Skip to main content
SEO InsightsTechnical SEO

There is a second AI search network and almost nobody audits it

Summarize with ChatGPT
JK
John Kyprianou
July 28, 2026
7 min read
Updated: September 2, 2026
Diagram of the Microsoft search index feeding Copilot, DuckDuckGo and Yahoo Scout, the second AI answer network, by SEO Turtle

Ask most business owners how they are doing in AI search and they will tell you about AI Overviews and ChatGPT. Ask them when they last logged into Bing Webmaster Tools and you get a blank look.

That gap is the point of this post. There is a second distribution network for AI answers, it runs on Microsoft's index, and checking your position in it costs an afternoon.

Diagram of the Microsoft search index feeding Copilot, DuckDuckGo and Yahoo Scout, the second AI answer network, by SEO Turtle

Yahoo Scout is the reminder that this network exists

In January 2026 Yahoo launched Scout, an AI answer engine built on Anthropic's Claude and grounded in Yahoo's own data ecosystem, with Microsoft's Bing grounding API supplying the web search layer (Yahoo Inc, Search Engine Land).

What makes Scout worth noticing is not the model choice. It is the link behaviour. Scout puts sources front and centre, with hoverable citations through the answer and a featured source block, rather than burying attribution behind a chevron.

That is the opposite of the pattern everyone has been complaining about for two years. An answer engine that is trying to send traffic out is one you want to be visible in.

The map most people have never drawn

Here is the part that surprises clients. A single index sits underneath several places people are getting answers.

  • Microsoft Copilot, inside Windows, Edge and Microsoft 365.
  • DuckDuckGo, which sources its traditional web links and images from Bing and layers its own crawler, Instant Answers and AI features on top (DuckDuckGo help pages).
  • Yahoo, including Scout.
  • The Windows search box and Edge's default experience.

None of those are Google. All of them are downstream of how well Microsoft has crawled, rendered and understood your site.

If you have never verified your domain in Bing Webmaster Tools, you have no idea what any of them can see. We run into this constantly with new client sites: strong Google presence, and a Microsoft index footprint that is stale, partial, or missing entire sections.

Our take: this is not a "Bing strategy"

We want to be careful here, because the SEO industry has a habit of turning every observation into a new discipline with a new acronym.

This is not that. We are not telling you to build a Bing content plan, or to write differently for Copilot, or to buy a rank tracker for DuckDuckGo. Microsoft's share of classic search is what it has always been, and chasing it as a traffic channel has never made sense for most businesses.

What changed is that the index stopped being a search engine and started being plumbing. It now feeds answer surfaces that reach people who never type a query into bing.com. Being absent from plumbing is a different kind of problem from ranking eighth.

So the correct response is proportionate: a one-off hygiene check, not a strategy. Half a day, mostly free tools, then move on.

The audit, in the order we run it

1. Verify the domain in Bing Webmaster Tools. It is free, and if you already have Search Console you can import the property rather than re-verifying from scratch (Bing Webmaster blog). Most sites we look at have simply never done this step. Once you are in, open the AI Performance report, in public preview since February 2026, which lists the URLs cited in Copilot and in Bing's AI-generated summaries and the grounding queries that pulled them in (Bing Webmaster blog).

2. Compare indexed page counts against Google. Not for vanity. You are looking for a gap that points at a structural problem: a section that never got crawled, a sitemap that was never submitted, a template that renders empty.

3. Check robots.txt for accidental blocks. Bingbot is the obvious one, but plenty of sites have picked up copy-paste blocks over the years, and AI crawler blocking has made robots.txt files messier than they used to be. Take two minutes to test Bingbot against a few real URLs rather than reading the file by eye. If you are unsure what you are actually allowing, our guide to AI crawlers and robots.txt walks through it.

4. Test rendering, not just fetching. If your content only exists after JavaScript executes, treat that as a risk on every index that is not Google's. This is where we find the ugliest problems, and it is the same class of issue covered in our JavaScript SEO troubleshooting notes. For a quick per-page look, our Chrome extension that audits a single page covers the on-page basics in one click.

5. Wire up IndexNow. It is an open protocol supported by Bing, Yandex, Naver, Seznam and Yep that lets you push changed URLs instead of waiting to be crawled (IndexNow). Most modern stacks and CMS plugins support it, and it is a genuinely small piece of work.

6. Read the crawl diagnostics once. Not weekly. Once, properly, and fix whatever it flags.

That is the whole thing. If your site is healthy, you will find nothing and you will have bought certainty cheaply. If it is not, you have just found visibility that was sitting there unclaimed.

Where the simple story breaks down

One correction, because it gets repeated as gospel: "ChatGPT runs on Bing, so fix Bing and you fix ChatGPT" is out of date.

OpenAI operates its own crawler, OAI-SearchBot, to surface sites in ChatGPT search (OpenAI), and its retrieval now blends multiple sources rather than passing through a single index. The overlap between what ranks in Bing and what ChatGPT cites is looser than it was in 2024. We went through what ChatGPT actually reads, and how rarely it opens the live page, in ChatGPT almost never opens your page.

We would treat Microsoft's index as necessary but not sufficient. It is a hard prerequisite for Copilot, DuckDuckGo and Yahoo Scout. It is one contributing input for ChatGPT. Neither of those is a reason to skip the check, and neither is a reason to expect the check alone to earn you citations. For that side of the work, structure and source quality still do the heavy lifting, which we covered in how to structure content so LLMs cite you.

Why this matters more in Cyprus and small US markets

Two reasons, and they are practical rather than philosophical.

First, in smaller markets the competitive set is thin. On a query where three local businesses could plausibly be cited, being one of the two that an answer engine can actually read is decisive in a way it never is in a crowded national vertical.

Second, Microsoft's index skews toward desktop and workplace contexts. Copilot inside Microsoft 365 and Edge is a business-hours surface. If you sell B2B services, professional services, or anything bought from an office desk, that audience composition is worth more to you than its raw share suggests.

None of this displaces Google. Google is still where the volume is, and the broader AI search market picture has not been rewritten by a Yahoo launch. It just means the cost of ignoring the second network went up while the cost of fixing it stayed at roughly zero.

The bottom line

Answer engines are consolidating onto a small number of indexes. Google runs one. Microsoft runs the other. OpenAI is building a third, and Apple quietly tripled its crawl capacity for a fourth in July.

Everyone is spending their AI search energy on the first and the third. The second is unglamorous, free to check, and quietly feeding answers to people using Windows, Edge, DuckDuckGo and now Yahoo Scout.

Verify the domain. Compare the index counts. Check what robots.txt is really doing. Turn on IndexNow. Then go back to the work that actually needs your attention.

If you would rather someone else ran the check, that is part of what we do in a technical SEO engagement, and it is one of the things we look at in a free SEO review.

John Kyprianou

John Kyprianou

Founder & SEO Strategist

John brings over a decade of experience in SEO and digital marketing. With expertise in technical SEO, content strategy, and data analytics, he helps businesses achieve sustainable growth through search.

Related Articles

Diagram of a Google Search Console page indexing report with URLs sorted into technical faults and quality judgements, by SEO Turtle
Technical SEO

Crawled, currently not indexed: how to work out what is actually wrong

Google has looked at the page and decided not to keep it. Most advice treats that as a technical fault and sends you off to hammer Request Indexing. Here is the order that actually finds the cause: an afternoon of cheap plumbing checks, a look at what Google chose as the canonical, and then the harder question of whether the page deserves to exist.

September 8, 2026
Diagram showing Applebot's published IP pool growing from 2,400 to 7,056 addresses ahead of the Siri AI launch, by SEO Turtle
Technical SEO

Apple quietly tripled Applebot's crawl capacity weeks before Siri AI ships

Apple's published Applebot address pool went from 2,400 IPs to 7,056 with no blog post and no explanation, weeks before the rebuilt Siri ships in iOS 27. Most sites have never looked at how they treat Applebot, and a lot of them are blocking the wrong user agent. Here is what changed and what to check this week.

August 27, 2026
Diagram of ChatGPT's three retrieval layers, its own index, a shared read cache and rare live page opens, by SEO Turtle
Technical SEO

ChatGPT almost never opens your page. Here is what it reads instead

New research pulled apart how ChatGPT actually fetches web pages, and the answer is uncomfortable. It runs its own index that barely overlaps with Bing, serves most answers from a cached copy of your page, and only truly opens about one page in eighty. Here is what that means for how you write and structure a page.

August 18, 2026
Illustration of a stack of duplicate web pages marked with a red cross resolving into one clean authoritative page that feeds an AI answer, by SEO Turtle
Technical SEO

Most AI visibility wins are just technical debt you finally paid off

Businesses are buying GEO tools to fix problems a 2019 site migration created. AI search did not add new technical requirements, it just stopped compensating for the old ones. Here is our practitioner take on why the technical SEO backlog is now the AI visibility roadmap, and why that favours smaller sites.

August 17, 2026
Google retires FAQ rich results in 2026, what it means for structured data and schema, guide by SEO Turtle
Technical SEO

Google killed FAQ rich results. What that actually tells you about schema

On May 7, 2026 Google stopped showing FAQ rich results, then told everyone in its new AI search guide that special schema is not needed for AI features either. Both moves point the same way. Here is our practitioner read on what structured data is actually for now, and what to stop wasting time on.

June 18, 2026

Continue Your SEO Journey

Explore more expert insights and take action on your SEO strategy