Field study, November 2026
The state of 50 New London County small business websites.
In November 2026 I measured 50 independent small business websites across New London County, Connecticut, for mobile speed, tap-to-call, contact forms, reviews and freshness, then scored each against Google’s E-E-A-T framework.
The typical site scores 73 of 100 on mobile, takes 6.7 seconds to show its main content, and scores 63 of 100 on E-E-A-T. Only 7 rate Strong. 8 of 49 have no tap-to-call link, 19 show no reviews, and just 15 of 50 published anything dated in the past year.
Measured November 6, 2026. Updated November 7, 2026.
7 numbers. The readout
73/ 100
Median mobile score
Lighthouse performance, 49 sites
6.7s
Median time to main content
Google's target is 2.5 s
84%
Have tap-to-call
8 of 49 do not
88%
Have a contact form
Typical form: 5 fields
61%
Show reviews on the home page
19 of 49 show none
30%
Published something in the past year
15 of 50
63/ 100
Median E-E-A-T score
7 of 50 rate Strong
Most sites are slow on a phone. Mobile speed
The typical site scores 73 on mobile and takes 6.7 seconds to show its main content on a mid-range phone over slow 4G. Google’s target is 2.5 seconds, and 4 of 49 meet it.
The problem is weight, not code. The median page is 2.7 MB and 22 of 49 are over 3 MB: desktop-sized photos, video backgrounds and plugins loading on every page.
Images sized and compressed for a phone, 1 font family instead of 3, and fewer scripts get a site under 3 seconds without losing a photo.
1 in 6 sites makes you dial by hand. Tap-to-call
On 8 of 49 home pages the phone number is not tappable: 6 show it as plain text to copy or retype, and 2 show no number at all.
For a contractor, a dentist or a restaurant, the call is the conversion, and most searchers are on a phone. A number that is text instead of a link quietly loses some of those calls. Every site I build has tap-to-call on every screen.
Most sites have a form. The typical one asks 5 questions. Contact form
43 of 49 sites have a contact form, typically 5 fields with 3 required. 16 ask 4 or fewer, 11 ask 8 or more, and 6 have no form at all.
Every field is a reason to stop. Usually 4 are enough: name, phone or email, what they need, and where. No form is fine if the phone number is a tap away, which on 2 of those 6 sites it is not.
2 in 5 home pages show no proof. Reviews on the page
30 of 49 home pages show reviews, testimonials or a rating, 11 of them pulled in live. 19 show nothing.
A Google review with a name on it does more than any paragraph about quality. Every dentist and chiropractor in the sample shows reviews, and 1 salon in 5 does. A live feed of your Google reviews is a small addition, and it updates itself.
Most sites look untouched. Freshness
Only 15 of 50 sites published anything dated in the past 12 months. Of the 30 with a copyright line, 10 still say 2024 or earlier.
A site does not have to blog. But a page that has not changed in 2 years tells visitors, search engines and AI answers that nobody is home.
A new photo, a new review, current hours and this year in the footer are enough, and they are what a monthly update covers.
Technically fine is not the same as trusted. E-E-A-T
The typical site scores 63 of 100 on Google’s E-E-A-T framework: 7 of 50 rate Strong, 31 Partial and 12 Weak. Trust scores highest and expertise lowest. Most sites prove they exist and can be reached; few show who does the work or what qualifies them.
E-E-A-T is Google’s shorthand for what its quality raters look for: experience, expertise, authoritativeness and trust. Google publishes no score, so I built a rubric a script applies the same way to every site: 20 signals, 5 per component, each scored 0, 3 or 5 from the home, About and contact pages. The full list is in the method below.
Sites do well on the mechanical signals, like HTTPS, hours and a phone number, and poorly on the human ones. 28 About pages name nobody. 10 sites show a licence number. 2 tell the owner’s story, and 17 more say “family owned” and stop there. 1 explains what happens after you get in touch.
None of that costs more than a conversation with the owner, and it is what separates a site a stranger believes from one they scroll past.
How many of the 49 readable sites clearly show each one.
2 in 5 are on WordPress. Platforms
WordPress sites had the lowest median mobile score of the common platforms: 67, against 70 to 77 for the builders.
That is about how they were built, not WordPress itself, which I build on every week: heavy themes, page builders, plugin stacks and untuned hosting. Built with care, WordPress scores in the 90s on the same test.
5 sites each, so read these as sketches. By category
10 kinds of business, 5 sites each. The numbers describe this sample, not every business of its kind in the county.
| Category (5 sites each) | Median mobile score | Median time to main content | Show reviews | Tap-to-call | Median E-E-A-T |
|---|---|---|---|---|---|
| General contractors and remodelers | 73 | 10.7 s | 3 of 5 | 3 of 5 | 61 |
| Roofing, siding and HVAC | 69 | 6.5 s | 3 of 5 | 5 of 5 | 89 |
| Plumbers and electricians | 73 | 6.7 s | 3 of 5 | 4 of 5 | 71 |
| Landscapers and tree services | 60 | 10.2 s | 4 of 5 | 5 of 5 | 57 |
| Dentists and chiropractors | 77 | 5.1 s | 5 of 5 | 5 of 5 | 69 |
| Auto repair | 79 | 5.2 s | 3 of 5 | 4 of 5 | 60 |
| Restaurants and cafes | 70 | 10.8 s | 2 of 5 | 3 of 5 | 49 |
| Salons, spas and barbers | 77 | 6.0 s | 1 of 5 | 3 of 5 | 42 |
| Real estate agents and small brokerages | 71 | 12.0 s | 3 of 5 | 5 of 5 | 64 |
| Law offices, accountants and insurance agents | 74 | 4.6 s | 3 of 5 | 4 of 5 | 72 |
The bar is low, and that is the opportunity. What it means
The typical county site is a little slow, a little stale and a little harder to act on than it needs to be, and so are its competitors.
A site that loads in under 3 seconds, shows its reviews, puts the phone number a tap away and changes every month is ahead of 4 in 5 sites in its town.
That is how I work: get the 3 jobs right, found, trusted and contacted, then keep measuring after launch. Send your site for a Pre-flight Check and I’ll go through it by hand and reply within 48 hours.
How the 50 were chosen and measured. Method
The sample: 50 independent small businesses with their own website, physically in New London County, in 10 categories with 5 sites each: general contractors, roofing and HVAC, plumbers and electricians, landscapers, dentists and chiropractors, auto repair, restaurants, salons, real estate, and law, accounting and insurance.
They are spread across Norwich, New London, Groton and Mystic, Waterford and East Lyme, Stonington, Ledyard, Montville, Colchester and 4 other county towns.
Sites were found by web search for the category and town and taken in order of appearance, skipping directories, franchises and chains, and businesses with only a Facebook page. 7 first-choice sites were dead (a domain that no longer resolves, a 404, an expired builder subscription) and were skipped; that is a finding of its own.
Independence: my own clients were excluded by name before sampling, and every sampled home page was checked for a credit to me or Movy Media. None matched. No business is named on this page; the results are reported as a group.
The measurements, all on November 6, 2026: Lighthouse 13.6 in mobile emulation (a mid-range phone on slow 4G), a single lab run per site, so medians are the number to quote.
Tap-to-call, phone text, the contact form’s visible and required fields, reviews and ratings, the copyright year, dated posts, the sitemap and the Last-Modified header were all read from the home page and the contact page by script, with the evidence kept per site. Nothing was submitted or clicked.
HTTPS was checked too: 49 of 50 certificates were valid, so it is not reported as a finding. The 1 site with an expired certificate could not be measured beyond it, so the page-level percentages use 49 sites.
The E-E-A-T pass, same day, same 50 sites: 3 pages per site at most, the home page, the About page and the contact page, read by script, nothing clicked. 20 signals, 5 per component, each scored 0, 3 or 5 with the evidence kept per site. Totals out of 100; Strong is 80 and up, Partial 50 to 79, Weak below 50.
Experience: real photos rather than stock, a portfolio or gallery, years in business, the owner’s own story, local knowledge beyond the address.
Expertise: a licence or certification, 3 or more service pages, educational content, named people with roles, the process explained.
Authority: an About page that names people, memberships and press, structured data, links to the business’s own profiles, reviews with names.
Trust: HTTPS, phone plus email plus address, a privacy policy, hours plus a way in, freshness.
6 of the signals (photos, the owner’s story, local knowledge, named people on the home and About pages, memberships and press) are judged from word and name patterns and can be wrong for a single site in either direction, one more reason the page reports the group and not any business. The rubric is additive and equal-weighted by choice: reproducible, not canonical.
Limits: a single lab run is not real-user data; the home page is not the whole site; 5 per category describes this sample only. The study will be repeated in late 2027 with the same method.
Questions about the study. Common questions
Which businesses were measured?
They are not named on the page. The study reports the 50 as a group, because the point is the county, not any one business.
The sample is 10 kinds of local business, 5 sites each, found by searching the way a customer would, with my own clients excluded so the numbers are independent.
What is E-E-A-T, and how did you score it?
E-E-A-T stands for experience, expertise, authoritativeness and trust: the qualities Google's search quality raters are told to look for in a site.
Google publishes no score, so I wrote a 20-signal rubric, 5 per component, and applied it the same way to every site's home, About and contact pages. Each site scores out of 100, and Strong is 80 and up.
Can you measure my site the same way?
Yes. Send it for a Pre-flight Check on the home page and I'll go through it by hand on a phone, the same measures plus whatever else I notice, and reply within 48 hours with what I'd fix first. No automated report, just a person looking.
Is the Lighthouse score the same as real-world speed?
No. It is a lab test: a single run on a simulated mid-range phone over a slow connection, which is a fair stand-in for a customer on a cell signal but not a measurement of your actual visitors.
That is why the page quotes medians across the sample rather than any single score, and why real-user data from Google is the better guide for one site.
What would you fix first on a typical site here?
Speed, then the phone number. Sizing the images for a phone and dropping unused scripts usually takes a 6 second page to under 3. Making the number tappable takes a minute.
Then reviews on the home page, a shorter form, and a date that moves. Most of that is a month of ordinary maintenance, not a rebuild.
