI features inside your helpdesk or email platform. Have it label each message as an order question, a booking, a refund, a new lead or a complaint. Have it draft replies for the first three. A person reads and sends them. Complaints and anything involving money go straight to a human.
The test I would use: could a new casual answer this with a one-page cheat sheet? If yes, AI can draft it. If it needs judgement or an apology, a person writes it.
Playbook part 2: Write the offer in your voice, then check the maths
AI is good at producing ten versions of an offer in a minute. It is bad at knowing your margin. Both facts matter.
Use it for wording, subject lines and SMS drafts. Feed it three real emails you have sent and ask it to match the tone. Then do the sums yourself, on paper, before anything goes out. A 20% discount on a product with a 30% margin cuts your profit by about two thirds on every sale. Volume has to more than triple just to break even. AI will not warn you about that. I think most small businesses over-discount, and the fix is a spreadsheet, not a prompt.
Rule of thumb: let AI write it, let a human price it.
Playbook part 3: Aim at people you already know
Big retailers win on ad budgets. Small businesses should not play that game. Your best audience is people who have already bought, enquired or asked for a quote.
Export your customer list. Ask an AI tool to group it by what people bought and when. Then write one message per group, not one message for everyone. A tradie might split past customers into "had a quote, never booked" and "had work done last year". A retailer might split into "bought once" and "bought three times". Two or three groups is plenty.
One caution. Do not paste customer details into a free consumer AI tool without checking its privacy settings. Use a business plan, or strip names and emails first.
Playbook part 4: Plan the follow-up before the first sale
The weekend is the cheapest time you will ever have to win a customer, and most small businesses waste it. Buyers arrive, buy and vanish, because nothing was set up to bring them back.
Before the sale starts, decide three things. What will you capture from a buyer: email, consent, what they bought? What is the first message they get after the sale, and when? When do you ask for a review? Have AI draft these messages now, while you are calm. Schedule them. Then on Monday you are reviewing, not scrambling.
What I would skip
I would skip a custom chatbot that answers customers on its own, if you have never run one before. I would skip any new tool you have not tested before the sale. I would skip anything that makes a promise to a customer in your name without a person checking. Sale week is a bad time to learn a new system. Use AI where a mistake is cheap and a person can catch it.
What to do differently this week
Pick the one job that broke last year. Fix that first, with a sort, a set of drafts or a schedule. Price your offer by hand. Plan the follow-up before you launch.
That is where workflow automation fits. It means taking repeat jobs such as sorting enquiries, drafting replies and triggering follow-ups off a person's desk, with a human still checking the parts that matter. Discovery is free, no obligation from the first chat. Bring a list of the three jobs that went wrong or ate the most hours in your last sale, and we can look at which ones are worth automating. Every enquiry gets a reply within 24 hours.
Sources
Frequently asked questions
- Is it too late to use AI for Black Friday this year?
- No. Aim at the jobs that go wrong in sales week, such as sorting enquiries, drafting replies and planning the follow-up. These can be set up in days. Leave big rebuilds until January.
- Where should a small business start with AI for BFCM?
- Start with the job that cost you the most time or goodwill last sale. For most small businesses that is the inbox, or orders and refunds.
- Can sole traders and tradies use a BFCM playbook too?
- Yes. The year-end enquiry spike works like a sale for tradies and contractors. The same steps apply: prepare the inbox, set the offer, and plan the follow-up.
