Exploratory data, 20 October 2026
A dated sample of 141 UK small-business homepages (20 October 2026)
On 20 October 2026 we checked the homepages of 141 UK shop websites listed in OpenStreetMap, to see how many answer, use HTTPS, include a mobile viewport tag and show a recent copyright year. This is a small, non-random, exploratory sample: read it as a snapshot, not as a measure of all UK small businesses.
Results
| Measure | Count (denominator stated) |
|---|---|
| Hosts in sample (unique, after de-duplication; 0 duplicates found) | 141 |
| Skipped because robots.txt disallowed the study bot | 3 of 141 |
| Homepages we attempted to fetch | 138 |
| Requests that failed or returned an error (observed on that day only) | 31 of 138 (22%) |
| Homepages that returned a page | 107 of 138 (78%) |
| Of those, HTML pages | 106 of 107 (99%) |
| HTML pages with a parked-domain signal | 1 of 106 (1%) |
| Returned pages not served over valid HTTPS (final address) | 13 of 107 (12%) |
| HTML pages with no mobile viewport meta tag | 4 of 106 (4%) |
| HTML pages with a copyright year found in the footer area | 59 of 106 (56%) |
| latest year 2024 or earlier | 12 of 59 (20%) |
| latest year 2025 | 2 of 59 (3%) |
| latest year 2026 | 45 of 59 (76%) |
| HTML pages where no copyright year was found (missing data) | 47 of 106 (44%) |
Failed requests (31): DNS lookup failed 12, connection failed 9, HTTP 403 7, HTTP 404 2, HTTP 302 not followed 1. These are what our single request saw on one day. A 403 may be a firewall blocking automated requests; a failed request does not prove a business has closed or has no website.
Method and limits
- Source of sites: OpenStreetMap shop objects with a website tag, from the Overpass API on 20 October 2026 (06:21 to 06:30 UTC). Each query was capped at 90 results per area and returned in the API's own order, so the sample is not random. Queries for 16 areas were attempted; 5 failed and nothing was used from them, and the crawl stopped after 141 hosts, so the sample covers only Greater London (89 sites) and Manchester (52 sites). It says nothing about other places.
- Hosts were de-duplicated by hostname, with and without "www": no duplicates were found among the 141.
- Each homepage was fetched once (robots.txt checked first, 10 to 15 second timeouts, at most 3 sites at a time) with a user agent naming the study and a contact address. Sites that disallowed automated access were skipped.
- "HTTPS" means the final address was HTTPS with a valid certificate. "Viewport tag" means the HTML has a viewport meta tag, not that the site looks good on a phone. "Copyright year" is the latest year found after the first 60% of the page text; pages without a year are counted as missing, not as old. Only the homepage was checked.
- Businesses are not named and no site addresses are published; the CSV holds one anonymous row per site.
Download the data
uk-small-business-website-sample-2026-10-20.csv (141 rows, UTF-8). Licence: Open Database Licence (ODbL) 1.0. Contains information from OpenStreetMap, © OpenStreetMap contributors (openstreetmap.org/copyright). If you reuse the data, credit both and share any derived database under the same licence.
Why it matters
If your own site is one of the 13 without valid HTTPS or the 4 without a viewport tag, a visitor on a phone may see a warning or a cramped page. See what a website costs over 1 and 3 years in our UK website costs by trade guide or the free cost calculator.
Published by Millwards Digital (millwards.shop), the studio that sells the websites mentioned here, so we have an interest in the topic. Collected and written by an automated assistant working for Millwards Digital; figures come from the log of the 20 October 2026 run.