Audit whether your key company facts agree across sources

A tribunal held Air Canada responsible after its chatbot contradicted its own webpage. A five-step audit to find and fix disagreements in your company's key facts.

If your website says one thing, your chatbot says another and an old directory listing says a third, customers and AI systems have to guess which is right. Sometimes the guess costs money.

A ruling against Air Canada shows what can happen when a company’s own information disagrees with itself. This article explains the case and gives a simple audit method.

The Air Canada ruling

In February 2024, British Columbia’s Civil Resolution Tribunal decided Moffatt v. Air Canada. The dispute was small, but the reasoning travelled widely.

In November 2022, after a family death, Jake Moffatt used a chatbot on Air Canada’s website to ask about bereavement fares. The chatbot said customers could apply for a reduced rate within 90 days of the ticket being issued.

Air Canada’s own “Bereavement travel” webpage said the policy did not apply to requests made after travel was completed. The chatbot even linked to that page.

Moffatt booked flights, applied within the window and was refused. An Air Canada representative later admitted the chatbot had used “misleading words”.

What the tribunal said

Air Canada argued it could not be held liable for information from its chatbot. The tribunal member called this “a remarkable submission”.

The decision said the chatbot was still part of Air Canada’s website. “It should be obvious to Air Canada that it is responsible for all the information on its website,” the tribunal wrote.

It also asked why customers should have to double-check information found on one part of a website against another part. It found Air Canada had not taken reasonable care to make its chatbot accurate.

Air Canada was ordered to pay $812.02, including $650.88 in damages, interest and tribunal fees.

Why marketers should care

The sum was small. The reputational lesson was not, and it applies well beyond airlines.

Your company’s facts now appear in many places you do not fully control: chatbots, help centres, partner sites, review platforms, business directories and AI-generated answers.

Google’s guidance on AI features in Search asks site owners to make sure structured data matches visible text, and to check that Merchant Center and Business Profile information is up to date. Consistency is part of the documented advice.

The key facts audit

The aim is to find where your company’s important facts disagree, then fix the source. Plan a half-day for a first pass.

Step 1: list your key facts

Choose 15 to 25 facts that customers, journalists or AI systems are most likely to need. Typical candidates:

• Official company name, legal entity and headquarters
• Leadership names and titles
• Products or services, with current names
• Prices, fees or how pricing works
• Policies customers rely on: refunds, cancellations, warranties, service levels
• Locations, opening hours and contact details
• Founding date, ownership and major milestones

Prioritise facts where an error could cost a customer money or create a legal risk. Air Canada’s problem was a policy, not a logo.

Step 2: name the source of truth

For each fact, decide which page or document is authoritative. Record its owner: legal, finance, product or customer service.

If nobody can say which version is correct, that is your first finding.

Step 3: check every place the fact appears

Search for each fact across:

• Your own website, help centre, FAQs and old landing pages
• Chatbot and support macro answers
• Your Organization structured data and Business Profile
• Partner, reseller and distributor pages
• Directories, review sites and trade association listings
• Press releases and your press kit

Then ask several AI assistants the same questions a customer would ask. Record what they say and any sources they show. Treat these answers as a snapshot; they can change.

Step 4: record and score disagreements

Use a simple sheet with five columns: fact, source of truth, where it disagrees, risk level, and owner of the fix.

Mark risk as high when a wrong fact could lead a customer to act on bad information, as Moffatt did.

Step 5: fix the source, not the symptom

Correct your own pages first, including structured data. Google’s Organization structured data guidance describes how markup can help it understand your administrative details and disambiguate your organisation.

Then update profiles you control, and ask partners and directories to correct theirs. You cannot edit AI answers directly; you can only improve the sources they may draw on.

Chatbots need the same governance

If you run a customer-facing chatbot, treat its answers as published content. Review it whenever a policy changes, and test it with the questions customers actually ask.

Give it the same source of truth as your website. Where it cannot answer reliably, it should hand over to a person.

A worked example

Hypothetical: a Belgian facilities management firm runs the audit. It finds three different cancellation notice periods: 30 days in its contract, 60 days on an old FAQ, and “one month” in its chatbot.

A partner’s website still lists a managing director who left two years ago. Two AI assistants repeat the old name.

The firm fixes its FAQ and chatbot the same week, asks the partner to update its page and adds a quarterly check to its calendar. The details are illustrative.

Make it routine

Run the full audit once a year and a short version each quarter. Add a step to every policy change: who updates the website, the chatbot and the press kit?

That small habit is cheaper than a dispute, and it makes your company easier to describe accurately.