SiteSignal
AI engines get a company's contact details wrong far more often than any other fact, and the pattern holds across unrelated industries. SiteSignal compared two independent studies, UK uniform and workwear suppliers and Sri Lankan banks, and found the same weak spot in both.
Support email addresses were wrong 30.0% of the time for workwear suppliers. Public contact emails were wrong 56.9% of the time for banks. Identity basics such as company name, founding date and headquarters were the most reliable facts in both sectors.
What held in both industries
- Contact details fail most: the worst-performing field in both studies, found independently. Neither study was designed to look for the other's result.
- AI almost never cites criticism: negative-sentiment citations stayed under 1% in both (0.5% for workwear, 0.2% for banking).
- Gemini hedges hardest: it declined to state facts most often in both sectors (71.3% for workwear, 29.2% for banking on a smaller sample).
What differed
- Where AI gets its information: workwear suppliers' own websites lost 38 percentage points of citation share over eight months. Banks' share moved just 4 points.
- Social sources: workwear saw a surge in Reddit citations; banking saw almost no social-platform citations at all.
- Market shape: workwear has a long tail of 890 suppliers with a very small stable core. Banking has a settled group of six leaders.
What it means
"We didn't expect two industries this different to share the same failure," said Niraj Appuhamy, founder of SiteSignal. "It suggests a simple rule for any business: check the contact details AI gives out about you first. They're the facts people act on, and the facts AI gets wrong most."
About the data
This comparison combines two SiteSignal studies that used the same measurement method: UK Uniform & Workwear: The State of AI Visibility and The State of AI Visibility in Sri Lankan Banking. Full analysis: Workwear vs. Banking: What AI Visibility Has in Common.
Press contact: hi@sitesignal.app