On 11 November 2026, Search Engine Journal published a column by Bill Hunt, President at Bisan Digital. He describes one client project. His team ran 38 criteria-based prompts across several product categories and buyer segments, on five AI surfaces. That produced 190 responses.
The company was mentioned in 98.4% of them. It appeared as a primary recommendation or shortlisted option in 93.2%. But 73.7% of the responses had weak or missing evidence behind important claims about the company. Its lead pipeline was shrinking anyway. Hunt says the analysis could not prove the gap caused every lost sale.
What the article is claiming
Hunt’s argument is that visibility tells you whether you are in the consideration set. It doesn’t tell you whether you stay there. As buyers add specific criteria, such as integrations, workflows and outcomes, the AI needs evidence to keep a vendor qualified. He proposes “decision readiness” as a measure to track next to visibility.
He also makes a useful distinction. A company’s own documentation is usually enough to show that a feature or integration exists. Claims about superiority, effectiveness, scale or measurable outcomes need independent corroboration.
What is new and what isn’t
The metric is the smaller part. The new thing is where the buyer’s checking now happens. A shopper or a procurement person asks an AI for a shortlist, then asks follow-ups. Does it work with our system? How does the process run? What results has anyone got? The shortlist is the first answer and the follow-ups decide whether you survive it.
What hasn’t changed is the old rule that vague claims don’t persuade anyone. Hunt’s examples are ordinary. A bullet claiming an “end-to-end process” with no description of how data moves between systems. An efficiency claim with no test or customer result behind it. The AI is doing what a careful buyer does.
What I’d tell a small business
Say you run a clinic, a repair shop or a small agency. Your website says “fast, reliable, trusted service.” An AI model can read that sentence and has nothing to quote from it. Then a buyer asks a follow-up: how fast, and how do you know?
Numbers survive and adjectives don’t. These are the kind of statements an engine can repeat:
- “We reply to WhatsApp enquiries within 15 minutes between 9am and 6pm.” That has a number you can check.
- “We connect to Xero and Shopify.” That names the systems.
- “In October 2026 we completed 212 bookings with a 4% no-show rate.” That has a date and a count.
Date every one of them. AI answers change week to week, and a stale number is worse than no number. Put in only what you can show. Hunt warns against “evidence-shaped content” written only to please a machine. If a knowledgeable customer wouldn’t find it useful, it shouldn’t be on the page.
Your pages get you cited. Other people’s pages keep you there.
For claims about results, a statement on your own site carries less weight than the same statement somewhere else. That is Hunt’s point about independent corroboration, and it matches what I’ve watched in AI search.
In August 2026, AeroChat ranked third on Google for “best Shopify chatbot” but wasn’t named in the AI Overview. I first put it down to the AI preferring a neutral article over a seller’s own page. A month later I went through the full list of sources behind that answer. One article named AeroChat first, but none of the other sources mentioned us at all, so the AI didn’t either.
So the work has two parts. Publish specific facts on your own site. Then get them repeated: in a customer’s review that names the result, a partner’s integration page, a local directory, a trade article, an interview. The wording matters less than the facts being the same everywhere. If your site says 15 minutes and a directory says “quick”, the engine has one number and one adjective.
What to do
- Write down the five questions a buyer asks after “who is good at this”. Integrations, how a typical job runs, price range, timing and results are the usual ones.
- Answer each with a plain sentence containing a number, a name or a date. Put it on one page where a person or an engine can find it without hunting.
- Ask the people who know the facts to check them. Hunt notes that content teams cannot invent missing facts, and he’s right. Sales, operations and the owner hold the real numbers.
- List where each fact appears outside your site. Pick two or three places to fix this month, such as partner pages, directories and reviews.
- Ask your AI of choice the follow-up questions a buyer would ask, and see which answers come back thin.
Source: Search Engine Journal: Visibility Gets You On The List, Evidence Helps You Stay There