Four months ago, my mobile Performance score sat at a stubborn 58. Today it reads 97. Yet the most useful lessons on the report lie in the numbers that did not move.
In June, I wrote about a frustrating paradox on this blog. Google PageSpeed Insights gave Rise&Inspire green marks for Accessibility, Best Practices and SEO, but only 58 for mobile Performance.
Recently, I ran the same test again. This is what it reported:
| Category | Mobile | Desktop |
| Performance | 97 | 99 |
| Accessibility | 96 | 96 |
| Best Practices | 100 | 100 |
| SEO | 92 | 92 |
| Agentic Browsing | 2/2 | 2/2 |
A climb from 58 to 97 is pleasing. But I have come to think that the 97 is the least instructive number on this report. Here is why.
1. What Changed Between June and Now?
Only one number moved. Mobile Performance rose by 39 points. Accessibility at 96, Best Practices at 100 and SEO at 92 stand exactly where they stood in June.
I cannot credit the rise to any single change. A laboratory score responds to many things at once: plugins, images, theme updates and quiet improvements by the host. What the comparison does show is that a poor score is a starting point, not a verdict.
2. What Does the Performance Score Actually Measure?
The Performance score comes from a laboratory test, not from real visitors. On mobile, PageSpeed Insights simulates a mid-range phone on a throttled 4G connection, which is why mobile scores usually trail desktop ones.
The score blends five lab measurements: First Contentful Paint, Speed Index, Largest Contentful Paint (LCP), Total Blocking Time (TBT) and Cumulative Layout Shift (CLS).
One distinction is easy to miss. Interaction to Next Paint (INP), Google’s measure of how quickly a page responds to a tap or click, cannot be captured in a lab, because no one is tapping. Lighthouse uses Total Blocking Time as its stand-in. Real INP is collected from actual visitors and shown in the field-data section of PageSpeed Insights and in Search Console, once a site has enough traffic.
Google’s targets for a good experience are simple to remember: LCP within 2.5 seconds, INP within 200 milliseconds and CLS no higher than 0.1.
3. Why Are the Unchanged Numbers the Real To-Do List?
A score that has not moved in four months is very likely pointing at the same few failed checks. Accessibility at 96 and SEO at 92 mean that specific audits are still failing on this page, and PageSpeed Insights names each one. Expand the category, and the failures are listed.
Accessibility covers matters such as alternative text, colour contrast, heading order and descriptive link names; SEO covers the page title, the meta description and crawlable links. Neither score is a certificate. Automated tools catch only part of what can block a reader, and an SEO score says nothing about whether the writing deserves to rank.
Best Practices at 100 confirms a secure connection and modern web standards on the tested page. It does not make a site immune to attack. Updates, strong passwords and backups still matter, as I set out in my guide to blog hygiene.
4. What Is the New Agentic Browsing Badge?
The small 2/2 beside the four circles is new this year. In May 2026, Google added an Agentic Browsing category to Lighthouse, the engine behind PageSpeed Insights. It asks whether an AI agent, software that browses and acts on a person’s behalf, could read and use a page reliably.
Unlike the other categories, it gives no score out of 100, only a count of checks passed, and Google describes it as experimental. Its checks look at the accessibility tree (the structured map of headings, links and buttons that assistive technology reads), layout stability, an optional llms.txt summary file and WebMCP, a proposed way for sites to offer tools to agents. Checks that do not apply are left out of the count, so my 2/2 means both counted checks passed.
Google has not made it a ranking factor, so there is no reason to chase it. What struck me is something else: agents rely on the same accessibility tree that screen readers use. Every descriptive link and well-ordered heading written for a reader who cannot see the screen now serves a newer kind of visitor as well.
5. Why Is a Fast Page a Form of Hospitality?
Many traditions treat the welcome of a guest as a moral matter, and they agree on one detail: the host prepares before the guest arrives.
- The Taittiriya Upanishad, an ancient Sanskrit teaching text, closes the student’s training with the instruction atithi devo bhava: let the guest be as a god to you.
- In the Book of Genesis, Abraham sees three strangers in the heat of the day, runs from his tent door to meet them, and hurries to have bread made ready (Genesis 18:1–6).
- Japanese omotenashi, the spirit of hospitality associated with the tea ceremony, prizes meeting a guest’s needs before they are spoken.
Performance work is that kind of hospitality. The reader never sees the compressed image or the deferred script. They notice only that nothing kept them waiting at the door.
6. Should I Chase a Perfect 100?
The case for chasing it deserves a fair hearing. For a large online shop, a fraction of a second can affect sales, and its engineers are right to fight for every millisecond. For a reading blog already at 97, the last three points would cost hours better spent on the next post.
So my next steps are modest:
- Open the failed Accessibility and SEO audits and fix what they name.
- Test other kinds of pages: an image-heavy post, a category page and the archive.
- Watch the Core Web Vitals report in Search Console, which reflects real readers over time.
- Run the test again after every new plugin or theme change.
The broader speed fixes, from plugin clean-up to image compression, are set out in the June post.
Conclusion: What Does the Report Really Reveal?
The 97 tells me the door opens quickly. The 96 and the 92 tell me where the hinges still need oil. The 2/2 tells me that a new kind of visitor has begun to knock. The real test, though, never appears on the report: whether a reader arrives, stays to the last line, and comes back tomorrow.
When you run your own blog through PageSpeed Insights, do you stop at the scores, or do you read the audits beneath them?
John Britto Kurusumuthu
Founder, Rise & Inspire
Explore more at the Rise & Inspire archive | Tech Insights
© 2026 Rise & Inspire.
