agentspeed.
← all tools
Tool

AI Search Citation-readiness

Score how well-positioned a page is to be cited by AI answer engines for a given query. We do NOT call ChatGPT, Perplexity, or Claude live. We score the agent-readable signals AgentSpeed can audit deterministically: query-term match, schema types matching query intent, authority signals, freshness, and llms.txt presence. Same input → same score, every time.

Tool

Deterministic. Not a live ChatGPT / Perplexity call.

https://nextjs.org/docs/app/getting-startedquery: how to set up next.js app routerinformational
Citation-readiness · deterministic, not a live model call
71/ 100medium likelihood
Term coverage

4 substantive query terms after stop-word filtering. title 3/4, headings 1/4, body 4/4.

TechArticle

Findings

  • Strong title matchtitle.match

    3/4 query terms in the title.

  • Query terms appear in headingsheadings.match

    1/4 terms found across H1/H2/H3.

  • Body content covers the querybody.match

    4/4 query terms in the body.

  • No schema types relevant to informational queriesschema.intent_match

    Expected one of: Article, BlogPosting, FAQPage, HowTo, WebSite, Organization. Detected on this page: TechArticle.

    fix Add JSON-LD for one of: Article, BlogPosting, FAQPage.

  • No Organization or WebSite schemaschema.authority

    Agents need a structured publisher signal to attribute citations to. Without it, attribution falls back to text inference.

    fix Add Organization + WebSite JSON-LD blocks site-wide.

  • No publication datefreshness

    For informational queries agents prefer recent content. No `article:published_time` or schema datePublished was detected.

    fix Add `<meta property="article:published_time" content="…">` and / or a datePublished field on Article JSON-LD.

  • llms.txt present on hostllms_txt.present

    Agents that consult /llms.txt before grounding answers have a curated index pointing at this page (or its cluster).

Suggested optimised answer block

Drop this somewhere on the page (or paste it into your CMS). Replace placeholder copy with real content. Agents weight short, direct answers near matching headings heavily for citation.

## What is set next app router?

<!-- Suggested optimised answer block for the query: "how to set up next.js app router". -->
<!-- Drop this somewhere on the page. Agents weight short, direct -->
<!-- answers near matching headings heavily for citation. -->

**Short answer.** Replace this paragraph with one or two sentences that directly answer "how to set up next.js app router". Use the actual query terms verbatim where possible.

**Why this matters.** One short paragraph of context that an agent can quote when expanding on the short answer. Avoid marketing copy; agents skip it.

**Related reading.** Link to two or three internal pages an agent could cite for follow-ups.
Watch this URL — get alerted when this changes

Turn this one-off check into a scheduled watch. We re-run it, diff the result, and alert you on a regression through the same channels as your agent alerts. Watches are a paid feature.

Manage watches at /watches. You’ll be asked to sign in if you haven’t already.

Examples

Click an example to run the tool with pre-filled inputs.

Why this matters for AI agents

Agents pick which sources to cite by reading the same signals search engines have always read, plus a few new ones. Pages that match the query in the title, name the answer in a heading, and declare structured authority typically beat pages that bury the answer in prose, no matter how good the prose is.

This tool is honest about what it is. It is not a live test of whether ChatGPT would cite you tomorrow. It is a deterministic score of the signals that agents read before deciding. Treat the score like Lighthouse: a useful proxy, not a guarantee.

Common fixes
Title does not match the query
Rephrase the page title so the primary query terms appear verbatim in the first 60 characters. Agents weight title match more heavily than body match.
No query terms in any heading
Add an H2 or H3 that names the query intent. Agents skim by heading; an answer block is only as good as its label.
Wrong schema for the query intent
Pricing queries → Product / Offer. Informational queries → Article / FAQPage / HowTo. Use the structured-data validator to add the right schema.
No Organization or WebSite schema
Add Organization + WebSite JSON-LD blocks site-wide. Without a structured publisher signal, agents fall back to text inference for attribution.
Old article with no datePublished
Add `<meta property="article:published_time" content="…">` and a datePublished field on Article schema. Agents prefer recent content for informational queries.
Linked rubric checks
FAQ
Is this a live ChatGPT / Perplexity call?
No. We do not call any model. The score is a deterministic readiness signal, so the same input always produces the same output. Predicting whether a specific model will cite a page on a specific day requires a live call that is rate-limited, costly, and non-deterministic. We deliberately do not pretend to do that here.
How is the score computed?
Six axes. (1) Query-term match in title: up to 20 points off if no match. (2) Query-term match in headings: up to 12 points off. (3) Body coverage: up to 15 off. (4) Schema types matching inferred query intent: up to 15 off. (5) Authority schema (Organization / WebSite): up to 8 off. (6) Freshness (informational queries only): up to 6 off. Plus a 6-point penalty if no /llms.txt on the host. Score clamped to 0–100.
How do you infer query intent?
Heuristic word presence: words like "buy", "price", "pricing", "subscribe" → transactional. Words like "what", "how", "why", "compare", "best", "guide" → informational. Words like "site", "login", "official" → navigational. Otherwise: unclear, treated as a navigational baseline.
What does "low / medium / high" likelihood mean?
A simple bucket on the score: 0–49 low, 50–74 medium, 75–100 high. The label is a convenience for sharing. The underlying deterministic score is what counts.
Will improving these signals guarantee citation?
No. Citation is also influenced by domain authority, content recency, the user's prompt phrasing, model temperature, and dozens of other variables we cannot audit. What this tool measures is the floor: the minimum signals an agent needs to consider you at all. Sites scoring high here are eligible; sites scoring low typically are not.
Run a full agent-readiness scan
Full scan

Score the entire site against 30 agent-readiness checks across discoverability, readability, structured data, actionability, and performance.

Get the full report for nextjs.org →