Single-Run Lab Audits of 96 Local Business Sites: What the Numbers Actually Say

In March 2026, a laboratory audit was conducted on 96 websites of local Ottawa businesses, including plumbers, electricians, HVAC companies, cleaners, and contractors. The audit was repeated on November 3, 2026, and the findings were compared. A single PageSpeed Insights API v5 lab run was performed on each site using mobile and desktop emulation with standard Lighthouse settings. The results were published anonymously, focusing only on the aggregated data and not revealing business names or URLs.

The median mobile PageSpeed score across the 96 sites was 65.5 out of 100. Half of the sites scored below this median, while the other half scored above it. The distribution of scores revealed that 10 sites (10.4%) scored between 90-100, 67 sites (69.8%) scored between 50-89, and 19 sites (19.8%) scored below 50. Most sites fell within the "needs improvement" band, with 10 scoring in the "good" band and 19 in the "poor" band.

The study identified several recurring issues that affected most sites. Unused JavaScript and CSS code were found on 68.8% of the sites, with an estimated time cost of around 460ms for JavaScript and 440ms for CSS per flagged site. Redirect chains were also prevalent, affecting 49% of the sites, with an estimated time cost of about 780ms per flagged site. However, it is important to note that these flags indicate potential issues, not confirmed performance waste.

Another significant finding was the poor performance in Largest Contentful Paint (LCP). For 72 of the 96 sites, the LCP took longer than 4 seconds, which is considered poor according to Google's guidelines. The median LCP time was 6.3 seconds, while the good threshold is 2.5 seconds and needs improvement runs to 4.0 seconds.

However, when the audit was re-run on November 3, 2026, the overall mobile score improved to a median of 71.0, an increase from 65.5 in March. Among the 88 sites that produced measurable scores in both editions, the share of sites scoring above 90 increased from 9.1% to 26.1%, while the share of sites scoring below 50 decreased from 20.5% to 15.9%.

Despite these improvements, it is crucial to recognize the limitations of this study, as it only provides a single-run lab measurement for each site on a specific day, and run-to-run variance is normal for PageSpeed.