Search readiness
Title and description length, one H1 and an outline that does not skip, alt text on images, a canonical that points at itself, an indexable robots meta, Open Graph tags, structured data, robots.txt and sitemap.xml.
Pikkola Speed & Search Studio, tool No. 20
One request from the edge measures what a visitor waits for, and reads what a crawler reads. Time to first byte, compression, HTML weight, the scripts that hold up the first paint, then the title, the description, the heading order, the alt text and the two files that tell search engines where your pages are.
A crawler reads your HTML and your crawl files, then decides what to index. A person waits for the first byte, the stylesheet, the script, and only then sees anything. The same page can do well on one and badly on the other, and fixing whichever is lower is usually a different afternoon's work.
Title and description length, one H1 and an outline that does not skip, alt text on images, a canonical that points at itself, an indexable robots meta, Open Graph tags, structured data, robots.txt and sitemap.xml.
Time to first byte from the edge, whether the response arrives compressed, how many kilobytes of HTML the browser must parse, and how many scripts and stylesheets sit in the head waiting to run.
No browser runs here, so there is no layout shift, no largest paint, no field data. JavaScript is not executed, so content rendered on the client is invisible to the crawl checks. Read the numbers as the cost of your response, not the whole experience.
robots.txt is fetched first and read for a wildcard block, a crawl delay and sitemap addresses. Then sitemap.xml at the conventional path, or wherever robots.txt points, whichever answers first. When neither answers, the audit writes both files for the audited site: a robots file that keeps admin and api paths out, and a sitemap that lists the page plus the internal links it found.
# robots.txt
User-agent: *
Allow: /
Sitemap: https://example.com/sitemap.xml
Watch reads up to ten domains once a day, the way this page reads one. When a response time doubles, a page stops answering, a certificate slips or a security header disappears, one message arrives in Slack, Discord or your own webhook.
$7a month
$70 a year, two months free
See what Watch checksTen domains, one check a day, $0 to start on this page.
The worker asks for the address and stops the clock when the response headers arrive. That is time to first byte. Redirects are followed one hop at a time and timed separately, so a chain that adds 900 ms is visible instead of hidden.
Up to 1.5 MB of the document, and nothing else. Scripts, images and fonts are never requested, so nothing on the page runs and no third party is contacted on your behalf.
The request leaves from Cloudflare's network, close to your hosting. Your visitor's own latency, their connection and their device are not part of this number, which is why it is a floor rather than a forecast.
No database, no account, no analytics, no log of audited addresses. The report lives in the page you are looking at and disappears when you close it. The audited site sees a request from Cloudflare, not from you.
No. Lighthouse runs a real browser on your machine, executes the JavaScript and models the layout. This makes one request from a server and reads the response. No render timing, no layout shift, no field data from real visitors. It answers in a second and works the same on a phone. Use it to fix the response, then use Lighthouse or Search Console for the rest.
The search half is not about speed. A missing description, two H1 headings, images without alt text or a sitemap no crawler can find cost points even when the page loads in 120 ms. Each finding names what was measured and what the value was, so read those before the total.
It goes to a Cloudflare Worker, which fetches it once and also asks for robots.txt and sitemap.xml. Nothing is written down. The target site sees a request from Cloudflare's network rather than from your connection.
A server that fetches any address it is handed can be pointed at a router page, an internal console or a cloud metadata endpoint. Every host is resolved and refused if it lands in a private range, and the check repeats on each redirect. A site that only exists on your laptop cannot be audited from here.
No. The request is anonymous, so you get the login page measured. Read those numbers as the cost of your sign-in screen, not of your product.
They are a correct starting point. The robots file keeps admin and api paths out, which may be wrong for your site, and the sitemap lists the page plus the internal links found on it, up to twenty. Change the paths and add the rest of your URLs before you publish.