What a Uniform Supplier and a Bank Have In Common, According To AI

What survives the comparison

• AI almost never cites a negative source, in either sector. 0.4% of 58,299 workwear citations (May to October 2026) carried negative sentiment; in the banking questions tracked, 0.2% of 34,532 (December 2025 to March 2026).
• Where AI gets its information differs by sector. Workwear answers lean heavily on community sources: since late August, 12.6% of workwear citations came from Reddit. In banking, a bank’s own website stayed the main source: about 64% of citations, between 60% and 68% every month from December to March. LinkedIn was about 1% and Reddit effectively zero.
• Both sectors give short lists. In banking, the two most-named banks appeared in about three answers in four on a fixed set of 16 questions. In workwear, the most-named supplier appeared in 70% of answers to its main “which supplier” question.

AI says “I don’t know” more often than it gets facts wrong, in both sectors

Both studies check what AI engines say about a business against the business’s own answers. Answers of “Unknown” count as no answer.

Two brands fact-checked in this sector, 18 May to 22 October 2026. Answers of “Unknown” count as no answer, and a policy link that differs only by “www.” or a trailing slash counts as right.

In workwear, AI often declines to give a specific fact (35% of checks) and is wrong about 5% of the time. Gemini declines most.

Banking: all 1,272 banking fact-checks came from one bank’s account, so they describe that bank, not the sector. AI gave no wrong email addresses; the only errors were an outdated phone number and a dead link, both in February and March 2026, and neither appeared from May.

Community sources matter in one sector, not the other

A uniform buyer has buying guides, comparison blogs and forum threads to read. A bank customer mostly has the bank’s own pages for rates and terms. That difference shows up in what AI cites. The two studies covered different windows and our citation capture changed in late August 2026, so read the levels as a contrast, not a precise comparison.

Two studies, read side by side

• Workwear figures come from neutral buyer questions tracked in two SiteSignal accounts in that sector. One account is The Uniform Consultants, a SiteSignal client.
• Banking figures come from questions tracked in one bank’s SiteSignal account (The State of AI Visibility in Sri Lankan Banking). Bank results use December 2025 to March 2026, when the questions didn’t change.
• The two studies used different questions, accounts and dates. Only patterns that hold regardless of those differences are stated here.

The takeaway: AI visibility depends on the category. Track your own buyers’ questions before borrowing numbers from someone else’s.