Free Website SEO Health Check — evidence, not a vanity score.

Free Website SEO Health Check by Jev for Technical SEO and Content Quality

A healthy website should be easy for people to use and easy for search engines to crawl, understand, and revisit. Yet problems are often distributed across dozens or hundreds of URLs: a sitemap can contain pages that no longer work, important content can be left without internal links, templates can omit titles or headings, and server responses can become slow or unreliable. Reviewing every URL manually is difficult, especially when a site has multiple sitemap files.

The nqzai Free Website SEO Health Check by Jev gives you a practical, evidence-based starting point. Enter a website URL and the tool fetches robots.txt, discovers same-site XML sitemaps, follows nested sitemap indexes, and checks up to 2000 URLs. It combines technical SEO diagnostics with content-quality, internal-architecture, and AEO/DEO signals, then presents a composite score, subsection scores, page-level evidence, real-time token telemetry, and an AI remediation prompt that can be shared with a developer or SEO agent.

This is a diagnostic crawl, not a Google ranking prediction. A higher score does not guarantee rankings, traffic, indexing, or conversions. The purpose is to reveal observable problems, explain why they matter, and give you a prioritised path to verification and repair.

Feature Comparison: nqzai Free Website SEO Health Check by Jev vs. Other SEO Crawl Tools

Pricing Comparison

What the Website Health Check Audits

The audit is organised around the parts of a website that most directly affect discoverability, technical access, structural clarity, editorial usefulness, and autonomous AI decision extraction. The report displays subsection scores so you can distinguish a crawl problem from a content problem instead of hiding every signal inside one number.

1. Crawlability and sitemap discovery

The scan begins with the site's robots.txt file. It records whether robots.txt is reachable, whether it contains a root-level disallow rule, and whether it declares sitemap locations. It then checks common sitemap entry points, including /sitemap.xml and /sitemap_index.xml, as well as sitemap URLs declared in robots.txt.

Many sites use more than one sitemap. A sitemap index may point to separate files for blog posts, products, pages, images, or different content types. The audit follows same-site nested sitemap references and inspects up to 50 sitemap files while collecting no more than 2000 page URLs. The report separates URLs discovered from URLs successfully fetched — being listed in a sitemap does not prove that a URL is accessible, indexable, canonical, or useful.

2. HTTP status and response-code health

Every discovered page is checked for its final HTTP response. The report groups responses into successful 2xx results, redirects, 4xx client errors, 5xx server errors, failed fetches, and no-content responses where relevant. A 404 or 410 may indicate a removed URL that still appears in a sitemap or internal link. A 5xx response can indicate a server-side failure that prevents crawling. The tool reports the observed status and leaves the editorial or engineering decision to the site owner.

3. Internal links and site architecture

The crawler extracts crawlable, same-origin links from fetched HTML. It counts inbound links, estimates the maximum click depth from the root page, identifies pages that appear to have no inbound internal links, and surfaces internal targets that were not among the fetched pages. The audit counts generic anchor labels such as "click here," "read more," or "learn more," because those labels say less about the destination than a specific, natural description. The architecture score is a deterministic crawl-graph signal, not a PageRank calculation.

4. Metadata, duplicate content and keyword cannibalisation

For each successfully fetched HTML page, the audit checks the presence of a title, H1, meta description, canonical link, viewport declaration, visible word count, and robots noindex signal. It also records response time, final URL, HTML size, and internal-link count. Missing titles, headings, or descriptions are useful template-level signals. Word count is treated as a thin-content heuristic rather than a target. Canonical and noindex values are surfaced as evidence for investigation.

The same subsection scores duplicate content and keyword cannibalisation, and it reports them as page pairs rather than as a count you cannot act on. Two comparisons run across every crawled page. The first builds a token fingerprint from the title, H1 and meta description together and flags any pair whose overlap reaches 0.82; that is the near-duplicate signal. The second compares the title alone — ignoring pages whose title carries fewer than two meaningful tokens — and flags any pair at 0.75 or above; that is the cannibalisation clue, because two pages whose titles say almost the same thing are usually chasing the same query. Tokens are lower-cased, stripped of punctuation, and words of two characters or fewer are discarded, so “the” and “of” never create a false match.

The report lists every flagged pair by URL, side by side, with each page’s word count — so the finding is a worklist, not a statistic. Neither signal is a verdict. Similar titles are a reason to open two pages and compare the query each is meant to answer; they are not proof that either page should be deleted, merged or redirected. A product series, a location series, or a genuine comparison set will legitimately share title structure. Confirm intent before consolidating anything, and remember that a redirect is much harder to walk back than a title edit.

5. Helpful-content and GEO-oriented signals

The audit includes lightweight heuristics inspired by people-first content principles and the existing GEO scorecard work. It looks for signals associated with clear authorship or review, dates or update language, caveats and alternatives, external source references, and content that appears unusually thin. Duplicate intent and cannibalisation are scored separately, in subsection 4 above. These are not AI judgments about truth or expertise — every heuristic requires human verification.

6. Sitemap accuracy and robots.txt alignment

The sitemap subsection compares discovered sitemap URLs with pages that were successfully fetched. It reports the number of sitemap files, whether the sitemap cap was reached, and whether URLs appear to resolve successfully. Keep sitemap files focused on URLs that are intended to be canonical and indexable. Remove stale, redirected, blocked, or broken URLs where appropriate.

7. Server response performance

For each fetched page, the scanner records elapsed server-side response time. It reports slow pages above a 1.5-second diagnostic threshold and calculates a median response time for successful responses. This is useful for identifying response outliers, overloaded routes, and broad server-side latency patterns. It is not a PageSpeed Insights report and does not include browser rendering, JavaScript execution, LCP, INP, or CLS.

8. HTTPS and basic security signals

The tool checks whether crawled pages resolve over HTTPS, whether robots.txt is reachable, and whether fetched HTML contains references to insecure HTTP resources that could represent mixed content. HTTPS is a baseline expectation for modern websites. Mixed-content references can create browser security warnings or prevent resources from loading. This crawl does not replace certificate-chain validation, expiry monitoring, or a security penetration test.

9. Mobile experience and Core Web Vitals

The scanner checks for a viewport meta tag but does not render every page in a mobile browser. It does not claim to measure mobile usability, responsive layout, tap targets, LCP, INP, or CLS. The report marks browser-based mobile and Core Web Vitals measurement as not measured rather than assigning a misleading score. For a complete performance review, pair this audit with PageSpeed Insights, Lighthouse, or Search Console's Core Web Vitals report.

10. AEO and Decision Engine Optimisation (DEO)

Search engines and generative answer engines (such as Perplexity, ChatGPT Search, and Google AI Overviews) consume websites differently than traditional crawlers. While traditional SEO focuses on keyword discovery and link authority, Answer Engine Optimisation (AEO) and Decision Engine Optimisation (DEO) optimize for extraction, structured comparison, and autonomous agent decision-making. Autonomous research agents look for structured entities, machine-readable comparison tables, verified social meta cards, transparent pricing signals, and explicit trust policies (such as terms, privacy, refunds, or SLA pages).

The health check audits six foundational AEO and DEO signals across every fetched page:

Sites that structure their key propositions into machine-readable formats are cited and recommended by AI decision engines with substantially higher confidence than sites relying exclusively on unstructured prose.

11. Real-Time Telemetry and Token Consumption (Powered by Jev)

Modern AI-native tools must be fast, cost-transparent, and lightweight. The Free Website SEO Health Check is powered by TypeSafe's Jev system, executing deterministic heuristic analysis and entity extraction in milliseconds without sluggish multi-second LLM waiting times.

At the top of every crawl report, a real-time telemetry box reports:

• Tokens Consumed: The exact volume of text and structural tokens analyzed across page bodies, title fingerprints, schema entities, and sitemap indices. This gives engineering teams full transparency into the computational budget of an automated crawl.
• Time Taken: Total wall-clock elapsed crawl duration in seconds, showcasing how bounded concurrency and asynchronous parsing deliver comprehensive multi-page audits in seconds.
• Evaluated by Jev: Verifying that the analysis ran on TypeSafe's fast cognitive engine rather than an unverified black-box simulator.

How the Scan Works

You enter a complete HTTP or HTTPS website URL. The server fetches robots.txt and gathers declared sitemap locations. Common sitemap endpoints and nested sitemap indexes are inspected. Same-site page URLs are deduplicated and capped at 2000. Pages are fetched server-side with a crawler user agent and bounded concurrency. HTML, links, metadata, response codes, timing, and selected content signals are extracted. Section scores are calculated with coverage awareness. The report renders summary cards, visual score bars, paginated audit findings, raw page results, and a remediation prompt.

The production implementation records actual crawl duration, page responses, sitemap files, fetch concurrency, and failed responses. A run that finishes quickly may reflect a small site, cached or fast responses, or concurrent fetching — it should not be confused with a simulated result.

Understanding the Composite Score

The composite score combines available subsection signals with weighted coverage. The tool considers content quality, crawl and indexability, HTTP status health, internal architecture, content hygiene, AEO and decision engine signals, sitemap health, performance, mobile availability, and security. Mobile and Core Web Vitals is excluded when browser measurement is unavailable rather than scored as a failure — this prevents an unmeasured category from distorting the result. Coverage is displayed separately so a 90/100 score from a partial crawl is not presented as equivalent to a 90/100 score from complete coverage.

Use the composite as a prioritisation aid. Start with critical crawl failures, server errors, blocked important pages, broken internal targets, and accidental noindex directives. Then address template-wide metadata gaps and architecture issues. Finally review content quality, performance improvements, and editorial opportunities. Always validate findings against the live site and source code before publishing changes.

How to Read the Report by Website Type

Different websites should interpret the same signal differently. On an ecommerce site, a missing product title or broken category link can affect many URLs because one template is repeated at scale. On a publisher or blog, the priority may be author clarity, update integrity, source quality, and links from high-level topic pages to detailed articles. On a SaaS site, a short pricing or login page is not automatically thin content — the important questions are whether public feature pages explain the product clearly and whether documentation is discoverable.

Local businesses should review location pages for unique usefulness rather than duplicating a city name across identical templates. International sites should pair this crawl with hreflang, language, and canonical checks. JavaScript applications should verify that important content and links are available in the HTML returned to crawlers. In every case, the score is a triage layer — the correct fix depends on the page's purpose, business context, and intended audience.

Measurement Integrity and Evidence Boundaries

The audit deliberately separates observation from interpretation. "HTTP 404" is an observation. "This page should be restored, redirected, or removed from the sitemap" is a recommendation that requires context. "Two pages have similar title tokens" is a duplicate-intent clue, not proof of cannibalisation. This distinction is especially important when a site uses templates, authentication, consent banners, client-side rendering, or request-dependent content. A server-side crawler may receive a different representation from a logged-in user or a mobile browser.

The AI Remediation Prompt

After the scan, you can copy an evidence-bound remediation prompt. It includes the audited domain, discovered page count, sitemap count, composite score, coverage, crawl timestamp, subsection signals, and a sample of priority page evidence. The prompt instructs a coding or SEO agent to verify observations, fix problems in priority order, preserve accurate claims, and return actions with evidence, affected templates, dependencies, owners, validation steps, and acceptance criteria. The prompt is a structured handoff, not an automatic deployment — review proposed edits, test them in a staging environment, and re-run the health check after release.

What This Tool Does Not Claim

The Free Website SEO Health Check by Jev does not guarantee rankings, indexing, traffic, Core Web Vitals, or compliance with every Google system. It does not access private Search Console data, infer search demand, evaluate backlinks, validate every structured-data implementation, or replace a browser-rendered audit. Its content and duplicate-page signals are heuristics. Its crawl sees the HTML returned to the server-side request and may not capture content injected only after JavaScript execution. The tool also does not make editorial decisions for you — a score should lead to investigation, not a blanket rewrite.

Recommended Workflow After Your Report

For developers

Export the failing URLs and group them by template. Verify status codes, redirects, canonical tags, robots directives, sitemap generation, and internal-link components. Fix shared infrastructure before making one-off page edits. Confirm the response and rendered experience after deployment.

For SEO teams

Compare sitemap URLs with Search Console indexing data. Review orphan candidates and broken targets against the intended information architecture. Map duplicate-like pages to search intent, then consolidate or differentiate only when evidence supports the change.

For editors

Review thin or weakly signalled pages for audience purpose, first-hand experience, sourcing, authorship, freshness, limitations, and alternatives. Add genuine evidence and useful explanation. Do not add filler paragraphs simply to increase word count.

For everyone

Save the report date, record the changes made, and re-run the same audit after implementation. A health check becomes more valuable as a repeatable baseline rather than a one-time score.

Frequently Asked Questions

Is this a free technical SEO audit?

Yes. It is a free website health and technical SEO diagnostic for up to 2000 same-site URLs discovered from sitemap files or same-site links, with no account or sign-up required.

Can it scan multiple sitemaps?

Yes. It follows nested sitemap indexes and sitemap locations declared in robots.txt, subject to bounded sitemap and page limits.

Does the tool crawl the entire website?

It crawls up to 2000 discovered URLs per run. For sites larger than 2000 pages, review sitemap segmentation and run focused audits where necessary. The report states when the cap is reached.

Does it use an AI model to judge every page?

No. The production health check uses a deterministic server-side crawl and heuristic analysis. It does not claim to make per-page LLM content judgments. The remediation prompt is generated from crawl evidence and is intended for human-reviewed follow-up.

What is Decision Engine Optimisation (DEO) and why does the tool check it?

Decision Engine Optimisation (DEO) is the practice of structuring website content and metadata so that autonomous AI agents (such as OpenAI Operator, Claude Computer Use, Perplexity, and Enterprise Copilots) can evaluate, compare, and recommend your product to human buyers. Unlike traditional SEO which targets search query rankings, DEO focuses on machine-readable parameters: side-by-side comparison tables, transparent pricing, Schema.org entities, and verified trust policies.

What does the 'Tokens Consumed' signal box measure?

The 'Tokens Consumed' counter at the top of the report measures the number of natural-language and structural tokens processed during your scan by Jev. This includes crawled body text, metadata fingerprints, Schema entities, and internal link graphs. It highlights the efficiency of Jev's fast evaluation pipeline compared to traditional heavyweight LLM calls.

Does a perfect score mean my site will rank?

No. The score reflects the checks available in this diagnostic. Rankings depend on many factors, including relevance, competition, quality, links, user experience, indexing decisions, and search context.

Why are Core Web Vitals marked as not measured?

Core Web Vitals require browser-rendered lab or field measurement. This server crawl reports response timing and viewport markup but does not pretend those are LCP, INP, or CLS results.

How should I use a duplicate or cannibalisation warning?

Open the relevant pages and compare their purpose, audience, claims, and search intent. The warning fires when a page pair reaches 0.82 token overlap across title, H1 and description together (near-duplicate), or 0.75 on the title alone (cannibalisation clue). The report names both URLs in each pair. Similar titles or metadata are a clue for review, not proof that pages should be deleted or merged — a series of product, location or comparison pages will share title structure by design.

Is the remediation prompt safe to paste into an AI agent?

It is designed as a careful starting brief. Paste it into an agent that can inspect the repository, but require verification, staging tests, human approval, and acceptance criteria before publishing changes.

Start Your Free Website SEO Health Check by Jev

Use the nqzai Free Website SEO Health Check by Jev to establish a clear technical and content baseline for your site. Enter your root URL, review the crawl evidence, inspect the subsection scores, expand the item-level findings, and copy the remediation prompt when you are ready to plan fixes. The most useful audit is not the one with the most alarming score — it is the one that connects a real URL to a verifiable problem, explains why the problem matters, and gives your team a safe way to resolve and re-test it.