Viewing the 2025 Web Almanac, the Current State of Accessibility and Web Performance
Kazuhito KidachiExecutive FellowWhen I see the word “Almanac,” the first thing that personally comes to mind is the fictional sports yearbook “Grays Sports Almanac” that appears in the movie Back to the Future Part II. Recently, however, the 2025 edition of the Web Almanac was released.
The Web Almanac is an annual report published by HTTP Archive, a project with the mission of tracking how the web is built. Since the first edition in 2019, it has been published almost every year (with the exception of the 2023 edition).
One of its most notable features is the sheer scale of its dataset. The 2025 Web Almanac is based on data collected in July 2025 and includes a dataset of more than 16 million websites, making it ideal for understanding the current state of the web.
Reflecting the size of its dataset, the Web Almanac itself also contains a vast amount of information. The report is organized into four parts, consisting of 16 chapters in total, written and contributed by 72 experts.
At this point, Japanese translations are not yet available for most sections, so I personally have not finished reading every chapter. In this Column, I will share my impressions of the chapters on accessibility and display performance, two aspects of web quality that our company has been focusing on.
Accessibility
Last year, 2025 marked the year when the European Accessibility Act was fully enforced (in Japanese). Some people may therefore assume that accessibility quality across the web must have improved significantly.
However, to state the conclusion first, the Web Almanac reports that—at least within the scope of its analysis—the situation has not changed dramatically compared with 2024.
Some indicators do show gradual improvement. For example, the median accessibility score in Lighthouse and the percentage of websites that properly consider color contrast have both improved slightly. In that sense, the situation may not be entirely pessimistic.
Nevertheless, the report also notes that many pages still fail to follow very basic rules. Examples include pages that do not specify the natural language used on the page via the HTML lang attribute, as well as pages that intentionally hide visible focus indicators.
Another point of concern is the steadily increasing percentage of pages adopting widgets intended to improve accessibility—tools often referred to as accessibility overlays. (Our company does not recommend implementing accessibility overlays. (in Japanese)).
For these reasons, it is difficult to say that accessibility across the web is clearly improving. To break through this situation, I strongly feel that we need better education and awareness initiatives, as well as further advancement of AI-powered accessibility checking tools.
Incidentally, we are also planning to cover the Accessibility chapter in more detail on our Accessibility Blog (in Japanese) at a later date.
Display Performance
Regarding display performance, what particularly stood out to me was the analysis based on Core Web Vitals, a set of metrics proposed by Google that can now be measured consistently across web browsers, not only in Chrome.
It appears that the percentage of websites that perform well across all three Core Web Vitals metrics—LCP, INP, and CLS—has been steadily increasing year by year, albeit gradually.
This may seem somewhat surprising when considered alongside a finding from the chapter on page weight, which notes that the amount of data required to display pages continues to increase steadily.
If richer content and larger file sizes have been accompanied by the adoption of techniques that ensure faster and more stable page display, that would certainly be a positive development.
However, some findings appear to contradict this assumption. For example, the image that occupies the largest portion of the viewport, the so-called LCP image, should generally not be lazy loaded. Yet according to the report, such an implementation was found on 17% of mobile pages.
Additionally, 62% of mobile pages had at least one image for which the width or height was not explicitly specified. It has long been considered best practice to specify the width and height attributes on img elements in order to prevent layout instability, so this figure was somewhat surprising to me.
Overall, while display performance seems to be improving gradually, my impression is that correct implementation of the techniques needed to achieve those improvements is still very much required.
For more information on our services, timeframes and estimates, as well as examples of our work, please feel free to be in touch.