Blog
Ideas, guides, and reflections on building websites that are faster, more accessible, and more trustworthy, for the people who visit them every day.
We write about web performance, accessibility, SEO, privacy, open source tools, and the principle that people should always come before technology. If a word feels unfamiliar, we'll explain it. No jargon gatekeeping here.
Test results in your codebase, working with the Webperf MCP server
Run the Webperf MCP server inside a coding assistant that has your repository open, and each finding can be traced to a file and a line. How to sort the fixes by where they live (code, server, DNS, third party), check the effect against the open-source rating code before deploying, re-test only what changed, and remember that automated tests are not infallible.
Group access: your whole agency, without the invitation dance
Agency staff now reach the whole client portfolio through their work email domain, instead of being invited to one client site at a time. What an agency administrator controls, why every client can open Users and see which agency is in their account, and the two boundaries that otherwise turn into support tickets: test credits belong to the website, and access is not support. Built because a customer described the onboarding as absurd rather than putting up with it.
Accessibility laws worldwide, and where to start your audit
A world tour of digital accessibility law, from the European Accessibility Act to the ADA, the UK, Canada and Australia, and the one standard they all point at. Then a practical way to begin an audit: automated testing for whole-site coverage, a hands-on guide to the manual testing anyone can do with a keyboard and a screen reader, and WCAG-EM 1.0 and 2.0 as the method that makes the result hold up.
The HIPPO in the room: the Highest Paid Person's Opinion
The Highest Paid Person's Opinion overrules the evidence in most meetings, not because it's right, but because it outranks everyone else. What the HIPPO costs in self-interest, vested interest and history, and why a neutral audit is a relief for the whole team, the highest-paid person most of all.
What a website audit finds, 2,000 pages of VisitDenmark
A worked example. We crawled 2,000 pages of a real, well-built site and read the results the way you'd read your own. The lab-versus-field gap that changes what the scores mean, the one accessibility fix that clears every page at once, and why a scary 37,000-error count collapses into a short list of root causes.
The cookies you set before anyone said yes
A cookie banner asks a question, but the answer only counts if nothing was already written to the visitor's device before they clicked. What "set before consent" really means, how an automated test can see it, and why the fix is usually less tracking rather than more.
Introducing the Webperf MCP server for your own AI
We said you should bring your own AI. Here's the door. A small, read-only server that lets Claude and other assistants answer questions about your sites, running on your machine, with your key in your keychain and every permission decision still made by the API.
Saying the inconvenient truth of a website audit out loud
Everyone already knows the checkout is slow; nobody wants to be the one who says it. Why an external audit can name what a colleague can't, why being free to wave away most of the findings is a feature rather than a flaw, and how that clears the way to fix the one thing that matters.
The discovery half of a privacy health check, automated
A GDPR health check is a lawyer and a technical specialist reading your site, and a good part of that week goes on discovery rather than legal reasoning. A crawler can keep that part current, which is why specialists run these checks themselves. What the audit can hand them before they start, and what only they can do.
Numbers tell you where, but people tell you why
Quantitative metrics show you where the problem is and how many it touches; qualitative methods show you why it happens and what to do. A tour of the research methods at both extremes, and why the strongest teams run them as a loop rather than picking a side.
Making the case for website quality to your leadership
Scores are how you find the problem; risk, revenue and cost are how it gets funded. How to translate audit findings for a leadership meeting. Bring a whole-site baseline, a competitor comparison, and a bounded ask with proof promised up front.
Being data-informed, not data-driven
"Data-driven" taken literally hands the steering wheel to whatever happens to be measurable. The better ambition is evidence on the table every time, with context and judgment in the chair. The difference in practice, with the habits that keep metrics honest.
Did the fix actually work? Measuring before and after
Optimisation without honest measurement is hoping with extra steps. Compare like with like, treat single runs as anecdotes, respect the 28-day lag in field data, and write down what you expect before you look.
Your first performance budget
Websites get slow the way a junk drawer fills up, one defensible addition at a time. A performance budget is the defence, with limits agreed in advance, anchored to your own baseline, and enforced automatically instead of by memory.
From audit to action, turning findings into a backlog
An audit that gets filed away changed nothing. Count pages before problems, sort by impact times effort, write tickets a developer can start, and close the loop with a re-test, until the audits get boring. That is the goal.
The hidden cost of third-party scripts
Every third-party script is code you don't control, running on your visitors' devices, on your reputation, billing you in speed, privacy exposure and resilience. How to inventory yours, ask three honest questions per entry, and keep the list from growing back.
Core Web Vitals in plain language
LCP, INP and CLS measure three things every visitor feels. The page shows up, it responds, it holds still. What each one means, the thresholds that matter, why the 75th percentile is what's being graded, and what to do about them.
Accessibility is a habit, not a checkbox (and now the law)
Accessibility helps far more people than most teams assume, and with the European Accessibility Act in force since June 2025 it's a legal duty for much of the private sector too. What automated testing can and can't catch, and why continuous beats one-off here more than anywhere.
Why Webperf Cloud and PageSpeed Insights scores don't match
Both tools run Lighthouse, so people expect identical scores. They won't match. Different servers, throttling, hardware and cold caches move the number, and a 10–15 point gap is normal. What actually differs, and why lab and field data answer two different questions.
Introducing Webperf Audit, a full-site website audit
A homepage score is a number; a whole-site audit is something you can decide from. Webperf Audit runs Lighthouse, Core Web Vitals, accessibility, SEO and a cookie & third-party check across every page, then points at the exact pages and elements behind each finding. And you keep all the raw data.
Why an open source foundation beats the closed box
Most website-quality tools ask you to trust a score you can't inspect and data you can't take with you. Because Webperf is built on an open source engine you can read, run, and leave with, the control stays with the organisation, not the supplier.
What's in the baseline report?
The baseline report is your one-time, whole-site snapshot. It contains Lighthouse scores, six months of real Chrome-user data, cookies set without consent, and the third parties reaching your pages.
Who actually needs the Standard tier?
Starter is the right place to begin for anyone curious. Standard is for teams with a real UAT process, performance budgets, and a need to be told the moment a score slips. How to tell which one is you.
Who is the Agency tier for?
Agency is built for teams looking after a whole portfolio. One website on Standard plus up to 20 on Starter, managed from one place, with portfolio-wide API access and Starter sites that get agency benefits.
Take back control of your web data
Owning your web data means more than access. It means the freedom to leave for the open source engine, to read everything through the API, and to draw your own conclusions instead of trusting a vendor's black-box "insights".
What is digital sovereignty?
You've probably heard the phrase. But what does "digital sovereignty" actually mean in practice, and why should it shape the tools you choose?
People before auditing tools
It's tempting to start with technology. But the best web teams start with the humans who use the site. A reflection on putting people at the centre of quality.
Why website quality matters
Speed, accessibility, SEO and privacy decide whether visitors stay and convert. Why continuous quality beats the occasional one-off check.
Introducing Webperf Cloud, open source website auditing
"Trust, but verify" is old advice, but it rarely matters more than when you choose the tools you plug into your website. Meet Webperf Cloud, the open-source auditing service that succeeds Webperf Premium, and try the engine yourself.
More posts coming
We're just getting started. If there's a topic you'd like us to cover, drop us a line at hello@webperf.cloud.