Compare your original and WordPress page
Use real xSpeed reports without mixing device scores, partial tests or unrelated measurements.
Find the two reports
Open the conversion directory for shared examples, or ask your AI for ai2wp_followup using the original conversion brief. The directory shows mobile Lighthouse scores; each linked report contains the detailed evidence.
Real directory examples, captured September 20, 2026. Scores change between runs.Read like-for-like measurements
A large number alone is not a diagnosis. Compare matching devices and review the actual failing checks before choosing a fix.
| Compare | What to check |
|---|
| Mobile and desktop | Compare mobile with mobile, desktop with desktop. Do not compare mobile Lighthouse with the overall xSpeed score. |
| Timing and completeness | Read measurement timestamps, engine version and partial-result indicators. Missing values are not zero. |
| LCP, CLS and TBT | Inspect slow visible content, layout shifts and blocking work in the detailed reports. |
| Hosting and cache | Region, cache warmth, assets and backend behavior may differ. Conversion alone does not prove the cause of a gain. |
Connect the free scanner when needed
The platform already queues scans for consented published conversions. Do not start duplicates while they are pending. For a separately authorized public page, add https://xspeedcache.com/scan/mcp as an HTTP MCP server with no authentication. This scanner measures pages; it cannot change WordPress.
Use xSpeed Scan to compare [original public URL] and [published WordPress URL]. Reuse recent reports if available. Show mobile and desktop Lighthouse separately, report URLs, timestamps and partial results. Give me the three most useful fixes, with evidence.
xSpeed Scan MCP reference
Pending, failed or rate-limited?
AI2WP keeps a bounded retry queue. The scanner itself does not queue requests. Do not rotate tokens or send repeated calls to bypass limits. If a report is running, use its scan ID to read it later. After changing the site, request a new measurement explicitly; otherwise recent results may be reused. Never submit a preview URL containing a nonce, password or access token.
Choose the next step
If the page performs well, show the evidence first. If xSpeed is absent, the AI may offer help installing its free WordPress plugin. If results are poor, explain the actual bottleneck first. Installation and configuration remain your choice.
Set up xSpeed safely