Search, AI readiness, speed, accessibility, design and conversion, measured and compared with the four web designers that rank first for “web design Dover”.
The site is well built: every AI crawler can read it, it passes almost every accessibility check, and on desktop it's as fast as anything in the comparison. It scores 84 against a competitor average of 68, the highest of the five. But it isn't being found. It's missing from the top 20 for “web design Dover” while all four competitors are in the top 4, the Google listing's name belongs to another business, and on a phone the main content takes 3.5s to 5.4s to appear. Those three are where the work is.
Overall score
84
Competitor average 68; best 81
Findings
12
2 critical, 4 serious, 6 medium
Position for “web design Dover”
20+
Competitors: A 1, B 4, C 2, D 3
Main content on a phone
4.0s
Web Services; Google's “good” is 2.5s
Fix these first
1You're not in the top 20 for "web design Dover"Check Search Console for the impressions Web Services gets for Dover searches, so you start from real numbers. Effort: two to four hours to start; reviews and links build over months.Critical
3Pages take 3.5 to 5.4 seconds to show their main content on a phoneSelf-host the four font files and preload the two used above the fold; show the first screen without the reveal animation; load Analytics after the page has painted, or switch to a lighter tool. Effort: half a day.Serious
5Search snippets get cut shortRewrite the 15 descriptions to under 155 characters with the answer first, and shorten the 7 post title tags.. Effort: two hours.Serious
What's working
All 32 pages carry their text in the HTML: 23,450 words an AI crawler can read without running any script.
160 of 160 fetches as GPTBot, ClaudeBot, PerplexityBot and others got exactly what a browser gets. Two of the four competitors turn GPTBot away at the server.
Lighthouse accessibility 98 on every page tested, 1,456 controls checked, 0 without a name.
Desktop speed 93 to 99 out of 100, and the lightest page weight bar one competitor.
Prices on the page and in the structured data. Only one competitor shows any price.
Where it's behind
Search: not in the top 20 for three local searches (F1), and a name clash on the Google listing (F2).
Phones: main content in 3.5s to 5.4s; Competitor C manages 2.0s (F3).
Proof: no rating or review count; Competitor C puts “5.0 on Google · 65 reviews” in its first screen (F4).
On a phone, the first screen of Web Services has no button to get in touch; the form starts below it.
02
How you compare
The four competitors are the four web designers that hold positions 1 to 4 for "web design Dover" in the search snapshot below. In a client's report they are named, with screenshots. This sample is public, so they are labelled A to D and their screenshots are left out.
In short
rgleeson.com scores 84, 16 points above the competitor average. It leads on AI and agent readiness, accessibility, design and visual quality, and trails on search visibility, where every competitor outranks it. Competitor C is the one to beat: fast, clear, and showing its reviews.
Overall score
rgleeson.com84
Competitor C81
Competitor A68
Competitor D65
Competitor B57
By category
Site
AI 20%
Search 20%
Speed 15%
Accessibility 15%
Design 10%
Content 10%
Trust 10%
Overall
rgleeson.com
100
65
76
99
78
85
80
84
Competitor A
60
93
67
65
58
47
69
68
Competitor B
50
83
55
38
55
75
34
57
Competitor C
85
75
100
68
73
90
79
81
Competitor D
60
90
62
70
53
42
54
65
Each category is out of 100; the overall score weights them by the percentages shown. Bold is the best in the column. The rules behind every score are in section 12.
Who the competitors are
Label
What it is
Position for “web design Dover”
Built with
Competitor A
website builder (Gridbox), Dover
1
Powered by Website Builder Gridbox
Competitor B
WordPress studio, East Kent
4
WordPress 7.1.4
Competitor C
agency, Dover, static site
2
Static, on Netlify
Competitor D
WordPress freelancer, Dover
3
WordPress 7.1.4
03
Search visibility
Where the site appears for the searches a local customer would make, and whether its pages are set up to rank.
In short
The pages are technically sound: every canonical correct, one h1 each, a clean sitemap, no broken internal links. The problem is that the site doesn't appear for any of the local searches tested. The brand search is worse: “RG Web Design Dover” returns a different business called RG Web Design, in Sussex, as result 1.
Search snapshot, 4 November 2026
Search
rgleeson.com
A
B
C
D
“web design Dover”
Not in top 20
1
4
2
3
“website designer Dover Kent small business”
Not in top 20
1
2
3
6
“web design Dover Kent small business website designer”
Not in top 15
1
2
3
9
“RG Web Design Dover”
Not in top 10
2
9
3
6
“AI readiness website audit UK”
Not in top 20
—
—
—
—
Firecrawl web search API, location set to Dover, England; top 20 organic results recorded (top 10 for the brand query). Not a personal Google search: no history, no map pack, no ads.
On-page set-up, all 32 pages
Check
Result
Titles
25 of 32 within 60 characters; 7 post titles run long (F6)
Meta descriptions
17 of 32 within 160 characters; 15 run long, including Web Services at 189 (F6)
Canonicals
32 of 32 correct
One h1 per page
32 of 32
Heading order
0 of 32 without a skipped level (F9)
Structured data
16 of 32 pages (F7)
Sitemap
32 URLs, all returning 200, declared in robots.txt
www and https in one hop, except http://rgleeson.com, which takes two (F12)
04
AI and agent readiness
Whether ChatGPT, Claude, Perplexity and Google can read the site, and whether an AI agent could use it on someone's behalf.
In short
This is the site's strongest area. All the text is in the HTML, every crawler gets the real page, and the interface passes the agent-friendly checks on every page. Two competitors look open in their robots.txt but turn GPTBot away at the server, so they're invisible to ChatGPT without knowing it. What's missing here is an action an agent can take: there's no way to book (F5).
Can the AI crawlers read the page?
Site
robots.txt
Fetched as GPTBot
Words in raw HTML
llms.txt
Structured data
rgleeson.com
✓ Allows
✓ Same page
971
✓ Yes
✓ ProfessionalService
Competitor A
✓ Allows
✕ Refused (429)
804
✕ No
✓ LocalBusiness
Competitor B
✓ Allows
✕ Connection dropped
1,127
✕ No
✓ Yes
Competitor C
✓ Allows
✓ Same page
1,012
✕ No
✓ Organization
Competitor D
✓ Allows
✓ Same page
613
✕ No
✕ No
Each site's page that ranks for “web design Dover”. GPTBot fetches re-tried twice by hand; both refusals repeated.
No page shows a visitor more than it sends a crawler
Controls without a name
0
of 1,456 at 375 and 1440px
llms.txt links working
29/29
5,126 bytes, plain text
Page
Words in raw HTML
Words a visitor sees
/
321
233
/websites.html
972
955
/audit.html
437
426
/contact.html
125
117
/project-action-carpets.html
777
759
/post-website-invisible-to-ai.html
1,127
1,118
The raw count is higher because it includes text hidden until needed, such as the mobile menu.
05
Speed
Lighthouse lab tests on a simulated mid-range phone on a slow 4G connection, and on desktop. Lab numbers, because the site has too little traffic for Google's real-user data.
In short
On desktop the site is quick: 93 to 99 on every page. On a phone it's slow to show its main content: 3.5s to 5.4s, against Google's 2.5s. The pages are light; the delay comes from what has to load before anything paints (five servers for two fonts and the page) and from the fade-in animation on the first screen. A half-day fix (F3).
Main content painted on a phone
rgleeson.com4.0s
Competitor A5.6s
Competitor B6.9s
Competitor C2.0s
Competitor D7.6s
rgleeson.comCompetitorsGoogle's “good”: 2.5s
Six pages of rgleeson.com
Page
Score, phone
Score, desktop
Main content, phone
First paint, phone
Layout shift
Weight
Requests
Home
84
99
3.7s
3.0s
0.040
671 KB
18
Web Services
81
98
4.0s
3.1s
0.001
505 KB
18
AI audit
75
93
4.8s
3.3s
0.000
306 KB
14
Work
90
99
3.5s
1.6s
0.022
864 KB
21
Action Carpets case study
80
99
4.2s
3.2s
0.000
392 KB
17
Journal post
72
99
5.4s
3.5s
0.000
404 KB
16
What the page weighs, and who it comes from
Site
Total
Images
Scripts
Fonts
Other
From other servers
rgleeson.com
505 KB
202
163
109
31
53%
Competitor A
739 KB
222
293
157
67
27%
Competitor B
1,833 KB
757
640
193
244
18%
Competitor C
266 KB
190
1
34
40
0%
Competitor D
1,458 KB
706
341
344
68
33%
Kilobytes transferred on a phone. “From other servers” is the share loaded from third parties, such as analytics and font services.
Why it's slow on a phone
Before the first paint the browser has to reach five servers: this site, api.fontshare.com and cdn.fontshare.com for Cabinet Grotesk, fonts.googleapis.com and fonts.gstatic.com for Inter. On a slow connection each new server costs a round trip.
Google Analytics is 158 KB, the largest single file on most pages, and Lighthouse finds 44% of it unused. Scripts are 163 KB of the Web Services page; the site's own script is 5 KB of that.
The hero headline fades in word by word and sections start invisible. The browser can't count the main content as shown until the script has run.
Journal covers are 2,000px wide, fill a 364px slot on a phone, and are lazy-loaded though they're in the first screen (F8).
06
Accessibility
The same checks an AI agent depends on: contrast, names, labels, landmarks and tap targets, measured on the rendered page.
In short
Close to clean. Lighthouse scores 98 on every page; the only items it raises are the footer headings skipping a level (F9). Across the site, six links are a few pixels under the 24px tap target (F10). One competitor fails contrast on 101 of 131 text elements, mostly orange on white.
Site
Lighthouse, phone
Text failing contrast
Small targets, phone
Images without alt
Main and skip link
rgleeson.com
98
1 of 141
0
0
✓ Yes
Competitor A
90
4 of 36
0
0
✕ No main
Competitor B
76
101 of 131
82
2
✕ No main
Competitor C
96
16 of 155
11
0
✓ Yes
Competitor D
79
0 of 55
22
0
✕ No main
Contrast measured on every visible text element at 1440px against the background behind it: 4.5:1, or 3:1 for large text. Text over images is left out, not counted as a pass.
rgleeson.com's one contrast failure is the decorative “·” between a name and a job title (#D0D0D0, 1.54:1). Across 15 pages, a separate contrast audit on 4 November measured 2,832 text elements with 10 failing, all of them minor.
07
Design review
Colour, type and layout, measured from the rendered page and then reviewed by eye at 1440 and 375px.
In short
A disciplined system: one accent colour, two typefaces (plus a monospace in one illustration), consistent components, comfortable body text. It has the top design score by the rules, but by eye Competitor C's page is the stronger one: bolder type and proof in the first screen. Four things would lift it: more of the text in ink rather than grey, fewer stray type sizes, proof in the first screen, and showing the first screen without waiting for an animation.
Colour palette, as used on Web Services
Ink#0F1014Headings, primary text, dark bands21% of text · 9% of the page19.01:1 on white
Muted grey#5C5C5CBody and supporting text60% of text6.69:1 on white
Subtle grey#6A7185Labels and captions1% of text4.87:1 on white
Coral orange#D04205Buttons, links and labels1% of text · 5% of the page4.70:1 on white
Bright orange#F36435Large display words, accents on ink3.14:1 on white
Page tint#F9FAFFAlternate section background84% of the page1.04:1 on white
White#FFFFFFPage and card background6% of text1.00:1 on white
Border#E2E2E2Dividers and card edges1.30:1 on white
Share of text is by characters; share of the page is by pixels in a full-page screenshot at 1440px. The accent covers 5% of the page: enough to lead the eye, not enough to tire it.
Each bar shows the main colours by area, photos included, with the typefaces alongside. Competitors A and D get most of their colour from stock photos; B from orange text.
Typography
Site
Typefaces
Sizes in use
Body size / line height
Characters per line
Headline
rgleeson.com
Inter, Cabinet Grotesk, ui-monospace
14
16px / 1.6
47
80px
Competitor A
Montserrat
9
19px / 1.6
49
65px
Competitor B
Roboto Condensed, Exo
8
16px / 1.7
106
36px
Competitor C
Archivo
16
16px / 1.6
97
108px
Competitor D
Open Sans, Poppins
12
15px / 1.59
51
60px
Comfortable reading: body 16px or more, line height 1.4 to 1.8, 45 to 85 characters per line. B and C run body lines past 95 characters on desktop.
Type sizes on Web Services, by share of text. Grey bars carry under 3% each.
3%
12%
3%
8%
42%
1%
4%
15%
8%
0%
2%
2%
0%
1%
1213141516171820243236445280
First screen
Home, 1440px. Written for hiring managers; Web Services is one link among six.Home on a phone.
Expert review
Four questions, 0 to 5 each: is the hierarchy clear (one thing to read first, one thing to do)? Are colour, type and components used consistently? Do the images show the actual business and its work? Would you remember it next to the others? Scored from the first screen and a full scroll at 1440 and 375px.
Site
Hierarchy
Consistency
Imagery
Distinct
Total /20
Notes
rgleeson.com
4
5
3
3
15
Clear headline, one accent colour used sparingly, every component from one system. The first screen has no picture of real work and no proof, and the layout reads like many well-made studio sites.
Competitor A
2
2
1
2
7
A stock photo of a woman making goggles with her fingers fills the first screen under a purple-to-blue gradient; the cookie banner covers the call to action.
Competitor B
1
1
2
2
6
Seven-item navigation, a clip-art logo, three typefaces and body lines over 100 characters long. A photo of Dover's guns is local, but doesn't show the work.
Competitor C
5
5
3
4
17
Huge, confident type, two calls to action and a proof bar (5.0 on Google, 65 reviews) in the first screen. The strongest page of the five.
Competitor D
3
2
1
1
7
A stock close-up of an eye behind the headline; "Website Design in Dover" repeated as the h1, h2 and h3. A standard theme with little changed.
Recommendations
R1. Most of the text is grey, not ink
60% of the characters on Web Services are muted grey #5C5C5C and 21% ink #0F1014. The grey passes contrast at 6.69:1 on white.
Do: Keep grey for supporting text, but set the opening paragraph and card descriptions in ink. Contrast between levels is what makes the hierarchy, and right now most text sits on the second level.
R2. 14 type sizes on one page
Text on Web Services uses 14 sizes from 12 to 80px; five of them (12, 14, 17, 32 and 52px) carry under 3% of the text each.
Do: Fold the stray sizes into the nearest step of a 9-step scale (13, 15, 16, 20, 24, 32, 44, 64, 80) and name each as a token, so new pages can't add more.
R3. No proof in the first screen
The first screen at 1440 and 375px has a headline, a paragraph and the enquiry form. The one testimonial is halfway down the page.
Do: Add one line of proof under the headline: a client result ("Vancat: sales up 200%") or, once there are enough, the Google rating and review count.
R4. Content waits for an animation
Sections start at opacity 0 and fade in as they scroll into view; the hero headline fades in word by word. Lighthouse counts the page as painted only once that finishes.
Do: Show the first screen without any reveal, and keep the fade for content further down. This is part of finding F3.
08
Content and conversion
What a visitor finds when they're deciding whether to get in touch.
In short
Web Services gives a visitor more than any competitor page: prices, a form, a phone number and plain-English copy (reading ease 68). What it lacks is proof near the top. One client quote, halfway down, against a competitor showing 65 Google reviews in its first screen. And on a phone, the first screen has no button to get in touch.
Site
Contact in first phone screen
Tap to call
Form
Prices
Reviews
Reading ease
Words
rgleeson.com
✕ No
✓ Yes
✓ Form
£400, £150
✕ Quotes only
68
958
Competitor A
✓ Contact Us
✕ No
✕ No
✕ None
✕ Quotes only
57
576
Competitor B
✓ Let’s Talk
✓ Yes
✓ Form
✕ None
✕ Quotes only
15
1,078
Competitor C
✓ Talk to us →
✓ Yes
✕ No
£2k, £6k
✓ Rating
64
852
Competitor D
✕ No
✓ Yes
✕ No
✕ None
✕ Quotes only
49
588
Reading ease is the Flesch score of the main text: 60 to 70 is plain English, under 30 is hard going.
What I'd change
Put a “Start a project” button in the phone's first screen, pointing at the form, so nobody has to scroll to find it.
Move the Vancat result (sales up 200%) up beside the headline, and add the Google rating once there are five reviews (F4).
Name the towns you've worked in (Dover, Deal, Folkestone) in the case studies that link from this page. It helps local searches and tells a visitor you're nearby (F1).
09
Trust and security
The signals that tell browsers, search engines and visitors that the site is safe and the business is real.
In short
Sound. HTTPS everywhere with HSTS, the main security headers present, Lighthouse best practices 100, and the business described in structured data with its address and prices. The one gap is a Content Security Policy (F11), which only Competitor A sends.
Site
HSTS
CSP
nosniff
Frame protection
Business in structured data
Best practices
rgleeson.com
✓ Yes
✕ No
✓ Yes
✓ Yes
✓ ProfessionalService
100
Competitor A
✓ Yes
✓ Yes
✓ Yes
✓ Yes
✓ LocalBusiness
96
Competitor B
✕ No
✓ Yes
✕ No
✕ No
✕ No
71
Competitor C
✓ Yes
✕ No
✕ No
✕ No
✓ Organization
96
Competitor D
✕ No
✕ No
✕ No
✕ No
✓ ProfessionalService
96
10
Findings
Three questions per finding: does it stop someone doing what they came for, does it affect many visitors or searches, and will it keep happening until it's fixed? Three yeses = Critical, two = Serious, one = Medium.
CriticalF1· Search
You're not in the top 20 for "web design Dover"
Stops a task: yesMany affected: yesKeeps happening: yes
What I found
Three local searches on 4 November 2026 ("web design Dover", "website designer Dover Kent small business" and one longer variant): rgleeson.com appears in none of the top 15 to 20 results.
All four competitors in this report appear in all three, in positions 1 to 9; for "web design Dover" they hold positions 1 to 4.
The Web Services page is one page out of 32. The home page and 12 case studies are about product design, so most of the site doesn't mention Dover or web design for local businesses.
Why it matters
People looking for a local web designer search for it, and AI assistants build their answers from the same results. A page that isn't in the top 20 isn't seen, however good it is.
What to do
Check Search Console for the impressions Web Services gets for Dover searches, so you start from real numbers. Then: ask your website clients for Google reviews; publish the Action Carpets and Downtogroundworks case studies with their towns named; list the business on the local directories that appear in these results (FreeIndex, Kent Small Business, Business Magnet); and link to Web Services from the home page's first screen.
Effort: Two to four hours to start; reviews and links build over months
CriticalF2· Search
Your Google listing's name is missing from your site, and another business owns it
Stops a task: yesMany affected: yesKeeps happening: yes
What I found
The Google Business Profile trades as "RG Web Design" (from a profile screenshot, 24 September 2026). The name appears 0 times across the 32 pages in the sitemap.
A search for "RG Web Design Dover" returns rgwebdesign.co.uk as result 1: a different web design business, based in Sussex. rgleeson.com is not in the top 10.
The structured data calls the business "Richard Gleeson — Web Design & Build", and its sameAs links point only to LinkedIn.
Why it matters
Google and AI assistants match a business across sources by its name. Here the listing's name points at someone else's website, and your website never claims it.
What to do
Pick one public name. If it stays RG Web Design, show "Trading as RG Web Design" on every page, add it as alternateName in the structured data and add the Google profile to sameAs. If the clash is a problem, rename the listing to match the site.
Effort: An hour, plus the naming decision
SeriousF3· Performance
Pages take 3.5 to 5.4 seconds to show their main content on a phone
Stops a task: noMany affected: yesKeeps happening: yes
What I found
Lighthouse, mobile, simulated slow 4G: largest contentful paint 3.5 to 5.4s on all six pages tested. Google's "good" line is 2.5s. Competitor C does it in 2.0s.
Before the first paint the browser has to visit five servers: this site, two for Cabinet Grotesk (Fontshare) and two for Inter (Google Fonts).
Third-party files are 53% of the Web Services page's weight and 88% of the AI audit page's: Google Analytics (158 KB, 44% of it unused) and the fonts.
The hero headline fades in word by word and sections start invisible, so the main content isn't counted as painted until the animation runs.
Why it matters
On desktop the same pages score 93 to 99, so this is a phone problem, and most local searches happen on phones. Speed is also one of the signals Google uses to rank pages.
What to do
Self-host the four font files and preload the two used above the fold; show the first screen without the reveal animation; load Analytics after the page has painted, or switch to a lighter tool. Re-run Lighthouse to confirm under 2.5s.
Effort: Half a day
SeriousF4· Content
There's no rating or review count anywhere on Web Services
Stops a task: noMany affected: yesKeeps happening: yes
What I found
Web Services shows one client quote (Vancat, "sales have increased 200%"), halfway down the page. There's no Google rating, no review count and no AggregateRating in the structured data.
Competitor C puts "5.0 on Google · 65 reviews" in its first screen and in its structured data. Competitor A has a "Top rated reviews" section of client quotes.
Why it matters
Someone choosing between five web designers looks for proof that other local businesses chose you. Reviews are also one of the things that decides who appears in Google's map results.
What to do
Ask the clients on the Work page for a Google review. Once there are five or more, show the rating and count under the Web Services headline and add AggregateRating to the business's structured data.
Effort: An hour to ask; showing it takes twenty minutes
SeriousF5· AI and agents
An AI agent can't book a call with you
Stops a task: yesMany affected: noKeeps happening: yes
What I found
contact.html and websites.html offer enquiry forms only. There's no booking link on any page and no booking action in the structured data. None of the four competitors has one either.
Why it matters
An agent acting for someone will read a contact form but won't fill in a stranger's form unprompted. A booking page with real slots is an action it can take. Few people use agents this way yet, which is why this isn't Critical.
What to do
Turn on bookings in the Google Business Profile, or publish a booking page, and link it from Web Services, Contact and the structured data.
Effort: An hour, plus deciding on available slots
SeriousF6· Search
Search snippets get cut short
Stops a task: noMany affected: yesKeeps happening: yes
What I found
15 of 32 meta descriptions are over 160 characters, including Web Services (189); the longest is 293.
7 of the 10 post titles are over 60 characters; the longest is 76.
Why it matters
Google trims long titles and descriptions in results, often mid-sentence, and assistants that quote them do the same.
What to do
Rewrite the 15 descriptions to under 155 characters with the answer first, and shorten the 7 post title tags.
Effort: Two hours
MediumF7· AI and agents
Half the pages carry no structured data
Stops a task: noMany affected: noKeeps happening: yes
What I found
16 of 32 pages have no JSON-LD: Work, About, Skills, Contact and all 12 case studies. Every block that does exist parses.
Why it matters
The case studies are the evidence behind the services. Without structured data, nothing tells an assistant that each one is work by the same person and business.
What to do
Add ProfilePage to About, ContactPage to Contact, and a CreativeWork to each case study, all pointing to the existing #richard and #business entities.
Effort: Half a day
MediumF8· Images
Images are heavier than they need to be
Stops a task: noMany affected: noKeeps happening: yes
What I found
152 images, 12.7 MB in total. Five case studies and the Work page carry 1.0 to 2.0 MB of images each; WEC is the heaviest at 1.96 MB.
Journal covers are 2,000px wide and fill a 364px slot on a phone; Lighthouse puts the saving at 86 KB per cover. The covers are also lazy-loaded although they're in the first screen.
Why it matters
Every extra kilobyte is a wait on a phone signal, and a lazy-loaded image in the first screen is fetched late on purpose.
What to do
Serve 800, 1,200 and 2,000px versions with srcset and sizes, and drop loading="lazy" from any image in the first screen.
Effort: Two hours, scripted
MediumF9· Accessibility
Footer headings skip a level on every page
Stops a task: noMany affected: noKeeps happening: yes
What I found
Every page jumps from the last h2 to the footer's h4 headings ("Site" and "Contact"). Lighthouse flags heading order on all six pages it tested.
Why it matters
Screen readers and agents use the heading outline to understand a page. A skipped level suggests a missing section.
What to do
Make the footer headings h2, styled as they are now, or plain labels.
Effort: Fifteen minutes
MediumF10· Accessibility
Six links are under the 24px tap target
Stops a task: noMany affected: noKeeps happening: yes
What I found
The six "Open on its own" links on the betCollab case study measure 134 × 21px. Every other control on the site meets 24px or is an inline text link.
Why it matters
Small targets are hard to tap, and an agent working from a screenshot may not register them as controls.
What to do
Add vertical padding so the links are at least 24px tall.
Effort: Ten minutes
MediumF11· Security
No Content Security Policy
Stops a task: noMany affected: noKeeps happening: yes
What I found
Responses send HSTS, nosniff, a referrer policy and frame protection, but no Content-Security-Policy header. Competitor A sends one.
Why it matters
A policy stops injected scripts running if anything on the page is ever compromised, such as a third-party script.
What to do
Add a CSP in the hosting config that allows this site, the two font services and Google Analytics, in report-only mode for a week first.
Effort: An hour
MediumF12· Search
One address takes two redirects
Stops a task: noMany affected: noKeeps happening: yes
What I found
http://rgleeson.com redirects to https://rgleeson.com, then to https://www.rgleeson.com (two 308s). The other two variants take one.
Why it matters
Each hop adds a round trip and a chance for a crawler to give up.
What to do
Point the http apex straight at https://www in the hosting settings.
Effort: Ten minutes
11
Action plan
Every finding and recommendation, in the order I'd do them. Most of the first week is under an hour each.
You're not in the top 20 for "web design Dover" Check Search Console for the impressions Web Services gets for Dover searches, so you start from real numbers. Then: ask your website clients for Google reviews; publish the Action Carpets and Downtogroundworks case studies with their towns named; list the business on the local directories that appear in these results (FreeIndex, Kent Small Business, Business Magnet); and link to Web Services from the home page's first screen.
Two to four hours to start; reviews and links build over months
Your Google listing's name is missing from your site, and another business owns it Pick one public name. If it stays RG Web Design, show "Trading as RG Web Design" on every page, add it as alternateName in the structured data and add the Google profile to sameAs.
Pages take 3.5 to 5.4 seconds to show their main content on a phone Self-host the four font files and preload the two used above the fold; show the first screen without the reveal animation; load Analytics after the page has painted, or switch to a lighter tool. Re-run Lighthouse to confirm under 2.5s.
There's no rating or review count anywhere on Web Services Ask the clients on the Work page for a Google review. Once there are five or more, show the rating and count under the Web Services headline and add AggregateRating to the business's structured data.
An AI agent can't book a call with you Turn on bookings in the Google Business Profile, or publish a booking page, and link it from Web Services, Contact and the structured data.
Half the pages carry no structured data Add ProfilePage to About, ContactPage to Contact, and a CreativeWork to each case study, all pointing to the existing #richard and #business entities.
Images are heavier than they need to be Serve 800, 1,200 and 2,000px versions with srcset and sizes, and drop loading="lazy" from any image in the first screen.
No Content Security Policy Add a CSP in the hosting config that allows this site, the two font services and Google Analytics, in report-only mode for a week first.
An hour
Medium
Design polish4 items
Item
What to do
Effort
R1
Most of the text is grey, not ink Keep grey for supporting text, but set the opening paragraph and card descriptions in ink. Contrast between levels is what makes the hierarchy, and right now most text sits on the second level.
An hour
R2
14 type sizes on one page Fold the stray sizes into the nearest step of a 9-step scale (13, 15, 16, 20, 24, 32, 44, 64, 80) and name each as a token, so new pages can't add more.
Two hours
R3
No proof in the first screen Add one line of proof under the headline: a client result ("Vancat: sales up 200%") or, once there are enough, the Google rating and review count.
An hour
R4
Content waits for an animation Show the first screen without any reveal, and keep the fade for content further down. This is part of finding F3.
Part of F3
Then re-measure
Re-run the same scripts a month after the fixes and compare. The scores, positions and speed figures in this report are the baseline.
12
Method and scoring
Every number in this report comes from a script or a dated test, and every score comes from the rules below, applied the same way to all five sites. Lighthouse 12.2.1; Chrome at 375 and 1440px; measured 4 November 2026.
What was measured
Measurement
Tool
Scope
Raw vs rendered words, crawler fetches, robots, llms.txt, landmarks, names, labels, tap targets, focus
The same measurements, the same benchmark against your local competitors (named, in your copy), the same report. £150, credited in full if you go on to a build. Email richardgleeson83@gmail.com or ring +44 7823 77 88 95.