Facebook tracking pixel Skip to main content

SEO and visibility

See why your page feels slow.

Checks how one live page performs: the three Core Web Vitals (LCP, INP and CLS), page weight, render-blocking files, mobile fit, images and the accessibility basics. With nothing connected, it reads the page source and flags likely trouble. Hand it a Lighthouse JSON report, a PageSpeed Insights API response or CrUX field data, and it reads the measured numbers from that file.

Request access Have us build it
  • Works with nothing connected
  • Says what was measured
  • Each fix names its vital

At a glance

Group
SEO and visibility
Command
/web-vitals
Uses
A page address, test data optional
Writes
A scorecard and a fix table
Sends
Nothing. Drafts only.

01When to use it

Run it when speed is the suspect.

  • A live page feels slow or janky, or converts below what its traffic suggests, and you want runtime evidence instead of guesses.
  • You want the three vitals, page weight, mobile fit, image hygiene and the accessibility basics covered in one pass.
  • You need the runtime picture the SEO skills miss. They read the HTML; this looks at what a browser does with it, measured with a browser connector or a test file, and inferred from the source without one.
  • A site audit calls it in as the lens for performance and accessibility.

Not for

02What it needs

One address is enough to start.

  • The address of one live page. On an agency setup, it must be a page of the client you are working for.
  • Your strategy and offers files, which say what the page is meant to win: a signup, a demo, a purchase or a lead.
  • Your team targets, if you set any. If not, it uses its default good marks: LCP under 2.5 seconds, INP under 200 milliseconds and CLS under 0.1.
  • Optional: a Preview or Chrome DevTools connector, which lets it load the page itself at phone and desktop sizes and take real readings.
  • Optional: a Lighthouse JSON report, a PageSpeed Insights API response or CrUX field data, plus any earlier run to chart against.

03What it writes

A report, screenshots and ledger rows.

final/web-vitals-{domain}-{date}.md
A header with the page, the date, the mode, the connector or file used, and whether the numbers are measured or inferred. Then a scorecard (the three vitals, weight, requests, mobile, accessibility) and the fix table: issue, severity, evidence, fix, the metric it moves and the estimated impact.
Screenshots
The page at mobile and desktop sizes, saved in the project folder when a browser connector was available.
Ledger rows
Rows in your numbers log, one per metric, appended and never edited. A value read from fetched HTML is marked inferred, and a reused old reading is marked stale.

04Limits

What it never does.

  • Rewrite the page, push a fix live, or touch the live site in any other way.
  • Make up a metric. A number it did not measure is labeled inferred, with a note on what a connector would confirm.
  • Give a number without its source: the connector, the tool or the fetched file it came from.
  • Sign off on legal compliance. Its accessibility notes are likely issues for a person to confirm.

What stays with you.

  • Making the changes. It hands over a fix list for someone else to apply.
  • Confirming each accessibility flag it raises.

05Sample output

Part of a fix table.

An excerpt for the product page of a made-up tea shop, run in Quick mode, so every number here was inferred from the page source.

final/web-vitals-example.com-2026-09-29.md Illustrative, not a client result.
Hero photo is the likely LCP element High
Evidenceoolong-hero.png, 1.8 MB, 2400px wide, shown at 600px
FixServe it as WebP or AVIF at display size
VitalLCP
Review widget blocks rendering Medium
Evidencereviews.js in <head> with no defer or async
FixAdd defer to the script tag
VitalLCP
Thumbnails have no set size Medium
Evidence6 images missing width and height
FixAdd width and height to each
VitalCLS

From the report

  • Mode: Quick. No connector and no test file.
  • Status: inferred. A browser connector would confirm the LCP element and its timing.
  • Conversion metric: purchases from this page, from the offers file.

06Related skills

Often run beside it.

07Questions

Before you run it.

Do I need Lighthouse or PageSpeed Insights first?

No. Quick mode runs with nothing connected and labels its findings inferred. A Lighthouse JSON report, a PageSpeed Insights API response or CrUX field data is an optional extra, and where one disagrees with the scan, the measured file wins and the scan finding stays only as a probable cause.

Which problems come first in the list?

The ones that hurt the vitals most for the most visitors, not the ones that weigh the most bytes. A field reading that fails for most real users outranks a minor lab-only finding, which outranks a risk inferred from the page source.

How deep does the accessibility check go?

It covers the basics: contrast on the main text and the main call to action, alt text on images that carry meaning, form labels, a visible focus state and the focus order through the conversion path. Contrast failures are noted by ratio against WCAG AA. For a dedicated WCAG 2.1 AA audit, run a11y-audit.

Have us build it

We can build it for you.

Start with the free plan. A person names the first fix.

Get your free plan Or run it yourself with the toolkit.