URLs

URLs are the core unit of a zued audit. Each URL you add is crawled, split into content chunks, scored for technical health, and tested against AI engine responses.

​

Adding URLs

Add pages on the Pages tab. Before your prompts are approved, you add them right there. Once they are approved, adding pages is part of Edit mode, except while the project has no pages (archived ones aside), when you can still add them right there. Adding pages is for the project’s owners and editors.

​

Managing pages

Once your prompts are approved, the Pages tab has an Edit mode, like the Prompts page. Adding pages, every action on a page, and every change in the topics card happen there, so reading your results never changes them. The one exception: while the project has no pages (archived ones aside), you can add pages without switching to Edit mode. Before your prompts are approved, or in a project with no pages at all, there is no Edit mode, and these actions sit on the Pages tab itself. Changing pages is for the project’s owners and editors. Viewers see every page, its results, and the reason when zued can’t analyse one, but they get no Edit mode, no page actions on the Pages or Personas tab, and no changes in the topics card. See Team members. Edit mode lists every page in three tabs:

• Active: the pages zued has read and measures, with how many prompts each one has.
• Needs attention: the pages zued can’t analyse, each with its reason and the actions that fit it.
• Archived: pages that are no longer tracked, with their history.

From there you can:

• Add pages by entering their addresses or importing a CSV. Export CSV downloads your active pages with their scores.
• Archive pages you no longer want to track. An archived page stops being measured: no more snapshots, prompts or analysis. Its page slot is freed, its prompts are archived with it, and all its history is kept.
• Restore an archived page to bring it back with its prompts. A page comes back when your plan has a free page slot and your prompt budget has room for its prompts. If you restore more pages than you have free slots, they come back in list order until the slots run out.
• Retry or Remove pages under Needs attention, as described in Pages we can’t analyse.
• Replace a page’s address, below.
• Bulk actions archive, restore, retry or remove many selected pages at once. zued reports what happened to each page: pages that couldn’t be changed stay selected, each with a note saying why.

While zued is reading pages, page changes are unavailable until it finishes. And while a project holds more pages than its plan allows, pages can’t be added, restored, retried or replaced until you archive or remove one.

​

Replacing a page

Replace gives a page a new address, for example when the page moved. For a page zued has read, you choose what happens to its prompts. There is no default, because the right answer depends on the new page:

• Keep its prompts, for a page that moved. The same prompts move to the new address with their answers, and those answers are measured against the new page’s content straight away, without asking the AI engines again. Choose this when the new page covers the same ground and makes the same claims. It is only offered when the page has prompts to keep.
• Start fresh, for a page about something else. The old page moves to Archived with its prompts and history, and the new address gets new prompts for you to review, like any page you add.

Either way, zued reads the new address first, while the old page stays live and measured. Nothing changes until the new address has been read and prepared for measuring. If it can’t be, the old page keeps its prompts exactly as they were, the new address shows up under Needs attention with its reason, and zued tells you the replace did not happen. Reading first means a mistyped or broken address never costs you a page you are measuring. The new address takes over the old page’s slot, so a project at its plan limit can still replace a page. When a replace does not happen, the new address holds a slot of its own until you remove it. When you keep a page’s prompts:

• Each prompt keeps its full answer history. Until its answers have been measured against the new page, it shows the alignment measured against the old one. A weekly snapshot that already ran keeps what it measured on the old address, and the next one measures the new page.
• The page’s own trend starts again at the new address. Page-level scores and week-over-week changes belong to an address, so the old address’s history stays with the old page under Archived.
• Restoring the old page later brings it back without the prompts that moved.

A page zued has read can’t be replaced while a measurement run is in progress, so no run measures a page halfway through its move: replace it once the run finishes. If a run starts while zued is reading the new address, the old page stays as it was and zued tells you how to try again. A page under Needs attention has no measured prompts to carry over, so replacing it asks nothing: zued reads the new address in its place, as described in Pages we can’t analyse.

​

Which URLs should you add?

zued is designed to audit the pages that AI systems directly cite when answering user queries. Prioritize pages that contain your core product, service, or expertise content. These are the pages that enter the retrieval pipeline, pass re-ranking, and get attached as citations.

​

Recommended

These formats perform best in AI retrieval and citation:

• Product and service pages: especially with clear purpose lines, selection criteria, and testable claims
• How-to guides with numbered steps and direct answers
• Comparison and “vs.” pages: AI systems frequently cite these for purchase decisions
• FAQ pages where each question is a heading with a direct answer in the first paragraph
• Case studies with before/after metrics and specific results
• Data-driven content with original research or statistics

​

Not recommended

Pages like impressum, about us, contact, privacy policy, terms of service, and career pages are not recommended for zued audits. These pages still hold value: they contribute to your overall domain trust, entity clarity, and E-E-A-T signals that AI models factor into their reasoning. However, they influence AI visibility indirectly. AI systems rarely retrieve or cite them directly in their answers to user queries. zued measures and optimizes direct visibility: whether your page is retrieved, cited, and surfaced in AI-generated responses. Since legal, contact, and corporate pages almost never appear in that citation pipeline, auditing them won’t produce actionable insights or move your visibility metrics.

​

What zued does with each URL

1. Crawls the page using a real browser, the same way AI crawlers access your site
2. Chunks the content into meaningful segments that preserve the page structure
3. Scores the page for Technical Health: bot accessibility, JS rendering, Core Web Vitals, structured data, and meta tags
4. Maps audit prompts to the most relevant content on your page, including prompts that span multiple URLs in your project
5. Analyzes alignment between AI engine responses and your content
6. Cross-URL analysis: for each topic several of your URLs cover, shows how your pages work together and where the internal linking gaps are
7. Fact-checks verifiable claims (prices, specs, features) against AI engine responses, showing which facts AI reproduces accurately and which it gets wrong

​

Pages we can’t analyse

When zued can’t analyse a page you added, the page doesn’t disappear. It stays in your list under Pages we can’t analyse, with the reason and the actions that fit it. Each one holds a page slot until you decide what to do with it. Once your prompts are approved, the actions are in Edit mode, on the Needs attention tab. What each action does:

• Retry reads the page again. If it still can’t be analysed, it gets a fresh reason. It is also worth a try on a page with too little text that looks full when you open it.
• Replace swaps the page for a different address. zued removes the old page and reads the new address like any page you add.
• Use new address replaces a redirecting page with the address it sends visitors to.
• Remove takes the page out of your project and frees its slot. A page that already has history moves to Archived in Edit mode with that history, so you can restore it later.

​

Use as topic

A page with too little text often still stands for a subject you want to win in AI answers. Use as topic keeps that subject and lets the page go:

• The subject joins your topics, the ground you want to win without a page for it yet. zued suggests the page’s title, or a name taken from its address, and you can edit it before saving.
• The page is removed and its slot is freed for a page zued can analyse.
• Once your market is described, zued can turn your topics into prompts for the AI engines. If you haven’t described it yet, you can do that in the topics card on the Pages tab (in Edit mode once your prompts are approved).

​

AI crawler access

Some checks are about your site, not any single page. zued runs these once per audit and shows the result on every URL, labelled site-wide.

​

Does your server treat AI crawlers differently?

Allowing a crawler in robots.txt and actually serving it a page are two different things. Bot protection, WAF rules and CDN defaults sit in front of many sites and can turn an AI crawler away long before your own rules apply. zued checks this by comparing, over one connection, how your site answers a normal browser against how it answers requests identifying as AI crawlers. A difference between the two is the only part a site owner controls, so it is the only part worth reporting. zued goes one step further, because it has something a scanner does not: your actual engine results. If the affected engines are still citing your pages, the verified crawler is clearly getting through, so zued reports the finding for your records and applies no penalty at all. The same applies when the difference only touches crawlers behind no engine your audit queries: worth knowing, but it cannot move a number. Evidence only ever works in that direction. A citation proves an engine reached you; an engine staying quiet proves nothing, because that is also what ordinary low visibility looks like. So a difference we cannot rule out stays a finding, not a verdict: it costs a few points in the technical score and is never treated as a proven block.

​

Content Signals

Some sites publish preferences in robots.txt about how their content may be used by AI systems. These are not all equal: If you did not intend to opt out of AI answers, this is usually the fastest fix in the whole audit, and it is invisible in every content-focused tool.

​

AI content endpoints

zued also checks for llms.txt, llms-full.txt, .well-known/ai.txt and whether your site serves Markdown to clients that ask for it.

​

Agent readiness

Separately, zued looks for the emerging conventions that let autonomous agents act on a site rather than just quote it: service and API descriptions, agent capability files, and, for sites that sell things, agent commerce protocols. None of this affects your score, and for most sites everything here will read “not found” for some time yet. It is shown so the question is on your radar before it becomes urgent. The commerce checks only appear when your site actually publishes product markup, so a services business never sees a checkout checklist it has no use for.