Dallas Small Business Website Speed Benchmark: 36 Sites Tested on Mobile
Dallas, TX • Serving all of Dallas–Fort Worth
We ran identical mobile lab tests against the homepages of 36 Dallas-area small business websites across five industries in August 2026. The median Largest Contentful Paint was 3.3 seconds. Only 11 of 36 sites — 31% — landed inside Google's 2.5-second "good" threshold. 16 sites, 44%, were slower than 4 seconds, which Google classifies as poor.
Weight was not the whole story. The median page transferred 1.21 MB over 70 requests, which is unremarkable. The failures were concentrated in a smaller group of sites carrying uncompressed hero media — the heaviest single homepage in the sample transferred 22.8 MB — and in sites with slow server response, where 10 of 36 took longer than 800 ms just to start sending HTML.
Results are reported as anonymized aggregates by industry. We are not naming the businesses tested; several are the kind of local company we would rather help than embarrass.
- Sites tested
- 36 Dallas-area homepages
- Median mobile LCP
- 3.3 seconds
- Inside Google's "good" threshold
- 11 of 36 (31%)
- Median page weight
- 1.21 MB over 70 requests
Disclosure: GH Consulting builds websites in Dallas, so a study showing that many Dallas websites are slow is convenient for us. Two things keep it honest: the businesses tested are anonymized, so this is not a lead list dressed as research, and the methodology above is specific enough for anyone to reproduce and challenge our numbers.
Headline results
Largest Contentful Paint (LCP) measures when the main content of a page becomes visible. Google's published thresholds are 2.5 seconds or faster for "good" and slower than 4 seconds for "poor".
| Metric | Result across 36 Dallas sites |
|---|---|
| Median LCP (mobile, throttled) | 3.3 seconds |
| 25th percentile LCP | 2.1 seconds |
| 75th percentile LCP | 5.3 seconds |
| Slowest site in sample | 18.2 seconds |
| Good (LCP ≤ 2.5s) | 11 sites — 31% |
| Needs improvement (2.5s–4s) | 9 sites — 25% |
| Poor (LCP > 4s) | 16 sites — 44% |
| Median first contentful paint | 1.5 seconds |
| Median full page load | 6.0 seconds |
| Median time to first byte | 604 ms |
| Sites with TTFB over 800 ms | 10 of 36 |
| Median transfer size | 1.21 MB |
| Sites over 3 MB | 6 of 36 |
| Median requests | 70 |
| Sites over 100 requests | 8 of 36 |
| Sites with a mobile viewport meta tag | 36 of 36 |
Results by industry
Roofing and exterior contractors were the fastest group tested and the only industry where a majority of sites passed. Dental and med spa sites were the slowest, driven by heavy imagery and third-party booking, chat and review widgets.
| Industry | Sites | Median LCP | Median weight | Median requests | Passing LCP |
|---|---|---|---|---|---|
| Roofing & exteriors | 8 | 2.3s | 1.27 MB | 53 | 5 of 8 |
| Legal | 7 | 3.2s | 1.14 MB | 56 | 2 of 7 |
| HVAC & plumbing | 6 | 3.8s | 1.47 MB | 112 | 1 of 6 |
| Landscaping & lawn | 8 | 4.5s | 0.80 MB | 59 | 2 of 8 |
| Dental & med spa | 7 | 5.2s | 1.39 MB | 93 | 1 of 7 |
Every site in the sample, anonymized
Sites are ordered fastest to slowest inside each industry and labeled by letter. The pattern worth noticing: weight and speed are only loosely correlated. Site A in landscaping transferred almost nothing and still took 1.9 seconds; the fastest roofing site rendered in 0.6 seconds with 0.9 MB; and one 22.8 MB homepage still hit 3.1 seconds because its hero image loaded before the rest of the payload.
| Site | LCP | TTFB | Transferred | Requests | Verdict |
|---|---|---|---|---|---|
| Roofing & exteriors — Site A | 0.6s | 0.06s | 0.9 MB | 20 | Pass |
| Roofing & exteriors — Site B | 1.5s | 0.90s | 0.3 MB | 22 | Pass |
| Roofing & exteriors — Site C | 2.1s | 0.70s | 0.5 MB | 36 | Pass |
| Roofing & exteriors — Site D | 2.3s | 0.53s | 3.9 MB | 80 | Pass |
| Roofing & exteriors — Site E | 2.3s | 0.06s | 0.8 MB | 99 | Pass |
| Roofing & exteriors — Site F | 3.1s | 0.92s | 22.8 MB | 137 | Needs work |
| Roofing & exteriors — Site G | 5.5s | 0.67s | 1.7 MB | 71 | Fail |
| Roofing & exteriors — Site H | 18.2s | 0.35s | 17.5 MB | 18 | Fail |
| Legal — Site A | 1.2s | 0.32s | 2.4 MB | 95 | Pass |
| Legal — Site B | 1.6s | 0.07s | 1.1 MB | 141 | Pass |
| Legal — Site C | 3.1s | 0.59s | 0.5 MB | 56 | Needs work |
| Legal — Site D | 3.2s | 2.02s | 1.1 MB | 69 | Needs work |
| Legal — Site E | 3.3s | 0.69s | 0.6 MB | 50 | Needs work |
| Legal — Site F | 4.5s | 1.16s | 1.4 MB | 53 | Fail |
| Legal — Site G | 5.3s | 1.68s | 2.6 MB | 38 | Fail |
| HVAC & plumbing — Site A | 1.4s | 0.32s | 1.4 MB | 140 | Pass |
| HVAC & plumbing — Site B | 3.0s | 0.40s | 1.3 MB | 112 | Needs work |
| HVAC & plumbing — Site C | 3.5s | 1.31s | 1.5 MB | 62 | Needs work |
| HVAC & plumbing — Site D | 4.1s | 0.55s | 0.5 MB | 37 | Fail |
| HVAC & plumbing — Site E | 4.4s | 0.56s | 4.2 MB | 116 | Fail |
| HVAC & plumbing — Site F | 12.4s | 2.49s | 7.7 MB | 112 | Fail |
| Landscaping & lawn — Site A | 1.9s | 0.63s | 0.01 MB | 18 | Pass |
| Landscaping & lawn — Site B | 2.0s | 0.06s | 0.9 MB | 38 | Pass |
| Landscaping & lawn — Site C | 3.0s | 0.67s | 0.7 MB | 92 | Needs work |
| Landscaping & lawn — Site D | 4.2s | 0.62s | 0.6 MB | 40 | Fail |
| Landscaping & lawn — Site E | 4.8s | 3.20s | 1.0 MB | 132 | Fail |
| Landscaping & lawn — Site F | 4.9s | 0.87s | 2.4 MB | 87 | Fail |
| Landscaping & lawn — Site G | 5.6s | 0.06s | 2.3 MB | 65 | Fail |
| Landscaping & lawn — Site H | 8.8s | 0.17s | 0.2 MB | 53 | Fail |
| Dental & med spa — Site A | 2.0s | 0.07s | 0.4 MB | 94 | Pass |
| Dental & med spa — Site B | 2.9s | 0.66s | 1.0 MB | 35 | Needs work |
| Dental & med spa — Site C | 3.1s | 0.07s | 1.4 MB | 93 | Needs work |
| Dental & med spa — Site D | 5.2s | 0.62s | 2.6 MB | 88 | Fail |
| Dental & med spa — Site E | 7.0s | 0.38s | 2.8 MB | 100 | Fail |
| Dental & med spa — Site F | 7.2s | 0.81s | 0.1 MB | 251 | Fail |
| Dental & med spa — Site G | 11.3s | 0.14s | 8.8 MB | 16 | Fail |
What made the slow sites slow
- Unoptimized hero media. The two heaviest homepages in the sample transferred 22.8 MB and 17.5 MB. Both were serving full-resolution photography to a phone. Compressing and correctly sizing those files is usually a one-day fix with the largest single payoff.
- Slow server response. Ten sites took over 800 ms to first byte and one took 3.2 seconds. No amount of front-end work recovers a slow origin — this is a hosting and caching problem.
- Third-party script pile-up. The heaviest request count in the sample was 251 requests on a page that transferred only 0.1 MB, meaning almost all of it was scripts and beacons: chat widgets, review embeds, booking tools, tag managers and several ad pixels.
- Render-blocking front ends. Several sites had a fast first paint and a slow LCP, which means the browser painted a shell quickly and then waited on JavaScript or fonts to place the actual headline and hero.
- Sliders and carousels above the fold. Multiple large images competing to be the LCP element is a common template default and a reliable way to fail the metric.
Why this matters for Dallas search visibility
Core Web Vitals are a documented Google ranking signal, and LCP is the one small business sites fail most often. In a metro as competitive as Dallas–Fort Worth, where several capable competitors rank for the same service-plus-city query, page experience becomes a tiebreaker rather than a nice-to-have.
The commercial cost is larger than the ranking cost. A Dallas homeowner comparing three roofers on a phone is not waiting 5 seconds, and 44% of the sites we tested took longer than 4 seconds to show their main content. That traffic is already paid for — through ads, referrals or hard-won rankings — and it leaves before the page appears.
- If your homepage transfers more than about 2 MB on mobile, media compression is the first place to look.
- If time to first byte exceeds 800 ms, the problem is hosting or caching, not design.
- If request count is over 100, audit third-party scripts before touching anything else.
- Re-test after every change, on a throttled mobile profile — not on desktop office wifi.
Benchmark study questions
How were the 36 sites chosen?
+
We searched for Dallas-area businesses in six industries, took the independent local companies that surfaced, and dropped national chains and franchise location pages. Nine of the 45 sites originally tested were excluded from the results — seven returned no measurable payload to an automated browser, and one industry, auto and collision, was left out of the reporting because only two of its sites remained measurable, which is too small to report.
Is this Google field data or lab data?
+
Lab data. Every page was loaded once in headless Chromium at a 390×844 mobile viewport, with a 4× CPU throttle and a simulated mobile network, from a single cloud location. This is comparable across the sample because every site got identical conditions, but it is not the same as Google's Chrome User Experience Report field data from real users.
Why is one site listed at 0.01 MB?
+
That homepage genuinely served almost nothing — it is a minimal page with text and one small asset — and it still took 1.9 seconds to render its main content. Lightness alone does not guarantee speed.
Will you tell me which site is mine?
+
If you own a Dallas business and want to know how your site would score under the same test, send us the URL and we will run it and send you the numbers. We are not publishing the mapping of letters to businesses.
How often is this study updated?
+
We plan to re-run the same sample periodically so the trend is comparable. This edition reflects tests run in August 2026.
Can I cite this study?
+
Yes, and we would like you to. Quote any table with a link to this page. If you are a journalist or researcher and want the underlying per-site measurements, contact us.
Methodology
Tests were run in August 2026 against the homepage of each site. Each page was loaded once in headless Chromium at a 390×844 viewport with a mobile user agent, touch enabled, a 4× CPU throttle and simulated mobile network conditions with 150 ms of added latency. Metrics were read from the browser's own Performance API: Largest Contentful Paint from the largest-contentful-paint observer, time to first byte and load timings from the navigation entry, and transfer size and request count from resource timing.
Verdicts use Google's published Core Web Vitals thresholds for LCP: good at 2.5 seconds or faster, needs improvement between 2.5 and 4 seconds, poor above 4 seconds.
Stated limitations: single-run measurements, one geographic test location, lab conditions rather than real-user field data, and transfer sizes that can be understated where a resource does not report its size to the browser. Sites returning no measurable payload to an automated browser — typically bot protection or interstitials — were excluded rather than scored, because a blocked load is not a slow load.
- Sample: 45 Dallas-area business homepages tested, 36 reported.
- Industries reported: roofing and exteriors, legal, HVAC and plumbing, landscaping and lawn, dental and med spa.
- Excluded: national chains, franchise location pages, and any site with an unmeasurable load.
- Test date: August 16, 2026. Anonymized aggregates only.
Sources
- Google — Largest Contentful Paint (LCP) thresholds — Source of the 2.5-second good and 4-second poor thresholds used for verdicts.
- Google — Core Web Vitals — Definitions of the metrics measured and their role in page experience.
- Google Search Central — Understanding page experience in Search results — Google's own documentation on how page experience factors into ranking.
- Chrome — Performance API: LargestContentfulPaint — The browser API used to capture LCP in these tests.
Citing this page: This study may be quoted or reproduced with attribution and a link to https://ghconsultingllc.com/research/dallas-website-speed-benchmark. Suggested citation: GH Consulting, "Dallas Small Business Website Speed Benchmark," August 2026.