The Sherborne Digital Report 2026

The non-social addresses include two branch or platform destinations. A public link does not establish ownership of the destination. This selected list is not a census of Sherborne.

A complete seven-day hours table was visible on 112 of 126 profiles. Three profiles displayed an explicit permanent-closure notice. Neither “closed now” nor the absence of a closure notice verifies trading status. Ten reviews is a descriptive threshold, not a Google requirement.

Review totals

103 addresses = 89 eligible responses + 12 blocked requests + two failed requests. The 14 unavailable outcomes are excluded from the feature counts. Six eligible final addresses had non-root paths, so these are linked-page checks, not necessarily homepage checks.

Details may use a different format, appear on another page or load with JavaScript. JSON-LD is one structured-data format. An HTTP final address does not establish that HTTPS is unavailable. A blocked automated request does not establish that customers cannot use the site. Read the check definitions

103 current linked addresses tested; 101 results included after response and content screening. 2 unavailable or excluded outcomes are not scored as zero. Each included page had one controlled, simulated mobile run. This browser check is separate from the HTTP-only check, so their eligible sets can differ.

The middle half of the recorded LCP values ran from 5.4 to 13.4 seconds. This is the spread across tested pages, not repeat tests of one page.

Recorded loading times

What the pages transferred

Local Lighthouse 13.5.0, Chrome 155, mobile emulation and simulated throttling. No September Lighthouse baseline exists, so these results do not show a September-to-October improvement or decline. How Lighthouse performance scoring works

Rules flagged

43 of 101 page runs flagged both the text-contrast and accessible-link-name rules. The counts overlap; they cannot be added together to count affected pages.

The accessibility rules cannot test every requirement or customer task. The SEO rules cover selected technical checks, not content quality or Google positions. A small number of element-level flags can affect a score; review the actual page before deciding the fix. What an accessibility score covers

1. Follow your public link.

   Open your Google profile as a customer would. Follow the website link, check the destination and make your public contact details, hours or appointment arrangements clear.

   How to edit your Google profile

2. Complete one task on a phone.

   Find a service, make an enquiry, book or buy. Test the form or checkout. Check that text is readable, key controls have clear names and the journey works with a keyboard. Publish location details appropriate to your business; a private address need not be added to satisfy a checklist.

3. Investigate the obstacle first.

   Start with an incorrect link, unreadable text, an unusable form or a slow important page. For loading problems, inspect the largest content element, image size and unnecessary resources. Check the result after making the change; a score alone does not show that the customer journey works.

4. Get a current page check.

   Our free website audit gives a fresh Lighthouse-based page check and practical next steps. It is separate from this selected local study and does not predict ranking gains.

   Open the free website audit

1. Sources and dates.

   The original list dates from 4 September 2026. We revisited its 127 names on 15 October 2026, using visible public Google Maps overview information and expanded weekly-hours tables. Linked-page HTML was collected between 09:48 and 10:14 UTC. Lighthouse tests ran between 09:54–10:36 UTC.

2. Selection and matching.

   The list is seeded and non-random. It is not a census, a representative town sample, verified business ownership or a strict six-mile boundary. The original discovery used a 15 km Google Places location bias, which is not a radius restriction.

   We checked retained identifiers against visible names and local details. Two matches were repaired after a focused name-and-locality search; original identifiers and evidence were retained. Six name differences were reconciled. One unresolved identity is excluded. A repaired match is a method change, not a business improvement.

3. What each count includes.

   Profile observations use 126 verified matches, including three with explicit permanent-closure notices. Ratings and review totals each use 119 observed values; seven are unknown. Weekly hours means a full seven-day table was visible. The 107 observed links include four Facebook links; the other 103 addresses form the page-test scope. HTML feature counts use 89 eligible responses. Lighthouse results use 101 eligible browser measurements, with each rule showing its applicable denominator.

   Unknown, blocked and failed values are not converted to zero. No photo totals, revenue, lost sales, ownership or actual search positions were measured. No individual business is named in the published findings.

4. The linked-page method.

   We requested the exact current public links using a consistent audit user agent, ordinary redirects, TLS verification and a 25-second timeout. No sign-in or page JavaScript was used for this stage. Same-day outcomes were reused only for exactly matching requested addresses after removing fragments: 101 reused outcomes and two changed-link requests. Response bodies, headers and statuses remain in the private research record.

   Eligibility requires a successful HTML response without an obvious error/interstitial, truncation or Facebook destination. Limited error screening cannot prove a complete or usable page.

5. The HTML detectors.

   Read the exact check definitions

   No UK postcode pattern detected. Original uppercase UK-postcode regex did not match extracted HTML text. This is not proof that no address/location is supplied.

   No UK phone-number pattern detected. Original UK telephone regex did not match BeautifulSoup-extracted response-HTML text. A pattern not detected is not proof no contact route exists.

   No JSON-LD marker detected. Case-sensitive application/ld+json string absent. No schema validity/type, alternative formats or JavaScript-injected data were tested.

   Empty or absent meta description. First matching meta description tag absent/empty in response HTML; not a search ranking or Google snippet verdict.

   Final fetched address used HTTP. Final response address did not start https://; this does not establish HTTPS availability or a complete security assessment.

   No viewport tag detected. Exact name=viewport meta tag absent. Not a test of responsive behaviour, zoom or accessibility.

   These are limited response-HTML checks. The postcode pattern is uppercase, the JSON-LD marker is case-sensitive, and the meta-description check uses the first matching tag. Copyright years are not treated as dates of content maintenance.

6. The browser-test conditions.

   Local Lighthouse 13.5.0 with Chrome 155.0.8059.40: one sequential mobile navigation per linked address, a fresh isolated Chrome profile and simulated throttling. Screen: 412 × 823 CSS pixels, device scale 1.75. Simulation: 150 ms RTT, 1,638.4 Kbps throughput and 4× CPU slowdown.

   Reports were screened for HTTP/runtime failures and interstitials using the captured page evidence. Ordinary consent overlays remain part of the observed experience. Dynamic resources, network conditions and the local machine can affect a run. Repeat runs are needed to assess variability. These are lab results, not real-user percentile measurements; Total Blocking Time is not INP and the report does not declare a Core Web Vitals pass.

7. Distributions and transferred resources.

   Quartiles use inclusive linear interpolation across the recorded values. Q1 and Q3 are the 25th and 75th percentiles across page runs, not field-user percentiles. Review totals use only visible values; unobserved totals stay unknown.

   The resource chart pools seven disjoint resource-summary types across 101 runs. Larger pages contribute more bytes. Third-party is an overlapping classification and is not added again. Resource-summary and total-byte-weight are separate audit totals: 21 reports differ, by at most 21,772 bytes. The transfer chart uses its own total of 413,572,880 bytes and does not estimate waste or savings.

8. How to compare and cite this update.

   September profile data came from Google Places API fields; October uses visible Maps observations and corrected matching. The sources and eligible sets differ. Do not read differences as business improvements or declines. There are no September Lighthouse results for a speed comparison.

   Cite: The Artist Studio, The Sherborne Digital Report 2026, updated 15 October 2026. State the measure and its denominator when sharing a finding. The source date is a measurement date, not a promise that a changing website or listing remains the same.

9. Source guidance.

   • Editing a Google Business Profile
   • Lighthouse overview and automated checks
   • Lighthouse performance scoring
   • Accessibility scoring and manual checks
   • Google’s structured-data guidance

Download the aggregate findings (CSV)

Counts, medians, denominators, definitions and source dates. No individual business rows.