Skip to content
A laptop showing charts, standing on a grey sofa

How to check your website’s security and speed, and read the results

Seven free website tests, our scores before and after the fixes, and the traps in reading them. Plus what to fix first and how often to test.

A free website test takes a minute or two. Reading the result well takes longer. A letter grade does not say what is wrong, and one speed measurement can differ from the next.

On 13 November 2026 we ran the same set of tests on our own site that we use on clients’ sites. Mozilla’s security test gave us a B+, 80 points out of 100. After one change to our script policy we have an A+ with 120 points. Below is how we got there and how to read each test.

Seven tests at a glance

Test What it checks Result
PageSpeed Insights speed, accessibility, best practices and SEO in Google’s lab, plus data from real Chrome users 0 to 100 in four categories
Mozilla HTTP Observatory security headers, including the script policy (CSP), HSTS and the redirect to HTTPS F to A+, with a score
SSL Labs the certificate, protocols and encryption of the connection F to A+
internet.nl the website (IPv6, DNSSEC, HTTPS, headers) and email (SPF, DKIM, DMARC, encrypted transport) a percentage and a list of checks
Rich Results Test the structured data Google can use for rich results a list of detected items
WAVE the accessibility of a single page: contrast, image descriptions, form labels, structure errors and alerts
Google Search Console how Google sees the site: the index, redirects, Core Web Vitals from real users reports over recent weeks

The first six need no account. Search Console asks you to verify that the site is yours.

Speed: PageSpeed Insights

PageSpeed Insights shows two different measurements. According to Google, lab data helps you debug, because it is collected in controlled conditions. Field data shows how real Chrome users experienced the page over the last 28 days.

The score from 0 to 100 comes from the lab run. Google treats 90 or more as good, 50 to 89 as needing improvement and anything below 50 as poor. Field data is judged on three Core Web Vitals:

  • LCP, the time until the largest element appears: 2.5 s at most,
  • INP, how quickly the page responds to a click or tap: 200 ms at most,
  • CLS, how much the layout shifts while loading: 0.1 at most.

A page passes when all three stay within their limits for 75 per cent of visits.

On 13 November we scored 100 in every category on desktop. On mobile we scored 98 for performance and 100 for the rest. There is no field data for us yet.

Two things to watch for:

  • A new site has no field data. When a page has too little of it, PageSpeed Insights shows data for the whole domain. When the domain has too little as well, it shows nothing. That is our case. Field data takes weeks to build up, and on a site with little traffic it may never appear.
  • One run is not enough. The Lighthouse documentation (Lighthouse is the tool that calculates the score) says the median of five runs is twice as stable as a single run. In one of our series of measurements, one run out of three gave an LCP of 1.8 s while the median was 1.5 s. Compare medians of several runs, always on the same type of device.

Security headers: Mozilla HTTP Observatory

Observatory checks the headers your server sends with each page. Its scoring starts at 100 and deducts points for each failed test. Bonus points for extra protection only count if the score is still 90 or more after the penalties. An A+ starts at 100 points, and the highest possible score is 145.

Our weak spot was the script policy, the Content Security Policy (CSP). It lists what the browser may run on a page. Ours allowed any script written into the page’s code ('unsafe-inline'). That weakens the protection against injected scripts, so the report deducted 20 points. We were left with 80, a B+. Bonus points for the other headers we already had did not count, because we were below 90.

After the fix, each page carries its own list of allowed scripts, identified by their SHA-256 fingerprints. A script that is not on the list does not run. The site now passes all 12 tests, the bonuses count and we have an A+ with 120 points. The report for our site is public.

Not every site can reach an A+ without larger changes. What matters more than the letter is which tests fail and which ones cost the most points.

Certificate and encryption: SSL Labs

SSL Labs checks the certificate, protocols and encryption of the connection to your server. Our site has an A+ on both of the server’s addresses, IPv4 and IPv6. Before you run the test, tick "Do not show the results on the boards", so that your result does not appear on the public lists on the test’s site.

According to the SSL Labs rating guide, an A+ needs a good configuration with no warnings and an HSTS header valid for at least six months. HSTS tells browsers to reach your site over HTTPS only. It has a side effect worth knowing. Under the HSTS standard, when the certificate has an error the browser does not let the visitor continue past the warning. If the certificate expires, people who have visited the site before cannot open it. Turn HSTS on only when the whole site works over HTTPS, and keep an eye on certificate renewal.

Domain and email: internet.nl

internet.nl is an initiative of the internet community and the Dutch government. It has a website test and an email test. The website test checks IPv6, the domain signature (DNSSEC), HTTPS and security headers, among other things. The email test checks SPF, DKIM and DMARC, which protect against others sending mail as you, and whether mail travels encrypted.

Some of the email checks depend on your mail provider’s servers rather than on you. Before you start fixing, find out which settings you can change in your domain panel and which ones belong to the provider.

Structured data: Google’s test and the schema.org validator

Structured data describes a page in a way search engines understand, for example a business, an article or a product. Google recommends starting with the Rich Results Test, which shows the rich results Google can build from the page. The schema.org validator checks every type of structured data, including the ones Google does not use.

A clean test result does not guarantee a rich result in Google. Test your home page, one article and one service page, because each carries different data.

Accessibility: WAVE

WebAIM’s WAVE marks accessibility problems on the page, such as low contrast, an image without a description or a form field without a label. WebAIM points out that no errors does not mean the page is accessible. Some things only a person can judge, such as whether an image description makes sense.

Test several kinds of pages, not just the home page. For an online shop, accessibility may be a legal duty. We cover this in our article on the Polish Accessibility Act.

Search Console: how Google sees your site

Search Console shows which of your addresses Google has indexed, where redirects lead and how the site performs for users. Its Core Web Vitals report uses the same Chrome user data as PageSpeed Insights. It groups similar addresses, and each group takes the rating of its worst metric. On a new site or one with little traffic, the report stays empty.

Our Search Console data is read by an AI agent through a service account. We describe the setup in our article on an AI agent in Google Cloud.

What to fix first

This is the order we follow on our own site and on clients’ sites:

  1. The certificate and HTTPS. Any error in SSL Labs, a certificate close to expiry or a page that loads without HTTPS.
  2. The headers that cost the most points in Observatory. For us it was the script policy. Also check that HSTS lasts at least six months.
  3. Core Web Vitals outside their limits. Look at Search Console first, because it has the data from real users. Then use the PageSpeed Insights lab run to find the cause.
  4. Errors in WAVE. Contrast, image descriptions and form labels.
  5. Structured data and email settings, the findings of Google’s test and internet.nl.

How often to test

  • After launching a new site, and again the next day.
  • Before and right after any change to headers, the certificate, redirects or scripts.
  • Once a month if nothing changes.

Keep the results with their dates. Without a record from before a change, you cannot tell whether it helped.

If your site scored below an A in any of these tests, or below 90 in PageSpeed Insights, send us its address. We will look at what causes it and what to fix first.

This article was written with the help of AI. We also used AI to compare its facts with the listed sources.

Sources

Tell us about your project

A few sentences are enough. We usually reply within 24–48 hours.

Contact us