Lesson 3 left you with a facts sheet and checked answers. This lesson turns one real job into a week of marketing: two posts, a short email and a few ad lines. The AI multiplies what you give it. You supply the facts and do the checking.
It also covers two sets of rules that matter more once content is quick to produce: truthful advertising, and Canada's anti-spam law for email and text. You finish with content ready to post, a consent log and one number that shows whether an ad can pay for itself.
Start with one true thing, then make five pieces
Skip the request for "marketing ideas". Start from something true and yours: a job finished this month, a product that just arrived, a question a customer asked this week, the season. Give the AI tool your Lesson 3 facts sheet and plain notes about the job, and pair the result with your own photo of the work. Local detail and proof are the parts an AI tool cannot supply. Where they are missing, it fills the gap with generic wording or makes something up.
From that one job, ask for five pieces:
- a post for your business listing
- a social caption
- a short email
- a new FAQ entry
- an ad angle
Two posts a week is about eight a month. Batch them in one sitting and that is one block of time, not eight separate scrambles.
Describe the job, not the customer. Names, addresses and contact details stay out of your notes and out of the AI tool.
Ad lines, and the number that decides whether an ad can work
Ad platforms mix and test short pieces of text, so ask for several, written to the platform's limits. Google's help page for responsive search ads lists up to 15 headlines of 30 characters and up to 4 descriptions of 90 characters (checked 6 October 2026). Ask for three angles (price, speed and trust) with a character count beside every line, and recount any line near the limit yourself, because AI tools can miscount. Then change one thing at a time, so you can tell what made the difference.
Before you spend, work out two numbers. The figures are examples only. Replace each with your own.
- Cost per lead = ad spend ÷ leads. Example: $300 ÷ 6 enquiries = $50.
- Break-even cost per lead = gross profit per job × close rate. Example: $400 × 1 job won from every 4 enquiries = $100. Gross profit is what the job brings in minus what it directly costs you to deliver, and close rate is the share of enquiries that turn into jobs.
In the example a lead costs $50 and the business can afford up to $100, so the ad has room. If your cost per lead sits above your break-even, better wording will not rescue the ad. Change the offer, the targeting or the landing page, or stop the ad.
Four questions before anything is published
Put every post, email and ad through four questions:
- Is it true? Check each price, date and claim against the facts sheet.
- Can I prove it to a customer standing in front of me?
- Am I allowed to say it in my trade?
- Does it reveal anything about a customer?
The first two are law. The Competition Act prohibits promoting a product or service with claims that are false or misleading in a material respect, and the Competition Bureau weighs the general impression as well as the literal words. Its guidance names specific traps: performance claims with no adequate and proper test done before the claim is made, a made-up "regular price" that dresses up a sale, a headline price that leaves out mandatory fees (government charges such as sales tax aside), and untrue low-stock or deadline claims. For a first violation, civil penalties are the ones Lesson 3 set out: up to $750,000 for an individual and $10 million for a corporation, or three times the benefit gained if that is larger. The Canadian Code of Advertising Standards also says every claim must be supported by competent and reliable evidence.
So strike "best" and "number one" unless you hold the proof. Publish a testimonial only with the customer's written approval and permission, and exactly as they gave it. Never show an AI-made image as your finished work or as a real customer. Platforms add their own rules: Meta says it puts an "AI info" label on ad images created or significantly edited with its generative AI features, or created or edited with third-party AI tools.
For question 3, regulated professions have stricter rules: the College of Physicians and Surgeons of Ontario's advertising policy, for one, bars testimonials and comparisons with other practitioners. If you answer to a regulator, read its advertising rule first. For question 4, the consent principle in federal privacy law (PIPEDA) expects meaningful consent to use personal information, and consent again for a new purpose. Alberta, British Columbia and Quebec have their own private-sector privacy laws, so check which one covers you. A photo taken for the job file is not yet cleared for an ad, so ask before showing a recognizable face, name, house number or licence plate.
This is general information, not legal advice.
Email and text: three anti-spam checks
Canada's anti-spam legislation (CASL) covers commercial messages sent to an electronic address, which the Act defines as an email, instant messaging, telephone or similar account. For you that means marketing emails, texts and direct messages. Each one must pass three checks. Public posts must still meet the truth rules above.
- Consent, with proof. Express consent means the person agreed, verbally or in writing, to receive your messages. It lasts until they withdraw it. Implied consent is time-limited: generally up to two years after a purchase and six months after an enquiry. Whoever claims consent has the onus of proving it, so keep a simple log: who, when, how.
- Identification. Each message names your business and gives a current mailing address plus a phone number, email address or web address, valid for at least 60 days after sending.
- Unsubscribe. Each message offers a clear, no-cost way to opt out, acted on within 10 business days at the latest. If your texts say "reply STOP", honour every STOP.
AI can write the message. It cannot create consent. Under the Act, a message asking for consent is itself a commercial message, so emailing strangers to ask permission does not get you around the rule. The Privacy Commissioner's e-marketing guidance says PIPEDA prohibits harvesting addresses with computer programs, with very limited exceptions, and that you stay responsible for the consent behind any list a supplier hands you. Treat bought and scraped lists as off limits. The maximum penalty per violation is $1 million for an individual and $10 million for a business.
This is general information, not legal advice. The rules have exceptions, so check the official pages linked here or ask a lawyer before you rely on one.
Honest limits
AI makes content cheap for everyone, your competitors included, so volume alone will not win attention. AI copy also drifts toward the same generic phrases and can get details wrong. Three habits keep it useful:
- Check every price, date and claim against the facts sheet.
- Read each piece aloud once. If it no longer sounds like you, rewrite it.
- Judge marketing by enquiries and bookings, not likes.
Try it: One job, one week of marketing (about 15 minutes on a phone)
- Pick one real finished job, product or customer question from the last month.
- Give the assistant approved in your Lesson 2 policy your facts sheet and a few plain notes about the job, with no customer names, addresses or contact details. Ask for a business-listing post, a social caption, a 90-word email, and three ad headlines with one description. The first prompt below does this.
- Run each piece through the four-question check. Strike out any claim you could not prove to a customer standing in front of you.
- Run the three anti-spam checks on the email. Write down exactly which contacts are allowed to receive it and why, and start a consent log with three headings: who, when, how. Keep the log in your own records, not in the AI tool.
- Work out your break-even cost per lead from your own gross profit per job and close rate.
- Schedule or post the two posts, with your own photo of the work. Check first that the photo shows no face, name, house number or licence plate you have not been given permission to show.
- Finish this week: go through your whole contact list. Remove anyone whose consent you cannot account for, and invite contacts whose implied consent is still running to opt in before it runs out. Fix your message footer so it shows your business name, a mailing address, one other contact method and a working unsubscribe. Save the week's content, the consent log and your break-even number as page 4 of your AI playbook.
Prompts you can copy
One job, five pieces
Ad lines to the character limit
Claim check before publishing
Message footer and sign-up wording
Before you move on
Sources checked for this lesson: Justice Laws Website (Government of Canada): Canada's anti-spam legislation, S.C. 2010, c. 23, full text • Justice Laws Website (Government of Canada): Electronic Commerce Protection Regulations (CRTC), SOR/2012-36 • Innovation, Science and Economic Development Canada: Getting consent to send email • Innovation, Science and Economic Development Canada: Texting for good client relations • Innovation, Science and Economic Development Canada: Understanding Canada's anti-spam legislation • Competition Bureau Canada: Misleading representations and deceptive marketing practices • Competition Bureau Canada: False or misleading representations • Competition Bureau Canada: Penalties and remedies for non-compliance • Competition Bureau Canada: Performance claims not based on an adequate and proper test • Competition Bureau Canada: Ordinary selling price • Competition Bureau Canada: Fake urgency cues • Competition Bureau Canada: Use of tests or testimonials • Competition Bureau Canada: Frequently asked questions about Canada's anti-spam legislation • Ad Standards: The Canadian Code of Advertising Standards (The Code Online) • Office of the Privacy Commissioner of Canada: PIPEDA Fair Information Principle 3 - Consent • Office of the Privacy Commissioner of Canada: Summary of privacy laws in Canada • Office of the Privacy Commissioner of Canada: Guidance for businesses doing e-marketing • Google Ads Help: About responsive search ads • Meta: How AI-generated images in ads are identified and labeled on Meta • Meta: Expanding GenAI Transparency for Meta's Ads Products • College of Physicians and Surgeons of Ontario: Advertising (policy)