How HR and employment lawyers show up in ChatGPT (and why most don't)

The question is being asked somewhere else now

Picture a restaurant owner in Saint John at eleven at night. A manager has to go, it has been building for months, and the owner wants to know what they are allowed to do. Five years ago they would have typed “terminating an employee New Brunswick” into Google and clicked whichever law firm ranked first.

Now they open an AI assistant and type the actual question: “Do I need a lawyer to terminate someone in New Brunswick, and how much notice do I owe?” The assistant answers in a few paragraphs. It mentions the Employment Standards Act, it mentions common law notice, it suggests talking to a lawyer, and, increasingly, it names a source or two. Sometimes a firm.

Whether that firm is yours depends almost entirely on what your website looks like to a machine. And for most employment and HR law practices in Atlantic Canada, the honest answer is that the website looks like nothing in particular.

What AI assistants actually reward

The mechanics differ between products and change often, so treat what follows as a description of direction rather than a rulebook. Across ChatGPT, Perplexity, Google’s AI results and Microsoft Copilot, the sources that get cited tend to share four properties.

• Clear authorship. A named person with a stated role and credentials, consistently presented across the site and across the web. Anonymous “by Admin” posts are weak signals.
• Structured content. Pages the machine can parse: a question in a heading, a direct answer underneath, supporting detail after that, and schema markup that labels what each part is.
• Specific jurisdictional answers. “In New Brunswick, the minimum notice under the Employment Standards Act depends on length of service” is citable. “Termination rules vary by province” is not, because it answers nothing.
• Consistent entity information. The firm’s name, address, phone, practice areas and people match everywhere: website, Google Business Profile, law society directory, LinkedIn. Inconsistency makes the machine less confident it knows who you are.

Notice what is not on the list: domain age, word count, backlink volume for its own sake, or how many times the phrase “employment lawyer Fredericton” appears. Those were the old game. The new game is being the clearest, most attributable answer to a specific question in a specific place.

What your content has to look like

Take the restaurant owner’s question and work backwards. For the assistant to cite you, there needs to be a page on your site that answers that question directly, for that province, in the voice of a named lawyer.

In practice that page looks something like this. A heading that is the question itself, or close to it. A first paragraph that gives the shape of the answer in plain language: whether a lawyer is required (it usually is not, legally) and when one is advisable (far more often than owners think). Then the sections an owner needs: statutory notice versus common law notice, with-cause versus without-cause, the role of a written contract, what a release is, common mistakes. Each section short, each one specific to New Brunswick, or Nova Scotia, or wherever you practise.

At the bottom, a clear statement that this is general information, not legal advice, and that the owner should speak to a lawyer about their situation. That is not just a liability line. It is also exactly the framing an assistant uses, which makes your page a natural fit to cite.

And it carries a byline. A real one, with a photo, a link to a full profile, and a one-line description of what that lawyer does. Not the firm. The person.

The jurisdiction problem

Many firms in the region practise across two or three provinces and write one page that tries to cover all of them. That page is useless to a machine answering a New Brunswick question, because it cannot tell which paragraph applies. Write one page per province per question. It is more work, and it is the work that gets cited.

The role of FAQ and Person schema

Schema markup is structured data embedded in a page that labels its contents for machines. Two types matter most for a law firm trying to be cited.

FAQPage schema wraps question-and-answer pairs so a crawler can pull the exact question and the exact answer without guessing which paragraph is which. If your termination page has five questions an employer asks, mark them up. The content must also be visible on the page as normal text; schema that describes content that is not actually there is a fast way to lose trust with search engines.

Person schema describes the author: name, job title, employer, credentials, a link to their profile, and links to the same person elsewhere (law society listing, LinkedIn). Combined with Organization or LegalService schema for the firm, it gives the machine a consistent entity to attribute the answer to. This is what turns “a page on the internet said” into “according to a lawyer at this firm in Halifax.”

None of this is exotic. A competent developer can add it to a site in days. The reason most firms have not is that nobody told them it was the difference between being quoted and being ignored.

What not to do

There is a style of law firm content that was built for the old search engine and is actively harmful in the new environment.

• Generic listicles. “Seven things every employer should know about terminations.” Not jurisdiction-specific, not authored, not structured, and indistinguishable from the thousand other pages with the same title. An assistant has no reason to pick yours.
• Thin location pages. The same five paragraphs republished for Fredericton, Moncton, Saint John and Bathurst with the city name swapped. Machines detect this and discount it.
• Content without a person. Articles attributed to the firm or to “the team” cannot be tied to credentials. Attribution is the whole point.
• Hedging everything. An answer that refuses to say anything concrete is not an answer. You can be specific about the law and still be clear that the reader’s situation needs individual advice.
• AI-generated bulk pages. Hundreds of near-identical question pages produced in an afternoon. They tend to be inaccurate on provincial detail, which in employment law is where the client’s risk lives, and they read as spam to the systems you are trying to impress.

The assistant is not looking for the firm that wrote the most. It is looking for the person who answered the question best, in the right province, in a way it can verify.

How to know if it is working

Measurement here is less precise than paid advertising, but it is not guesswork.

1. Ask the main assistants your target questions directly, once a month, and record whether and how your firm appears. Keep a simple log. This is crude, but it is the ground truth.
2. Watch referral traffic in your analytics from AI-related sources. Several of the assistants now pass identifiable referrers when a user clicks through.
3. Add “an AI assistant mentioned you” as an intake option when you ask how a new client found the firm. It will start appearing, and you want to know when.
4. Track which province-specific question pages are getting impressions in Google Search Console, since the same content that gets cited by assistants tends to earn long-tail search visibility as well.

Set the expectation internally that this is a quarters-long effort rather than a weeks-long one. The firms that are being cited today in Atlantic Canada are mostly the ones that started publishing specific, authored answers before anyone was measuring it.

What to do first

Pick one question. The one your employment group hears most often from new employer clients, phrased the way the client phrases it. Confirm with your law society’s marketing rules that the way you plan to publish is permitted.

Then write the answer for one province, under one lawyer’s name, structured as question, direct answer, detail, and a clear note to seek advice. Have your developer add FAQ and Person schema. Publish it. Ask the assistants the question a month later and see what changed.

That single page will teach you more about how your firm shows up in AI than any amount of reading about it, and it gives you a template for the next twenty.