Local Landing Page Differentiation Auditor
Find repeated service blocks, location-name swaps, thin city pages, reused claims, and missing local evidence across branch and service-area landing pages. The report separates content differentiation from local evidence coverage so every finding can be reviewed.
Do not include data you are not authorized to review. A normal refresh clears the working set.
- Add data
- Map and configure
- Preview extraction
- Analyze and review
- Export evidence
1. Add page data
Use one crawl CSV/TSV, multiple saved HTML files, a ZIP containing supported files, or pasted page records. Maximum: 10,000 pages, 50 MB per job, and 5 MB per HTML file.
--- PAGE ---. Supported headers: URL, Location, Type, Title, Description, H1, Address, Phone, Canonical, FAQ, Staff, Projects, and Testimonials.2. Map fields and configure comparison
3. Review extracted content
Confirm that main content is present and global navigation, scripts, styles, footers, and selected boilerplate were excluded. Up to 25 previews are shown.
4. Review results and evidence
Similarity and evidence are separate. Open individual issues to see observed evidence and a concrete verification or content action.
Portfolio overview
Use the summary above to size the review, then open the page matrix for priorities, the shared-block library to classify legitimate boilerplate, and the evidence matrix to find page-specific gaps. Thresholds control review priority only.
What the auditor compares
The analyzer preserves heading and paragraph boundaries, normalizes Unicode text, creates word shingles, and compares likely nearest neighbors with MinHash/locality-sensitive indexing on larger sets. It separately inventories NAP, directions, transport, accessibility, neighborhoods, service boundaries, staff, projects, testimonials, local events or constraints, images, FAQs, links, and local entity schema.
Supported inputs and safeguards
CSV/TSV, saved HTML/HTM, ZIP archives, and pasted records are accepted. Scripts, styles, navigation, headers, footers, forms, and selected simple selectors are removed from HTML analysis. Uploaded HTML is never rendered and external resources are not loaded. Encrypted ZIPs and malformed files fail with recoverable messages.
Deliberately unsupported in this MVP
The tool does not crawl live pages in bulk, run user regular expressions, execute uploaded code, verify factual accuracy, rewrite copy, invent testimonials or addresses, or interpret server configuration. Complex CSS selectors and dynamic JavaScript-only content extraction are not supported.
Frequently asked questions
Does this tool upload my crawl export or saved HTML?
No. Files and pasted text are processed locally in your browser and a browser Web Worker. The analyzer does not send file contents, URLs, titles, schema, or issue examples to the server or analytics.
Does a high similarity percentage mean Google will penalize a page?
No. Similarity is a review-priority signal, not a penalty prediction. Shared legal, service, and safety copy can be legitimate. Review whether each page still gives visitors useful and verifiable local information.
What should I map from a crawl CSV?
Map a URL and extracted main text or HTML first. Optional fields such as location name, type, title, description, H1, NAP, schema, inlinks, breadcrumbs, images, testimonials, staff, projects, and FAQs improve the evidence inventory.
How are large portfolios compared without checking every possible pair?
The analyzer normalizes block text, creates word shingles and MinHash signatures, then uses locality-sensitive hash bands to find likely nearest neighbors. Small sets use exact pair coverage; large sets compare indexed candidates instead of unbounded all-pairs combinations.
What does local evidence coverage prove?
It proves only that supplied text or fields contain evidence patterns such as an address, directions, staff, projects, FAQs, local images, links, or local entity schema. A person must verify accuracy, ownership, consent, and whether each claim applies to that location.
Can I mark legal or service boilerplate as acceptable?
Yes. In the shared-block library, mark a repeated block as acceptable boilerplate. The tool recomputes page metrics from normalized blocks without reparsing your files.
Reviewed Jul 2026 · Sources and limitations
Review details: 2026-07-13 · Marc LaClear · v1.0
Reference sources:
- Google Search Central documentation
- Google Search Central crawling and indexing docs
- Google structured data guidelines
- Schema.org vocabulary
- MDN Web Docs for HTTP and HTML references
Known limits:
- The tool detects presence and repetition; it does not verify that local claims, addresses, testimonials, staff, or projects are true.
- Similarity thresholds prioritize human review. They do not predict rankings, indexing, or a duplicate-content penalty.
- Saved HTML may omit content that a live page loads only after JavaScript runs.
- ZIP support requires a modern browser with DecompressionStream. Encrypted archives and unsupported compression methods are rejected.
- Custom exclusions support simple tag, .class, and #id selectors in the MVP; complex CSS selector syntax is deliberately unsupported.