Why AI keeps finding your competitors

Before rewriting your homepage, it helps to know where the gap is. Did AI find their company first? Did it find a more relevant page? Or did it already know their name?

Three recent developments make those questions more useful to ask.

1. Bing can show where other sites appear and yours doesn’t

On October 13, Microsoft introduced Topics Portfolio, a beta feature in Bing Webmaster Tools. It groups your site’s AI citation activity by topic and surfaces queries where other sites appear while yours is absent.

There are limits. Sites need enough data to see the feature, the figures are approximate, and some suggested opportunities involve searches for a competitor’s name. This is a view of Bing’s activity, with gaps of its own.

Why it matters: A total citation count can look encouraging while hiding the questions that matter commercially.

Imagine a consultancy cited for explaining Microsoft Copilot, but absent when someone searches for help implementing it. The first shows that its information is useful. The second is closer to a potential customer looking for a provider.

Our read: look at which buying situations your visibility covers before chasing a bigger number.

Source: Microsoft Bing

2. Google is making the source easier to reach

On October 15, Google announced that inline links would appear higher in AI responses more frequently. It is also testing a prominent section for tasks where one or two official websites provide the exact information or destination someone needs.

That official-page section remains a test. Google’s announcement does not establish that businesses will receive more visits.

Why it matters: The page linked inside an answer deserves attention.

Consider a buyer checking whether your software connects to their accounting system. A direct link to a clear integration page gives them somewhere useful to go. A broad homepage may leave them searching again.

Our read: check where the link leads and whether that page helps the buyer take the next step.

Source: Google

3. Familiar brands can get a head start

An October research preprint examined more than 158,000 AI answers across consumer and business software markets.

The researchers found that ChatGPT sometimes included familiar brands in the searches it generated. In controlled experiments, changing brand names could also change product rankings while the underlying facts stayed identical. For lesser-known business software vendors, visibility depended more heavily on evidence found through search. These findings concern the markets and systems tested.

Why it matters: A smaller company cannot assume AI will already know what makes it relevant.

Our read: make the evidence easy to find. Name the customers you serve, the problems you solve and the requirements you meet. Give those claims enough detail for someone to check them.

Source: University of Toronto researchers, via arXiv

What we’re watching

Whether Bing’s opportunity data becomes useful for smaller sites, whether Google’s clearer links bring relevant visitors, and how consistently unfamiliar companies surface when buyers describe a specific need.

The bigger picture

“Are we showing up in AI?” leaves too much unanswered.

A more useful question is: When someone needs exactly what we do, what helps AI find us, choose us and send them somewhere useful?

That gives a company something concrete to investigate.

The Million Token Window tracks what’s changing in how AI finds, compares and recommends companies.