WebMax Canada says AI usually gets Canadian business facts right

WebMax Canada’s latest study tested whether ChatGPT, Gemini, Claude and Augure could accurately describe 16 Canadian businesses after finding them in search. The company found high factual accuracy once the right business was identified, but more errors and gaps appeared when public information was outdated or contradictory.

Why it matters: - Canadian businesses can show up in AI results with the wrong address, phone number or company history if public information is stale or conflicting. - The study suggests businesses need to manage two separate risks: being hard for AI to find and being described inaccurately once found. - For consumers, the findings show AI can be useful for research, but not a substitute for checking official sources when details matter.

What happened: - WebMax Canada released the Canadian Business AI Information Accuracy Study 2026 on August 13, 2026. - The study tested ChatGPT, Gemini, Claude and Canadian AI platform Augure on 16 real Canadian businesses. - Researchers asked whether AI could correctly identify and describe business facts after locating the right company. - The testing covered business identity, official website, location, phone number, services, service area, company history, credentials and conflicting public information. - A reference answer key was prepared before testing and kept separate from the AI outputs being scored.

The details: - Claude posted 100% factual accuracy on the scorable fields it answered, with 96.3% coverage. - ChatGPT posted 99.2% factual accuracy, with 98.5% coverage. - Gemini posted 96.9% factual accuracy, with 97.0% coverage. - All three platforms correctly identified the intended business in all 16 cases. - Augure posted 93.8% factual accuracy on the fields it answered, but coverage was 17.8%. - WebMax Canada said several Augure sessions had search or result-delivery problems, so the company is treating Augure’s result mainly as a coverage issue. - Only two fields in the full locked scoring dataset were classified as outright incorrect. - Seven of the 16 businesses had known conflicts in their public information before testing began. - Those conflicts included different addresses, inconsistent company history and details that did not match across sources. - Under the study’s scoring method, ChatGPT identified or substantially preserved 92.9% of conflict cases. - Claude scored 64.3% on conflict cases, Gemini 28.6% and Augure 14.3%.

Between the lines: - The results show that factual accuracy can be strong once AI locks onto the correct business, but the harder test is deciding what to do when public sources disagree. - That distinction matters because many small and mid-sized businesses have old directory listings, prior addresses or outdated descriptions still circulating online. - Susan Jones of WebMax Canada said the challenge is not just finding one believable answer, but recognizing when credible sources conflict. - WebMax Canada is positioning the research as Canadian evidence for Canadian businesses, since much AI-search research focuses on U.S. companies or global datasets. - The company argues that Canadian owners need to know both whether AI can find them and whether the information AI surfaces is current.

What’s next: - WebMax Canada said the study is a starting point and that AI platforms and public business information can change over time. - The company plans to use the research to encourage Canadian businesses to review what the public internet says about them and correct important listings where possible. - WebMax Canada also said its earlier August 11 study found that 66.8% of market-specific businesses identified appeared on only one of the three major AI platforms, while 11.2% appeared across all three. - Together, the two studies point to a two-step task for businesses: be discoverable in AI search, then keep the underlying information clean and consistent.

The bottom line: - AI appears fairly reliable at pulling basic facts for Canadian businesses, but outdated or conflicting public data can still distort the picture.