Use Case – Onsite Audit

From:

SISTRIX Content Team

Modified: 18.08.2026

SISTRIX MCP Servers

• Use Case - Onsite Audit
• Use Case - Prompt Research Current Status
• Use Case - Topic Prompt Research
• Use Case – Initial SEO Analysis
• Use Case – Content Gap Analysis
• Use Case – Content Audit
• Use Case - Automatic Keyword Strategy
• Use Case - GEO Analysis
• Back to overview

Uses

• Immediate overview: of all critical errors without having to manually scroll through hundreds of lines.
• Priorisation based on degree of severity: Error, Warning, and Notice instead of an unsorted list.
• Directly usable list of affected URLs: including linked source pages, ready for use for developer tickets.
• Comparison of timelines: shows based on the crawl history if taken measures are actually working.
• All-encompassing audit without additional efforts: complete onsite-tech audit without further crawling or manual data evaluation.

How an onsite audit should be structured

A complete onsite audit covers five topic clusters with a total of 26 factors, that can all be directly retrieved via the SISTRIX MCP server. All retrievals are carried out via project_onpage for the relevant chosen project.

• 🔍 Crawling & Indexing: Sitemap.xml, URL redirection, 404 errors, canonicals, robots meta tag, crawling blockades, SSL certificate, Hreflang
• 📄 Content: Content quality and quantity, H-tags/headings, title, description, alt-description, placeholder text
• 🔗 Website & URL structure: Canonical URL, internal links, descriptive URL, broken links, click depths from starting page on, URL hygiene, mixed content
• ⚡ Performance: Picture formatting and size, JavaScript & CSS, loading times, data sizes
• 📣 Social & Meta: Open Graph

26 factors in 5 topic clusters, all directly retrievable with the SISTRIX MCP server without the need for manual clicking through the overview. The specific SISTRIX retrievals (issue keys) per factor can be found in the separate reference table.

Prompt

Process

1. Choose project: Simply name an already existing SISTRIX Onpage Project for the domain you want to check, for instance “Project X”. The model will automatically find the relevant data record.
2. Pull crawl overview: project_onpage(view: overview) delivers the error, warning and notice numbers in a historical timeline.
3. Retrieve all issues types: project_onpage (view: crawl) delivers the complete breakdown for the 105 potential error types. Usually, the model does this automatically without you having to specifically ask it to.
4. Pull the relevant URLs per factor: project_onpage(view:issue, issue...) delivers the affected URLs with the fitting issue key from the five topic section above, for instance page_not_found, missing_h1, thin_content, noindex, or open_graph_incomplete.
5. (optional) Prioritise: The model uses the amount, URL score and error type to delineate priority and recommendation for action.

Output / Result

Breakdown of errors by topic area, live example: sistrix.de

Data retrieved in real time via the SISTRIX Onpage project (project “sistrix de”, 38,765 pages crawled, 168 errors, 15,520 warnings, 13,155 notices):

Overview of errors for the live example sistrix.de

Type of error	Amount	Type	Priority	Recommended action
SSL certificate expiring	1	Notice	1	Already expires on 21/08/2026: renew in time, otherwise browser warnings and loss of trust are likely.
404, page not found	73	Error	1	Set up 301 redirects to suitable target pages or update internal links to the live URL; clean up the source pages identified in the drilldown first.
Image not found	71	Error	1	Replace missing image files or remove the tag to avoid broken-image icons.
Server error 500	4	Error	1	Escalate immediately to development: server errors completely block crawling and the user experience.
Title completely missing	1	Error	1	Add the title tag. With only 1 URL affected, this is done in a few minutes: a classic quick win.
H1 missing	5	Warning	1	Add the H1 tag; the small number of pages also makes this a quick win.
Incorrect URLs in sitemap	15	Error	2	Clean up the sitemap.xml: only include live, accessible, indexable URLs, then resubmit it to GSC.
Title too long	620	Warning	2	Shorten titles to around 55–60 characters, placing the most important keyword at the start, ideally via a template rule rather than manual editing.
Description missing	594	Warning	2	Add meta descriptions, prioritizing category and blog templates first, then individual pages.
Thin content (< 250 words)	51	Warning	2	Expand the content of the pages, consolidate them, or set them to noindex if they have no independent search value.
H1 identical across multiple pages	50	Warning	2	Assign an individual H1 to each page to avoid duplicate-content signals.
Alt attribute missing	213	Warning	2	Add alt attributes at a technical level; this is usually a template or CMS fix rather than manual maintenance.
Canonical points externally	43	Warning	2	Check whether the external canonical is intentional, e.g. for syndication purposes; otherwise correct it to a self-canonical.
Image too large	335	Warning	2	Compress images or switch to WebP, prioritizing high-traffic pages.
Alt attribute empty	2.406	Warning	3	Fill in alt texts gradually, prioritized by URL score or traffic rather than all at once.
Only 1 inbound link	98	Warning	3	Strengthen internal linking from thematically relevant pages.
Broken external links	8.950	Warning	3	Group by target domain and clean up or remove in batches. The high number suggests outdated link lists.
Noindex set	151	Notice	3	Check whether all URLs were deliberately excluded; an accidental noindex completely prevents rankings.
Open Graph incomplete	1.467	Notice	3	Add OG tags such as title, image, and url so that social shares and AI-generated snippets are displayed correctly.
Duplicate content (fuzzy)	103	Notice	3	Consolidate similar pages or point them to the main version via a canonical tag.

Combinations and Variations

• Traffic-light-reporting: Only retrieve Errors as a compact red-yellow-green overview to send to clients
• Progress tracking: project_onpage(view: overview) to compare across multiple crawl datasets to test the effect of actions taken
• New-client-pitch: Focus only on the most severe 5 to 10 Errors to quickly show your further value in an initial consultation.