January Price Increase? Let AI Draft Your Customer Notice

Raising prices in January is normal. Telling customers about it badly is what costs you. A vague email, a surprise line on an invoice, or a notice that reads like it was written by a lawyer having a bad day can turn a 6% increase into a cancelled account.

The good news: drafting that notice is one of the safest, most useful jobs you can hand to AI. It's a writing task with a clear goal, a human checks every word before it goes out, and nothing gets sent automatically. Here's how to do it well, and what the rules in Canada actually say.

Why this is a good first job for AI

Most small business owners in Burnaby and across Metro Vancouver write maybe one price notice a year. That's exactly why it's hard. You don't have a template, you second guess the tone, and it sits in your drafts folder until December 30.

A general purpose AI assistant is good at the parts that slow you down:

• Turning your bullet points (old price, new price, effective date, reason) into a clear, warm first draft
• Writing three versions in different tones so you can pick, instead of staring at a blank page
• Shortening a long draft into an SMS or a two line invoice note
• Producing a plain FAQ for your front desk so staff give the same answer every time
• Checking your draft for confusing sentences or a missing effective date

What it shouldn't do is decide the facts. You supply the numbers, the date and the reason. The AI only arranges them.

The rules that shape your notice in Canada

Before you open any AI tool, know the guardrails. Three are worth your attention.

Anti-spam law (CASL). If you email or text the notice, Canada's Anti-Spam Legislation applies. Subsection 6(6) of the Act says the consent requirement "does not apply to a commercial electronic message that solely" does certain things, including providing "notification of factual information about" a customer's "ongoing subscription, membership, account, loan or similar relationship." A plain notice that a member's monthly rate is changing may fit that exemption. The key word is solely. If your AI draft adds "and check out our new premium package!" you may have turned an exempt notice into a marketing message that needs consent. The exemption covers consent only. Subsection 6(2) still requires the message to identify the sender, give contact information and include an unsubscribe mechanism. When in doubt, ask your lawyer.

Honest pricing (Competition Act). The Competition Bureau's drip pricing page (last modified August 24, 2026) puts it simply: "Canadians should be able to trust that the price they see is the one they pay." Its first best practice is "Never promote an unattainable price." So if your new rate comes with a mandatory fee, the notice should state the full price, not a headline number with the fee tucked into the fine print. Government charges like sales tax are the stated exception.

Privacy (PIPEDA and the OPC). The Office of the Privacy Commissioner's principles for generative AI (December 2023) recommend using "anonymized or de-identified information within prompts to a generative AI system rather than personal information." In practice: draft the notice with placeholders like {first_name} and {plan_name}. Don't paste your customer list into a chatbot to "personalize" it. Let your email or billing system merge the names in after you've approved the copy.

A simple workflow that works

Here's the process we'd suggest for a small team. It takes an afternoon, not a week.

1. Write the fact sheet yourself. Old price, new price, which plans or services change, the effective date, how much notice customers are getting, and the honest reason (rent, wages, supplier costs, whatever is true). Add what customers can do: lock in a rate, switch plans, cancel, or call you.

2. Give the AI the fact sheet and a tight brief. Something like: "Using only the facts below, write a customer email announcing a price change. Plain language, under 200 words, friendly but direct. State the new total price and effective date in the first two sentences. Don't add offers, promotions or facts I didn't give you."

That last line matters. It keeps the message closer to the CASL exemption and stops the model from inventing things.

3. Ask for variations, not perfection. Request a warm version, a brief version and an SMS version. Then ask for a five question FAQ for staff.

4. Check every fact against your sheet. Price, date, plan names, phone number. Every one. More on why below.

5. Read it out loud. If it sounds like a press release, ask the AI to "make it sound like a local business owner talking to a regular customer."

6. Send through your normal system. Your email platform handles the unsubscribe link and sender details CASL expects. The AI never touches the send button.

Why a human must check every number

AI writing tools are fluent, and fluency is the trap. The U.S. National Institute of Standards and Technology, in its Generative AI Profile (NIST AI 600-1, July 2024), names "confabulation" as a core risk: "the production of confidently stated but erroneous or false content (known colloquially as 'hallucinations' or 'fabrications') by which users may be misled or deceived." NIST notes this happens because models generate outputs that approximate patterns in their training data, which can produce text that's "factually inaccurate or internally inconsistent."

For a price notice, that might look like a draft that rounds $64.50 to "about $65," moves your effective date from January 15 to January 1, or invents a "loyalty discount" that doesn't exist. Each of those creates a customer service problem, and the wrong price could create a Competition Act problem. The OPC principles make a related point about accuracy and ask developers to disclose "any known issues or limitations about the accuracy of generative AI outputs."

The fix isn't complicated. You own the facts, the AI owns the sentences, and you compare them line by line before anything goes out.

An illustrative example

The following is illustrative, not a real client. A New Westminster yoga studio is moving its unlimited monthly pass from $139 to $149 on February 1 because of a rent increase. The owner writes a five line fact sheet, asks the AI for a warm email and an SMS, and gets drafts in seconds. On review, the owner notices the SMS version says "starting next month" instead of the date, and the email offers a "founding member rate" nobody ever mentioned. They cut both errors, add the real date, and send. Total time: maybe 30 minutes, most of it reviewing. That review step is the whole point.

Where this doesn't apply

AI drafting isn't the right call for every price change. Be honest with yourself about these cases:

• Contract terms. If your customer agreements say how and when prices can change, the contract decides, not the AI. Have a lawyer or your own careful reading confirm your notice matches what you signed. BC consumer protection law may also apply to some contracts, and we're not giving legal advice here.
• Tiny customer lists. If you have 12 key accounts, call them. A personal conversation beats any email, however well written.
• Regulated industries. Utilities, telecom, insurance, financial services and some health services have sector specific rules. An AI tool doesn't know your regulator's latest bulletin.
• Bad news on top of bad news. If you just had a service outage or a staffing problem, a polished AI draft can read as tone deaf. Write that one yourself.
• When you don't have the facts yet. If you haven't decided the new price or the date, AI can't help. It'll just write confidently about nothing.

It's also worth admitting the time saving is modest. You're saving the drafting, not the thinking. If writing comes easily to you, the gain might be small.

The bottom line

A January price increase notice is a perfect small job for AI: low risk, quick to check, and the kind of writing most owners put off. Keep the facts in your hands, keep marketing out of the notice, keep customer data out of the prompt, and check every number before you hit send.

Sources

• Government of Canada, Justice Laws Website. An Act to promote the efficiency and adaptability of the Canadian economy by regulating certain activities that discourage reliance on electronic means of carrying out commercial activities (Canada's Anti-Spam Legislation), section 6. laws-lois.justice.gc.ca
• Competition Bureau Canada. Drip pricing. Page last modified 2026. competition-bureau.canada.ca
• Office of the Privacy Commissioner of Canada. Principles for responsible, trustworthy and privacy-protective generative AI technologies. 2023. priv.gc.ca
• National Institute of Standards and Technology. Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile (NIST AI 600-1). 2024. doi.org/10.6028/NIST.AI.600-1

Planning a price change for January? Autana Solutions helps Burnaby and Vancouver area businesses set up AI drafting workflows with the right guardrails, so your team saves time without sending anything you haven't checked. Book a free call and we'll walk through it with you.