We did it! We published the 2025 Web Almanac!
The Web Almanac analyses over 17 million websites and processes 244 TB of open-source data. Over 60 contributors have worked together over the past months to understand how the web evolves. The almanac this year has 15 chapters.
I co-wrote and edited the chapter on accessibility and I'd like to highlight some stuff for you, in case you don't have time to read the entire report. Although I'd encourage you to do so! It's a sobering and useful read for anyone responsible for digital products.
This year's chapter shows we created a web that is slowly getting better on the metrics we can easily track. But it's still excluding huge numbers of users with disabilities in the places that both matter most and are easy to fix. Colour, content and structure.
For product owners, it's clear we need to focus less on learning new techniques and more on deciding what we are willing to ship.
Overall, scores are up. The median Lighthouse accessibility passed 85, yet the same four issues keep showing up everywhere. Poor colour contrast, unclear link naming, broken heading hierarchies and missing or meaningless alt text on images take the lead.
We have legal pressure in full force. The European Accessibility Act and the updated ADA rules give accessibility "real teeth." In theory. Because the data suggests that regulation alone does not automatically translate into inclusive experiences.
On the testing front, as I've been saying, automation is necessary, but insufficient. Automated tools such as Lighthouse and axe-core are good, but they detect less than half of real-world issues and can only partially cover WCAG success criteria. They should be treated as guardrails and monitoring, not your definition of "done."
This year, we also dedicated an entire section in the accessibility chapter to Artificial Intelligence (AI). AI can help teams move faster, but these Large Language Models (LLMs) are trained on the same inaccessible web they are meant to repair.
AI should support human expertise and inclusive design, not replace them.
That is a good north star for product decisions.
Overlays still exist. About 2% of desktop sites now use accessibility overlays, but only 0.2% of the top 1,000 sites do so. These "magical fixes" are just a distraction. A risk, not strategy. They can help catch some low-hanging fruit, but they do not satisfy legal standards when the underlying site is inaccessible. And they certainly don't guarantee either accessibility or avoiding any legal troubles.
As I was reviewing the data and writing the piece, it dawned on me. The tooling is there. It's good enough. The legal pressure is real. So the remaining gaps are mostly about governance and the day‑to‑day decisions about what to prioritise.
As an author of the chapter, I couldn't really give me honest opinion on some of the findings. I needed to remain as objective as possible and not interpret things too much.
Alas, this is my newsletter and I'll write whatever I want!
In the emails that follow, I'll go deeper and not hold back. I'll give me 2 cents on where progress is real, where it's stalled and what this means for your backlog, your governance, budget and your conversations with stakeholders.
Stay tuned!
And in the mean time, read the damn thing, will you!?