The newsroom
Research and analysis for the AI search era.
We write the way a publisher runs a newsroom: named authors, dated posts, every quantitative claim linked to a primary source. We say what works, and we say plainly what does not.
More from the newsroom
Classic Google plus six AI answer engines: search in 2026
Classic Google, AI Overviews, ChatGPT, Claude, Perplexity, Gemini, and Grok each claim a slice of intent. They behave differently, and a single-engine SEO strategy is now a coverage gap.
Shimon Carroll ·
llms.txt is not a ranking signal: what actually gets you cited by AI
A thoughtful proposal that sounds like it should work, but as of mid-2026 no major engine honors llms.txt. We separate the inert from the load-bearing, with citations.
Shimon Carroll ·
Schema markup for AI citation: the types that matter
A practical guide to the Schema.org types that move the needle for AI citation, Organization, WebSite, Article, Product, LocalBusiness, FAQPage, BreadcrumbList, HowTo, and the entity work that ties them together.
Shimon Carroll ·
SSR, SSG, or CSR: which rendering modes AI crawlers can actually read
AI crawlers read the first HTML response and nothing after it. Server rendering, static generation, and ISR survive that; client-side rendering does not. A mode-by-mode guide, a five-minute test, and the semantic HTML upgrade that makes a readable page quotable.
Shimon Carroll ·
Why classic SEO audits miss AI search, and what to check instead
Classic SEO scanners grade the page Google ranked in 2018: titles, headings, links, schema validity. AI search adds five things they were not built to measure. Here is the gap, the evidence behind it, and a checklist that closes it.
Shimon Carroll ·
AI crawler directory 2026: every bot, what it does, what blocking costs
A verified reference to the AI crawlers that matter in 2026, from GPTBot and OAI-SearchBot to Claude-SearchBot, PerplexityBot, Google-Extended, and Applebot-Extended: what each one does, whether it obeys robots.txt, and two robots.txt templates.
Shimon Carroll ·
Local SEO when AI Overviews sit above the map pack
AI Overviews appear on most informational and hybrid local searches but few pure "near me" ones. What that means for local businesses, what the AI layer cites, and the six changes that win both the overview and the pack.
Shimon Carroll ·
Do named authors get cited more by AI? What the evidence shows
There is no published, controlled study showing that a named author makes AI engines cite a page more. There is clear guidance from Google on bylines and trust, and a cheap author setup that serves both. An honest look, including a number we retracted.
Shimon Carroll ·
Anatomy of a citable passage: how to write paragraphs AI engines quote
AI engines quote passages, so the unit of writing that matters is the paragraph. What peer-reviewed research and Google's own documentation say makes a passage citable, a before-and-after rewrite, and a checklist for your top pages.
Shimon Carroll ·
How Google AI Overviews pick citations, and how to become one
What Google documents about AI Overview citations, what independent data adds, why most cited pages no longer rank in the top 10, and a practical sequence for earning citations without chasing myths like llms.txt.
Shimon Carroll ·
AEO vs GEO vs AIO: what the acronyms mean and which one you need
AEO, GEO, and AIO describe overlapping slices of the same shift: being cited inside AI-generated answers. Where each term came from, how they differ in practice, and why the useful distinction is not the acronym but the engine.
Shimon Carroll ·
Unlinked brand mentions are the new backlinks for AI search
In Ahrefs' study of 75,000 brands, web mentions correlated with AI visibility roughly three times as strongly as backlinks. What that does and does not prove, the patent history behind "implied links", and a practical program for earning mentions.
Shimon Carroll ·
How AI engines pick sources: a working model of the pipeline
ChatGPT, Claude, Perplexity, and Google's AI Overviews all decide whether to search, rewrite the question into searches, retrieve from an index, and compose a cited answer. What each stage rewards, with the vendors' own documentation as evidence.
Shimon Carroll ·
FAQ schema in 2026: what changed, what still works, five mistakes
Google retired FAQ rich results on May 7, 2026, after restricting them in 2023. What that means for FAQPage markup, what no AI engine has confirmed, where a good FAQ section still earns its place, and the five mistakes that make FAQ content worthless.
Shimon Carroll ·
GEO vs SEO: what actually changes in the work, and what does not
SEO earns a ranked link; GEO earns a mention or citation inside an AI answer. The foundations are shared, but the unit of work, the off-site priorities, and the metrics differ. A side-by-side comparison with sources, and how to run both without doubling the effort.
Shimon Carroll ·
The E-E-A-T audit checklist for 2026, and what not to pay for
Google says E-E-A-T is not a specific ranking factor, and that trust matters most. A site-wide checklist built from Google's own guidance and the Search Quality Rater Guidelines, covering who, how, why, reputation, and the spam policies that sink sites.
Shimon Carroll ·
The Perplexity citation playbook: how to get cited in 2026
Perplexity cites sources on every answer from an index built by its own crawler. How that works according to Perplexity's documentation, why it sends comparatively more traffic, and a six-step playbook for earning citations, with the robots.txt details that trip people up.
Shimon Carroll ·
Core Web Vitals in 2026: INP is the one that bites
What Core Web Vitals do and do not do for rankings in Google's own words, how the web scores on each metric according to the 2025 Web Almanac, why INP is the hard one, and a practical order of fixes.
Shimon Carroll ·
Topic clusters in 2026: still working, but built for fan-out now
Topic clusters still work, and AI search makes them more valuable: AI Overviews and assistants fan a question out into sub-questions and cite the best answer to each. How to plan a cluster for that, link it properly, measure it, and avoid the scaled-content trap.
Shimon Carroll ·