Getting Found vs. Getting Cited: We Fact-Checked the Viral Chart

Why this chart deserves a careful read

Most search infographics are either too vague to be wrong or too confident to be right. This one is better than most. It takes the question every marketing leader is asking, whether AI search needs a different playbook, and answers it with fourteen concrete labels. That makes it useful. It also makes it testable.

So we tested it. For each box we asked two questions: does the claim match what the search and AI companies say in their own documentation, and does independent research support it? Where the answer was mixed, we say so. We are not grading the designer, who made a clear and well-structured graphic. We are grading the claims, because brands are already using charts like this to set budgets.

We have done this before with a viral SEO ladder in Rank, answer, cite. The pattern is similar. The instinct is right, and the details need work.

The scorecard

Ten boxes hold or mostly hold. Four need a fix. Here is the reasoning behind each verdict.

Checking the seven ways you get found

The engine row needs a footnote before the boxes do. Google, Bing and DuckDuckGo look like three engines, but DuckDuckGo says its traditional links are "largely" sourced from Bing, alongside its own crawler. For most brands, that row is two indexes, not three.

Crawlability holds, with a missing line

Robots.txt, sitemaps and rendering are the right three. One clarification: Google says robots.txt "is not a mechanism for keeping a web page out of Google." It manages crawling, not indexing. The missing line is AI search crawlers. OpenAI's crawler documentation says sites that block OAI-SearchBot "will not be shown in ChatGPT search answers," and Anthropic's documentation, reported by Search Engine Journal, warns that blocking Claude-SearchBot may reduce your visibility in Claude's search results. Crawlability belongs in both halves of the chart.

Keywords mostly hold

Intent and volume are sound. Difficulty is a score invented by SEO tool vendors to estimate competition, and every tool calculates it differently, so treat it as a planning aid, not a fact about Google. Keywords also behave differently in AI answers. The Princeton and IIT Delhi GEO study, presented at KDD 2024 and tested on 10,000 queries, found keyword stuffing scored below the unoptimized baseline in generative engines.

Core Web Vitals needs a fix

The box lists load time, mobile and stability. Google's Core Web Vitals are three specific metrics, and mobile is not one of them:

Responsiveness is missing from the chart entirely, and it is the metric that measures how quickly a page reacts when someone taps or clicks. Mobile friendliness still matters to page experience, and Google's AI features documentation lists page experience among its best practices for AI Overviews and AI Mode too. Just label it correctly.

Internal links hold

Clusters, anchors and hierarchy are well supported, and Google lists strong internal linking among its recommendations for AI features as well as classic search. We covered why in internal linking for SEO and AI visibility.

Content depth holds

Coverage, uniqueness and freshness all appear in Google's own explanation of ranking. One nuance: freshness is query-dependent. Google says freshness "plays a bigger role" for current news than for dictionary definitions. Updating an evergreen page just to change its date does nothing.

Backlinks mostly hold

Referring domains and relevance are right. Google says one quality factor is whether other prominent websites "link or refer to the content." "Authority" needs care. Google's John Mueller said in 2020 that Google does not use domain authority "at all" in its algorithms, while confirming PageRank is still one of many signals. Domain authority is a vendor metric. Do not buy links to move it.

Rankings hold

Positions, snippets and traffic are outcomes, not inputs. Listing them as a box is fine as long as nobody mistakes the scoreboard for the game.

Checking the seven ways you get cited

The engine row needs a footnote here too, and it is the most important one in the article. These AI tools are not separate from search. OpenAI says ChatGPT search draws on third-party search providers as well as content supplied directly by its partners. Google's Gemini API offers grounding with Google Search, which "connects the Gemini model to real-time web content" and returns citations. TechCrunch reported that Claude's web search appears to run on Brave Search, based on Anthropic's subprocessor list and matching citations, though Anthropic had not confirmed it. Each AI tool retrieves from a search index before it writes. Getting found is the entry ticket to getting cited.

Answer-first holds

Direct answers, clear structure and short paragraphs help readers and machines alike. The research behind GEO supports it. In the GEO study, adding quotations from credible sources was the best single method, improving visibility by about 41% over the baseline, with adding statistics, citing sources and improving fluency close behind.

Extractability needs a fix

Self-contained sections and a clear hierarchy are good writing. "Clean chunks" is where the box drifts. On Google's Search Off the Record podcast in January 2026, Danny Sullivan said of breaking content into bite-sized pieces for large language models: "We don't want you to do that." Write sections that stand on their own because a reader might land mid-page, not because you are trying to feed a model fragments.

Schema markup needs a fix

This box is the most oversold, and two of its three examples are dated. Google says no special schema is required to appear in AI Overviews or AI Mode. FAQ rich results stopped appearing in Google Search on May 7, 2026, and Google has since removed the documentation for the feature.

The citation evidence is weak as well. Search Engine Roundtable reported an Ahrefs study of 1,885 pages that added JSON-LD schema, compared with 4,000 control pages. It found "no major uplift in citations on any platform," and a small, statistically significant decline in AI Overview citations. Ahrefs sells SEO tools, and its own write-up notes the differences were small enough to be noise.

Original data holds, and it is the strongest box

Research, first-hand experience and proprietary numbers have the best support on the whole chart. Google's helpful content guidance asks whether content provides "original information, reporting, research, or analysis" and demonstrates first-hand expertise. The GEO study found statistics addition among the top performing methods. If a brand can only fund one box, fund this one.

Entity signals hold

Named authors and consistent naming are well supported. Google's guidance asks whether pages "carry a byline, where one might be expected" and whether bylines lead to more information about the author. Wikidata deserves a caution: it is a community-edited knowledge base, and an entry for a brand with little independent coverage is unlikely to help. Earn the coverage first.

Off-site authority holds, with a Reddit caveat

Media coverage and mentions have strong correlational support. An Ahrefs study of 75,000 brands found branded web mentions correlated with AI Overview brand visibility at 0.664, compared with 0.218 for backlinks. Ahrefs sells SEO tools, and correlation is not causation, but the gap is large.

Reddit is the shakiest of the three examples. Search Engine Land reported Promptwatch data showing Reddit's share of ChatGPT search citations falling from 3.83% to 0.52% in mid-August 2026, a finding the vendor called provisional. We think Reddit's role in AI answers resembles DMOZ's old role in search, a shortcut that fades as engines get better, and we made that case in Reddit, GEO and the DMOZ lesson. Earn real mentions where your customers talk. Do not try to manufacture them.

Commercial pages hold

Pricing, services and comparison pages answer the questions buyers bring to AI tools. When ChatGPT cut Reddit citations in August, an OtterlyAI analysis of 16 brands found official brand pages gained share in 10 of 15 reports. OtterlyAI sells AI visibility tracking, so read that as one vendor's sample. The direction still favors brands with clear, factual commercial pages.

What the chart leaves out

Three things are missing, and they matter more than any single box.

The first is the overlap. Five of the seven "found" boxes, crawlability, internal links, content depth, backlinks and page experience, are also preconditions for being cited, because AI tools retrieve from search indexes. Splitting them into two charts invites two budgets, two teams and two sets of metrics for what is mostly one job.

The second is measurement. The mug in the chart says "cited and paid, please." Citations and payment are different things. Pew Research found that when a Google result page showed an AI summary, users clicked a traditional result on 8% of visits, against 15% without one, and clicked a link inside the summary on just 1%. Track citations, referral visits and conversions as separate numbers, or you will celebrate visibility that never reaches revenue.

The third is the difference between platforms. The chart treats ChatGPT, Claude, Gemini and Perplexity as one audience. They use different search sources, different crawlers and, as the Reddit data shows, very different source preferences. A brand that only measures one of them is guessing about the rest.

One root system, not two trees

Here is how we would redraw the chart. Put a single foundation at the bottom: crawlable pages, open to both search crawlers and AI search crawlers, fast on the three real Core Web Vitals, and linked into clear topic hierarchies. Above it sits content written answer-first, carrying original data, credible quotations and a named expert. Around it sits reputation: coverage, reviews and mentions you earned. Search rankings and AI citations both grow out of that same base.

That is how we run SEO and GEO at Silverback, as one program with two reporting surfaces rather than two competing budgets. It is also why we do not chase single-box tactics, such as schema sprints, chunked rewrites or manufactured Reddit threads, that promise citations without building the thing citations are supposed to reflect.

How to act on this

The chart gets the destination right: brands need to be found and cited. It gets the map slightly wrong by drawing two roads. Build one foundation that search engines can rank and AI systems can trust, and both halves of the chart take care of themselves.