Cloudflare released a new tool on August 6, 2026, that measures whether AI assistants cite your website as a source. The AEO (Answer Engine Optimization) Visibility Dashboard sits inside Cloudflare's main platform and shows five metrics tracking how often your site appears in AI-generated answers. However, the tool has a significant limitation: it only monitors Anthropic's Claude and OpenAI's GPT, leaving out Gemini, Perplexity, Copilot, and Grok entirely.

What Exactly Does Cloudflare's New Dashboard Measure?

The dashboard produces five separate scores, each measuring a different aspect of how AI assistants treat your website. Understanding what each metric actually represents is crucial, because the numbers are calculated against different baselines.

• Citation Rate: The percentage of answers in your category that cite your site as a source, measured against sampled answers in the category Cloudflare inferred for you, not all AI answers globally.
• Prominence: When your site is cited, how much of the answer comes from your content and how early it appears, calculated only from answers where you were already cited.
• Mention Rate: How often assistants name your brand in their answer, whether or not they link to your domain, measured against the same sampled category answers as Citation Rate.
• Share of Voice: Your slice of citations compared to competitors, showing who wins the prompts you lose, based on all citations in the sampled category panel.
• Industry Fit: Whether an AI assistant views your site alongside your actual competitors, measuring which brands consistently appear together in the same category panel.

The critical detail most coverage misses: these numbers come from snapshot data, not live queries. Cloudflare runs its category panel once per category and reuses that baseline across all accounts in that category, so your scores load from a frozen moment rather than reflecting what Claude or GPT say about you today.

Why Does It Only Track Two AI Assistants?

Cloudflare explicitly states it probes "the leading assistants," which today means Claude and OpenAI's GPT. The company's own hedge is the word "today," suggesting this scope could expand. But right now, every visibility number carries that limitation. If your website appears frequently in Perplexity's answers or Gemini's results, this dashboard will not capture it.

This matters because different AI assistants have different user bases and use cases. Perplexity, for example, is specifically designed as an answer engine and attracts users looking for cited sources. Excluding it from a citation-tracking tool creates a significant blind spot for publishers and content creators trying to understand their visibility across the AI landscape.

How to Interpret Your Citation Scores Correctly

• Read Prominence alongside Citation Rate: A Prominence score that improves while your citation count shrinks is not progress; it is a smaller sample being measured. Always compare these two metrics together to avoid misinterpreting your performance.
• Understand that Mention Rate counts naming, not sentiment: If an AI assistant mentions your brand negatively or in passing, Mention Rate still counts it. The metric does not distinguish between favorable and unfavorable mentions.
• Verify your competitor set is accurate: Share of Voice depends on Cloudflare correctly inferring which companies are your actual competitors. If the inferred competitor set is wrong, the metric becomes unreliable.
• Check the AI Operator Activity panel for network-layer data: The genuine advantage of Cloudflare's position as a content delivery network (CDN) lies in the separate AI Operator Activity panel, which shows observed crawl and referral traffic from AI assistants, not in the Citation Rate metric itself.

Cloudflare's own press release criticizes the methodology the dashboard uses. It calls prompt sampling "limited in scale and prone to inconsistency," and the visibility metrics are produced by prompt sampling. This self-aware critique suggests the company recognizes the tool's limitations even as it launches it.

What About the Agent Readiness Checklist?

The launch also includes Agent Readiness, which Cloudflare describes as an "existing tool" that moved into the dashboard on August 6. The tool actually launched on April 17, 2026, as a free scanner at isitagentready.com. It grades websites on 20 checks to determine whether they are ready for AI assistant crawling and citation.

Of the six checks Cloudflare names in the Advanced integration bucket, four are unratified standards. WebMCP's own specification states it is not a W3C Standard. Web Bot Auth is an IETF draft. Cloudflare calls the MCP Server Card a proposal currently in draft, and the Agent Skills index is a Cloudflare proposal. Only the checks based on published RFCs (Request for Comments) represent finalized internet standards.

The Bigger Picture: Why This Matters Now

Cloudflare's pitch for this tool rests on a traffic argument. The company states that by its count, fewer than half of all HTML page requests now come from a human. It immediately qualifies this: "Not all of those machines are agents acting for a person, but that share is growing fast." Both parts matter. The denominator is HTML page requests observed across Cloudflare's network, not all internet traffic and not visits.

This reflects a real shift in how content gets discovered. As AI assistants become primary interfaces for information retrieval, websites that do not appear in AI-generated answers effectively disappear from a growing portion of search traffic. Cloudflare's tool addresses a genuine need: helping publishers understand their visibility in this new layer of the internet.

The dashboard is available in early access by request, and no pricing has been published. Cloudflare's forward-looking-statements section explicitly lists as uncertain the timing of when the AEO Suite or any of its related features will reach general availability, so there is no guaranteed launch date and none of it should be budgeted as a shipped, priced product yet.