It used to be that someone looking for an accountant in Trois-Rivières typed "accountant Trois-Rivières" into Google, looked at the top results and clicked. Today, more and more people put their question to ChatGPT or Perplexity, or simply read the summary Google shows at the top of the results. And they do not always click any more.
That changes the game for SMBs. If the AI answers on the customer's behalf and never names you, you do not exist in the conversation, however good your site is. This is called GEO, generative engine optimization. The goal is no longer just to appear in a list of links, it is to be the source the AI cites when it answers. Here is how.
What actually changes
Classic SEO ranked you in a list of ten blue links. The visitor chose. Now the AI reads dozens of pages, synthesizes an answer, and cites two or three sources. You are no longer competing for a rank, you are competing to be one of the sources judged reliable on a specific question.
The direct consequence: traffic to your site can fall even as your visibility rises. Someone can learn your name, your specialty and your city through ChatGPT without ever clicking, then call you directly. Old dashboards that only count clicks will lie to you. What counts now is being named.
Another change: AI likes clear, self-contained answers. It struggles with a page that circles the point. It easily lifts a paragraph that answers a question head-on. The structure of your content matters as much as its quality.
Answer specific questions clearly
First thing to do: write content that answers real questions, phrased the way your customers ask them. Not "Our plumbing services", but "What does replacing a water heater in Laval cost?" followed by a direct answer, with numbers, in the first paragraph.
AI models love that question-and-answer format because they can lift it and cite it as is. Give the answer first, the nuance after. A page that clearly answers one specific question stands a far better chance of being cited than a catch-all page skimming ten subjects.
You no longer optimize for a rank in a list, you optimize to be the source the AI judges reliable enough to name. Real authority wins, not tricks.
Building authority and reviews
AI models, like Google before them, try to separate credible sources from noise. They look at who talks about you, who cites you, what people say about you. A business with 200 detailed, recent Google reviews inspires more confidence in an AI than a ghost site with no trace anywhere else on the web.
In practice: build up genuine customer reviews, get mentioned by local media, by associations in your industry, by partners. Show clearly who is behind the business and what expertise those people actually have. Authority is not declared, it is proven through signals the machine can verify somewhere other than your own site.
Structured data and consistent information
Structured data markup is invisible code that tells the machine, in plain terms, "this is my address, this is my price, this is a review, these are my hours". It removes ambiguity. An AI that understands without guessing is an AI that cites you confidently. For an SMB, the basic markup — local business, reviews, FAQ — already makes a big difference.
Even more important: consistency of your information everywhere. Your name, address and phone — what is called the NAP — have to be identical on your site, on Google, in directories, on Facebook. An address written three different ways sows doubt. And a machine in doubt picks somebody else.
Local specificity is your best asset
For a Quebec SMB, local is an enormous advantage. An AI answering "best caterer for a wedding in the Eastern Townships" needs sources anchored in the territory. The more precisely localized your content and signals are — city, region, neighbourhoods served, in the language your customers actually use — the more relevant you are for the local queries big national chains serve badly.
Name your regions, talk about local realities, build up reviews from customers here. A large company based outside Quebec will struggle to look more credible than you on a question rooted in your own area. This is ground where an SMB can beat opponents far bigger than itself.
What does not change
For all the churn, the substance is unchanged. Quality content, written by someone who knows their trade, answering a real search intent — that is what won yesterday and what still wins. AI models are trained to recognize genuine expertise. The old principle holds: be the best answer to the question, for real.
The intent behind the search matters as much as ever. Someone asking "how" wants to learn, someone asking "best" or "price" wants to buy. Address the right intent with the right content and you stay relevant whether the answer arrives through a blue link or through an AI.
Mistakes to avoid
First mistake: panicking and tearing everything down. SEO is not dead, it is evolving. Your foundations — useful content, a fast site, reviews, consistency — still hold. Second mistake: falling back into old bad habits and stuffing your pages with keywords. AI models catch that even better than Google does. Stuffing makes you look less credible, not more.
The worst mistake: generating mountains of worthless AI content to "feed the machine". When everyone publishes the same generic text an AI spat out, nothing stands apart, and the engines learn to distrust those hollow pages. Ironically, in the age of AI what gets you cited is real human expertise, real numbers, the concrete examples only you can give.
GEO is not a new magic trick, it is discipline applied to shifting ground. If you want to know whether an AI cites you today, and what would need fixing so that it does, we look at it together during the free 15-minute diagnostic call.