Back to Top Start a Trial

Small-business homepage lab performance

Lab PageSpeed Insights on 12,536 unique homepages: median LCP 9.21s, median score 57, median page weight 3.05 MB. Untreated baselines, not product before-and-after.

By Alex MacArthur, builder of PicPerf · Published

Headline (2026-09-05): across 12,536 unique public homepages measured with PageSpeed Insights, median lab Largest Contentful Paint was 9.21 seconds. Median mobile Performance score was 57. Median total page weight was 3.05 MB. Google’s “good” LCP threshold is 2.5 seconds. These pages were measured as they exist on the origin — not after PicPerf optimization.

Median LCP
9.21s
3.7× Google’s 2.5s “good” line
Median score
57
45 p25 · 67 p75
Median page weight
3.05 MB
12,536 unique homepages
Median LCP by industry
Slowest at the top. Longer bars are worse. The dashed line is a “good” lab LCP (2.5s).
Median Performance score by industry
Same order as LCP (slowest LCP at the top). Longer bars are better.

Lab baselines at a glance

One row per unique URL, latest analysis kept. Completed and partial lab runs with a before Performance score are included. PageSpeed Insights lab data (mobile, worst of up to five runs).

Metric n p25 Median p75
Performance score 12,536 45 57 67
LCP (seconds) 12,536 5.54 9.21 15.03
Page weight (MB) 12,536 1.68 3.05 5.71

Performance score buckets

Score URLs Share
0–49 4,136 33.0%
50–89 7,817 62.4%
90–100 583 4.7%

Outreach sample vs. requested analyses

Most URLs come from PicPerf's outreach pipeline (local businesses selected for analysis). A smaller set comes from free Turbulence analysis form submissions.

Source n Median score Median LCP Median page weight
Outreach (promo) 11,435 57 9.15s 3 MB
Analysis form 1,101 53 9.8s 3.82 MB

Restaurants and hotels are slowest

Industry is set on outreach leads only (6,236 of 11,435 promo URLs). Rows below have at least 50 labeled URLs. Restaurants and hotels are slower on LCP than dentists, lawyers, and clinics in this sample.

Industry n Median score Median LCP Median page weight
Dentist 2,123 62 7.13s 2.17 MB
Restaurant 1,503 56 11.66s 3.65 MB
Lawyer / Attorney 645 61 7.43s 2.05 MB
Clinic / Doctor 499 61 6.55s 2.03 MB
Advertising Agency 456 67 5.84s 2.65 MB
Car Dealer 446 54 10.76s 3.74 MB
Real Estate / Realtor 399 55 10.35s 3.47 MB
Hair Salon 76 59 9.68s 3.18 MB
Hotel 52 48 13.51s 4.93 MB

Methodology

  1. URLs are public homepages that went through PicPerf’s Turbulence analysis pipeline (outreach and the free form at picperf.io/turbulence).
  2. PageSpeed Insights scores the origin URL on mobile, up to five runs. The stored before report keeps the worst run.
  3. This page uses only that before report: Performance score, LCP (Lighthouse largest-contentful-paint numeric value), and total byte weight. After scores and product transformations stay out.
  4. Duplicate URLs keep the latest analysis. Partial runs (before metrics only) are included when a score exists.

Machine-readable tables: JSON and CSV. Image payload for a separate homepage sample lives here: small-business homepage image weight. Paired before-and-after product results live here: Turbulence performance research.

What this does not claim

  • Not the whole web. This is a convenience sample of sites PicPerf analyzed, weighted toward local businesses (dentists, restaurants, law firms). It is complementary to HTTP Archive, not a replacement.
  • Not field data. Lab LCP is not CrUX LCP. Worst-of-runs also makes the distribution slower than a median-of-runs recipe would.
  • Not a product result. Nothing on this page is “after Turbulence.” Do not cite these medians as optimization gains.
  • Not HTML size. Document transfer size was not reliably populated in the stored PSI payloads, so it is omitted.

Frequently Asked Questions

  • In PageSpeed Insights lab runs on 12,536 unique homepages (snapshot 2026-09-05), median Largest Contentful Paint was 9.21 seconds. The 25th percentile was 5.54s and the 75th percentile was 15.03s. Google treats LCP under 2.5s as “good.” This is lab data, not Chrome UX Report field data.

  • The same 12,536-URL sample had a median mobile Performance score of 57 (p25 45, p75 67) and a median page weight of 3.05 MB. 4,136 URLs scored below 50; 583 scored 90 or above.

  • In this sample, hotels (median LCP 13.51s) and restaurants (11.66s) were slower than dentists (7.13s), lawyers (7.43s), and clinics (6.55s). Industry labels exist only on outreach-pipeline rows, not on every URL.

  • No. Every figure is PageSpeed Insights (Lighthouse) lab data on mobile. Before metrics pick the worst of up to five runs. It is useful for comparing sites under the same lab recipe. It is not a substitute for CrUX or RUM.

Related Guides

Turbulence rewrites every HTML page at the edge. No code changes.

Start a 14-day Turbulence Trial