Methodology revision: October 25, 2026. These are implementation limits, not a promise of rankings, traffic, revenue, or Google indexing.
A detailed page score and a bounded website crawl
The report contains two different evidence sets. The detailed page score evaluates the submitted page's fetched HTML and available response evidence. The website crawl separately inspects up to 100 public pages discovered on the submitted hostname, within discovery, request and time limits. The page score is not an average score for the entire website.
The crawl summary reports the pages actually inspected, including observed errors, repeated titles and other cross-page patterns. Fewer than 100 pages may be inspected. An undiscovered or excluded URL is not a passed check, and the report does not certify that every page on your domain works. Content inserted only after JavaScript runs may not be represented.
A high score does not mean your content meets every visitor's need or that your site is legally compliant, secure, accessible, or indexed. Use the affected URLs and evidence behind each finding to decide what needs attention.
How the page score is calculated
Each included check has a weight. A pass earns its full weight, a warning earns half, and a failure earns zero. The score is the rounded percentage of earned weight divided by total included weight. It is not a percentile or an industry benchmark.
score = round(100 × sum(weight × result) / sum(weight))
result: pass = 1; warning = 0.5; failure = 0For example, three illustrative checks weighted 10, 8, and 6 with pass, warning, and failure earn 14 of 24 points: 58/100. This miniature example is not an actual customer score.
Current HTML-check weights
| Check | Weight |
|---|---|
| HTTPS protection | 10 |
| Successful response | 10 |
| Page title | 10 |
| Meta description | 8 |
| Primary heading | 8 |
| Canonical URL | 8 |
| Mobile viewport | 8 |
| Indexing directive | 10 |
| Structured data | 6 |
| Structured data validity | 5 |
| Image alternatives | 6 |
| Internal links | 6 |
| Visible page copy | 10 |
| Document language | 4 |
| Character encoding | 3 |
| Site icon | 2 |
| Open Graph title | 3 |
| Open Graph description | 3 |
| Open Graph image | 3 |
| X/Twitter card | 2 |
| Section headings | 4 |
| Mixed content | 7 |
| Image dimensions | 4 |
| Off-screen image loading | 3 |
| Server response time | 7 |
| HTML document size | 5 |
| Cache instructions | 3 |
| Strict transport security | 4 |
| Content security policy | 3 |
| MIME sniffing protection | 2 |
| Referrer policy | 2 |
Presence checks are not quality guarantees
- Title and description lengths are heuristics, not Google’s fixed limits. A warning can be reasonable for a particular page.
- A detected canonical does not confirm that its destination is suitable or that Google selected it.
- The HTML indexing check does not replace a full review of robots.txt, HTTP headers, access restrictions, and Google’s own indexing decision. Use the Indexability Checker for additional technical evidence.
- An internal-link count does not test every destination. An alt attribute can exist without describing an image well.
- Valid JSON is not the same as eligibility for a Google rich result. Word count is not a content-quality measure.
- Optional performance or enrichment results have their own source and scope; do not treat them as a live user-experience measurement of every visit.
Missing evidence is not zero
A measured zero is different from an unavailable metric, an unconnected property, a provider error, or an expired report. Do not infer zero demand or zero traffic from a blank result. Google data depends on the selected property, permissions, date range, and data availability. Keyword volumes, difficulty, and observed rankings are provider estimates, not guaranteed live rankings or forecasts.
Freshness and repeat checks
Use the date on each result. Saved audit reports expire after 30 days; a saved report is not a new scan. Matching market reports can be reused for seven days, and Google evidence is cached for up to six hours. A fresh request may use your allowance. Changes after the original scan require fresh evidence, not just a refreshed display of the old report.
What Fix Verification establishes
For supported selected findings, Fix Verification compares owned, unexpired audit evidence for the same submitted page and requires a newer completed audit after the task was marked done. Missing or mismatched evidence requires review. Passing those selected checks does not establish that all issues are resolved or that Google has recrawled, indexed, or ranked the page. Manual checks remain necessary where a guide says so.
AI recommendations need human review
AI suggestions interpret the supplied evidence and can be incomplete or incorrect. Review proposed wording, facts, schema, and technical changes before publishing. SEOPlan does not silently apply fixes to your website. Check original evidence and business relevance before prioritizing an AI recommendation.
Explore the synthetic sample · Read practical fix guides · Google: title links · Google: canonicalization