<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Jeremy Powell</title>
    <description>The latest articles on DEV Community by Jeremy Powell (@powellhosts).</description>
    <link>https://dev.to/powellhosts</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4246540%2F9462ed0c-33b0-489f-a85a-dd875869a5ba.png</url>
      <title>DEV Community: Jeremy Powell</title>
      <link>https://dev.to/powellhosts</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/powellhosts"/>
    <language>en</language>
    <item>
      <title>A homepage check is not a usability verdict: lessons from 224 garage websites</title>
      <dc:creator>Jeremy Powell</dc:creator>
      <pubDate>Thu, 19 Nov 2026 11:32:49 +0000</pubDate>
      <link>https://dev.to/powellhosts/a-homepage-check-is-not-a-usability-verdict-lessons-from-224-garage-websites-409a</link>
      <guid>https://dev.to/powellhosts/a-homepage-check-is-not-a-usability-verdict-lessons-from-224-garage-websites-409a</guid>
      <description>&lt;p&gt;A website check is easy to make look more authoritative than it is. Count an HTML feature, turn it into a percentage, give the percentage a red badge, and a narrow observation starts to resemble a verdict on an entire business.&lt;/p&gt;

&lt;p&gt;That matters when you build for garages and trades. The owner wants to know whether customers can contact them. A developer wants evidence that a change is needed. Neither is helped by a score that claims more than the check actually established.&lt;/p&gt;

&lt;p&gt;I'm Jeremy Powell. Before building websites, I spent 15 years in the motor trade, including seven as an MOT tester. Our garage website study gave me a useful reason to think about the boundary between a repeatable check and the practical conclusion you draw from it.&lt;/p&gt;

&lt;h2&gt;
  
  
  Start with the denominator, not the headline
&lt;/h2&gt;

&lt;p&gt;The starting sample contained 238 active UK independent garage, MOT and vehicle businesses. They were registered limited companies and published a business email on their own websites. Of those websites, 224 loaded; 14 did not respond and were excluded from the findings denominator.&lt;/p&gt;

&lt;p&gt;The checks ran on 3 November 2026. Read-only visits used a normal desktop Chrome browser identity, visited each homepage, tried the HTTP address, inspected up to ten homepage images and examined available Internet Archive homepage history. The published study reports 41 unit tests for the checking tool.&lt;/p&gt;

&lt;p&gt;That describes the recorded method. It does not establish comprehensive browser coverage, a security audit or a representative national sample. Requiring a published email may favour businesses with better-maintained sites. Excluding non-responders also means you cannot silently include them in the denominator when discussing a particular homepage feature.&lt;/p&gt;

&lt;p&gt;The &lt;a href="https://powellhosts.co.uk/garage-website-index" rel="noopener noreferrer"&gt;Garage Website Index&lt;/a&gt; publishes the aggregate table and limitations. No sampled business is identified. There is one edition, so there is no measured improvement or decline to report.&lt;/p&gt;

&lt;h2&gt;
  
  
  A telephone number and a telephone link are different observations
&lt;/h2&gt;

&lt;p&gt;The homepage check found no tap-to-call link on 79 sites, reported as 35% of the 224 that loaded. That does not mean those sites displayed no phone number. Nor does it mean they lacked a telephone link on a contact page.&lt;/p&gt;

&lt;p&gt;For a developer, this is a useful distinction between content and interaction. A number rendered as text may be readable, but the visitor still needs a way to act on it. Conversely, a telephone link existing somewhere in the markup does not establish that its label is clear, its destination is correct or its position is useful on a small screen.&lt;/p&gt;

&lt;p&gt;I would write the follow-up acceptance criterion around the customer action: the published homepage offers an understandable route to calling the business on a phone. That calls for a human check alongside any markup check. A count of links is supporting evidence, rather than the whole acceptance decision.&lt;/p&gt;

&lt;p&gt;We did not measure abandoned enquiries or lost bookings. There is no honest conversion-loss figure to attach to the missing-link count.&lt;/p&gt;

&lt;h2&gt;
  
  
  Separate secure availability from a secure default
&lt;/h2&gt;

&lt;p&gt;Eighteen sites did not redirect their HTTP address to HTTPS, reported as 8%. Separately, 16 had no working HTTPS connection, reported as 7%.&lt;/p&gt;

&lt;p&gt;Those checks answer different questions. Can the secure version be reached? Does the plain address lead to it? A site can fail a redirect check without proving that the secure endpoint is unavailable. The findings can overlap, so adding the counts would misdescribe the evidence.&lt;/p&gt;

&lt;p&gt;The practical lesson is to keep the observations separate in a report and in the work handed back to a developer. "Secure endpoint unavailable" and "plain address stays unsecured" need different explanations. A browser problem is also not proof that a site has been hacked.&lt;/p&gt;

&lt;p&gt;This is particularly important in a customer-facing report. "Security failed" is a much broader statement than the request-level evidence supports. Say which address was checked and which behaviour was observed, then explain the scope of the follow-up.&lt;/p&gt;

&lt;h2&gt;
  
  
  Copyright dates are weak evidence of upkeep
&lt;/h2&gt;

&lt;p&gt;Thirty-seven sites displayed a footer copyright year before 2024, reported as 17%. A developer could make that finding disappear by changing a date. That would not establish that opening hours, service descriptions or contact details had been reviewed.&lt;/p&gt;

&lt;p&gt;The useful action is a content review, with the date as a prompt. A footer is not a maintenance log. A business might keep its important information accurate while leaving the footer alone, or show a current year while retaining obsolete instructions elsewhere.&lt;/p&gt;

&lt;p&gt;Archive evidence needs the same restraint. Fifteen homepages appeared unchanged for at least three years in the available captures. Archive coverage is incomplete, and small incidental changes count as changes in this check. The published method therefore describes an undercount, not a complete record of maintenance activity.&lt;/p&gt;

&lt;h2&gt;
  
  
  CMS detection should describe, not condemn
&lt;/h2&gt;

&lt;p&gt;The study identified 24 sites using Wix, reported as 11%. That is a platform descriptor. It is not a fault category, and it is not a complete breakdown of the remaining CMS mix.&lt;/p&gt;

&lt;p&gt;A report that places a platform name beside broken features can accidentally imply that the platform is itself broken. Keep inventory and defects distinct. If the actual problem is an unusable contact route, describe that route and the expected behaviour.&lt;/p&gt;

&lt;p&gt;The same discipline applies to mobile checks. Eight sites had no viewport setting, reported as 4%. Presence of that setting would not prove that menus, text and contact controls work well on a phone. Passing a narrow check does not justify a broad usability badge.&lt;/p&gt;

&lt;h2&gt;
  
  
  Turn the findings into an honest handover
&lt;/h2&gt;

&lt;p&gt;For small-business developers, the useful output is a short list of observable problems with a clear scope, not an alarming overall grade. Keep the date, page coverage and denominator beside the result. Separate observation from interpretation, and interpretation from a proposed fix.&lt;/p&gt;

&lt;p&gt;Our &lt;a href="https://powellhosts.co.uk/garage-tech-headaches" rel="noopener noreferrer"&gt;garage technology guide&lt;/a&gt; describes the owner-facing concerns behind this work: customers finding the business, using the site on a phone and making contact. Those are better acceptance goals than simply making every diagnostic indicator green.&lt;/p&gt;

&lt;p&gt;A site can pass an HTML check and still confuse a customer. It can fail a footer-date check while publishing accurate service information. Good reporting leaves room for both, so the owner pays for useful work rather than a tidier score.&lt;/p&gt;

&lt;h2&gt;
  
  
  Links
&lt;/h2&gt;

&lt;ul&gt;
&lt;li&gt;&lt;a href="https://powellhosts.co.uk/garage-website-index" rel="noopener noreferrer"&gt;https://powellhosts.co.uk/garage-website-index&lt;/a&gt;&lt;/li&gt;
&lt;li&gt;&lt;a href="https://powellhosts.co.uk/garage-tech-headaches" rel="noopener noreferrer"&gt;https://powellhosts.co.uk/garage-tech-headaches&lt;/a&gt;&lt;/li&gt;
&lt;/ul&gt;

</description>
      <category>webdev</category>
      <category>a11y</category>
      <category>testing</category>
      <category>beginners</category>
    </item>
  </channel>
</rss>
