We checked 50 managed service providers in 10 UK towns against 99 things buyers, Google and AI assistants look at. The better websites don't rank higher on Google. They are the ones AI recommends.
The median MSP website scores 39 out of 100. Most are easy to contact and hard to tell apart. Buyers still can't answer the three questions they arrive with: what will it cost, have you done this for a business like mine, and what happens when I get in touch?
Google doesn't reward the better websites. Firms that rank and firms that don't score almost the same on website quality: medians of 33.5 and 31 out of 70. Three of the ten best websites belong to firms that don't show up on Google's first page for their town.
AI does. The firms that ChatGPT, Gemini or Google's AI Overview recommended have clearly better websites: 35 out of 70, against 27 for the rest. Their sites say more plainly who they serve, show more proof and make it easier to get in touch. AI answers also lean on what buyers look for: 335 of 360 ChatGPT and Gemini answers mention Cyber Essentials, and 330 mention response times.
Three changes matter most. Say who you serve and how you charge. Prove it with a result and the accreditations buyers and AI look for. Give people one specific next step and say how fast you reply.
Every dot is one website. Firms that rank on Google (blue) and firms that don't (slate) overlap almost completely.
| Group | Websites | Scores (out of 100) |
|---|---|---|
| Visible in search (25) | 25 | 24.2, 27.5, 29, 29.5, 31, 35.8, 36.8, 38.2, 38.5, 38.5, 38.9, 39.5, 41.2, 44.5, 44.7, 45.9, 47.9, 49.6, 50.2, 50.4, 51.5, 53.3, 54.7, 60.5, 61.4 |
| Less visible (23) | 23 | 18, 20, 26, 26, 27.5, 27.5, 28, 28, 28.5, 29, 31, 35, 35, 36.3, 41.2, 43.5, 44.5, 44.6, 47, 49.3, 54.7, 59.3, 60.5 |
| All websites: median | 48 | 39 |
Search for most of these firms by name on a phone and Google shows their business card, but 23 of the 50 get no normal link to their website. Their name is shared with something else, or their homepage doesn't say clearly who they are.
Reviews are thin. 14 firms have no Business Profile we could find, only 8 have 25 or more reviews, and half the firms with reviews got one or none in the past year. Star ratings don't help buyers choose: 24 of the 36 profiles show a perfect 5.0. Directories take a quarter of the top-10 places for IT support searches in these towns.
Own-name searches on a phone in the firm's town; Business Profile and reviews from Google.
| Measure | Count (of 50) |
|---|---|
| No normal result for their own name | 23 of 50 |
| No Google Business Profile found | 14 of 50 |
| A Google review in the last 60 days | 11 of 50 |
| 25 or more Google reviews | 8 of 50 |
“A referral only works if the buyer can check you out. For a lot of these firms, a search for their own name turns up very little and a handful of old reviews. It's the cheapest gap in the study to close.”
We asked ChatGPT, Gemini and Google's AI Overview six questions a buyer would ask, such as "best IT support in Leeds", three times each in every town. 24 of the 50 firms were named in at least two of three runs. Gemini and AI Overview, which both draw on Google search, mostly named firms Google already shows. ChatGPT didn't.
What goes with being recommended is the website. Named firms score 35 out of 70 on website quality, against 27 for the rest, and the gap holds among firms that don't rank on Google. They're clearer about who they serve, show more proof and are easier to contact. AI answers mention Cyber Essentials constantly, in 335 of 360. Firms showing it were named more often, 17 of 26 against 6 of 22, but much of that gap goes once you allow for Google visibility. These things go together; we don't claim one causes the other.
The answers aren't stable. Four times in ten that one of our firms was named, it didn't come up again in the other two runs.
Six buyer questions per town, asked three times each. Named = in at least 2 of 3 runs.
| AI tool | Visible in search (25) | Less visible (25) |
|---|---|---|
| Any AI tool (all firms) | 24 of 50 | |
| Gemini | 12 of 25 | 4 of 25 |
| Google AI Overview | 7 of 25 | 1 of 25 |
| ChatGPT | 8 of 25 | 7 of 25 |
Answers mentioning each term, out of 360.
| Measure | Count (of 360) |
|---|---|
| Microsoft | 336 of 360 |
| Cyber Essentials | 335 of 360 |
| Response times | 330 of 360 |
| Being local | 325 of 360 |
| ISO certification | 281 of 360 |
| 24/7 support | 242 of 360 |
“The firms AI recommended had clearer websites, and that held for firms that don't rank on Google too. That's good news for smaller MSPs: it's work you can do on your own site.”
27 of 48 websites use two or more stock phrases on their main pages, such as "trusted IT partner", "peace of mind" or "tailored solutions". 15 use four or more. Only 3 name the sector they serve in the headline, and only 13 name one anywhere on the homepage.
The firms AI recommends score higher here too: a median of 8 out of 20 for clarity, against 6 for the rest.
Homepage headline and opening copy, read by AI and checked against its quoted evidence.
| Measure | Count (of 48) |
|---|---|
| Use 2 or more stock phrases | 27 of 48 |
| Say what size of business they serve | 31 of 48 |
| State an outcome in the headline | 20 of 48 |
| Name a sector anywhere | 13 of 48 |
| Name a sector in the headline | 3 of 48 |
“Most MSP homepages say the same few things in the same words. If a buyer could swap your logo for a competitor's and nothing else would change, the page isn't doing its job. Say who you work with and what changes for them.”
27 of 48 websites show nothing about price, not even whether they charge per user or per device. 17 explain how they charge without figures. Only 4 publish a figure for their support, from £25 to £58 per user a month.
Terms are rarer still. 19 give a specific response time, 19 explain onboarding, and only 7 say anything about contract length or notice.
Each website, by the most detailed pricing it shows; then what else it explains.
| Measure | Websites (of 48) |
|---|---|
| Pricing shown: Exact prices | 2 of 48 |
| Pricing shown: A starting price | 2 of 48 |
| Pricing shown: How they charge, no figures | 17 of 48 |
| Pricing shown: Nothing | 27 of 48 |
| Give a specific response time | 19 of 48 |
| Explain onboarding or switching | 19 of 48 |
| State contract length or notice | 7 of 48 |
“Buyers want to know whether you're in their budget before they pick up the phone. You don't need a price list: explaining how you charge, and what a typical client pays, answers the question most firms leave hanging.”
Badges are easy to find: 34 of 48 sites show a vendor partnership and 26 show Cyber Essentials. Evidence that it worked for someone is rare. 20 have a case study, and only 8 have one with a measurable result.
Proof goes with AI recommendations: named firms score a median of 7 out of 15 on trust, against 4 for the rest.
Websites showing each kind of proof.
| Measure | Count (of 48) |
|---|---|
| Vendor partner badge | 34 of 48 |
| Cyber Essentials (or Plus) | 26 of 48 |
| Named testimonials | 23 of 48 |
| Client logos | 22 of 48 |
| Real team photos | 21 of 48 |
| A case study | 20 of 48 |
| ISO 27001 | 9 of 48 |
| A case study with a measurable result | 8 of 48 |
“Badges tell a buyer you're qualified. They don't tell them you're good. One short case study with a real result does more than another vendor logo.”
17 of 48 homepages lead with "contact us" or "get in touch", and on 13 the main button is a phone number, an email address or missing. 23 offer something specific: a call, a free assessment or a quote.
Few say what happens next. 8 let visitors book a time online and 9 promise a reply time. 9 sites have no contact form we could find.
Main homepage call to action, and fields on the main contact form.
| Measure | Websites (of 48) |
|---|---|
| Main call to action: "Contact us" or "Get in touch" | 17 of 48 |
| Main call to action: Book a call | 11 of 48 |
| Main call to action: Free assessment or audit | 10 of 48 |
| Main call to action: Something else, or none | 8 of 48 |
| Main call to action: Get a quote | 2 of 48 |
| Contact form fields: 3-4 | 9 of 48 |
| Contact form fields: 5 | 12 of 48 |
| Contact form fields: 6-7 | 16 of 48 |
| Contact form fields: 8-9 | 2 of 48 |
| Contact form fields: none | 9 of 48 |
“'Contact us' asks the buyer to do the work. The best sites say exactly what happens next and how quickly. It's a small change with a direct line to enquiries.”
Every site uses HTTPS, but the next layer is patchy. 18 of 48 don't enforce DMARC, so email pretending to come from them can still reach inboxes. 19 run a plugin or script with known security issues, and 26 show which version of their CMS they run.
Only 9 send a Content-Security-Policy, and 28 of 48 get an E or F grade on security headers overall.
Passive checks only: public headers, certificate, DNS and the versions a page reveals. Anything serious is shared privately with the firm concerned.
| Measure | Count (of 48) |
|---|---|
| HTTPS with a valid certificate | 48 of 48 |
| DMARC set to quarantine or reject | 30 of 48 |
| No plugin or library with known issues | 29 of 48 |
| Strict-Transport-Security (HSTS) | 22 of 48 |
| CMS version not on show | 22 of 48 |
| Frame protection | 17 of 48 |
| Content-Security-Policy | 9 of 48 |
| security.txt published | 2 of 48 |
“Buyers can't audit your network, but they can see your website and your email set-up. For a firm that sells security, getting your own basics right is part of the pitch.”
The median homepage takes 6.2 seconds to show its main content on a mid-range phone, and weighs 2.8 MB. 38 of 48 take over four seconds.
31 of 48 load trackers before the visitor agrees to cookies, and 12 of those show no cookie banner at all. Analytics is often broken too: 8 still run Universal Analytics code Google switched off in 2024, and 19 have no GA4.
Lab speed tests on a mid-range phone (too few visitors for Google's real-user data); automated accessibility checks on key pages; cookie checks from a fresh browser.
| Measure | Websites (of 48) |
|---|---|
| take over 4 seconds to show their main content on a phone | 38 of 48 |
| have 3 or more serious accessibility issues | 27 of 48 |
| load trackers before cookie consent | 31 of 48 |
| still run Google Analytics code that stopped working in 2024 | 8 of 48 |
“Most of these sites are slow on a phone and load tracking before anyone has agreed to it. Neither is hard to fix, and both are things a careful buyer notices.”
The ten best websites aren't the ten best-ranked firms: three of them don't appear on Google's first page for their town. What they share is clarity, a next step and the proof AI looks for. Nine of the ten were recommended by an AI tool. We only name firms that agree.
| Trait | Top 10 | Other 38 |
|---|---|---|
| Recommended by an AI tool | 9 /10 | 14 /38 |
| Say who they serve (sector or business size) | 9 /10 | 25 /38 |
| A specific next step (call, assessment, quote) | 8 /10 | 15 /38 |
| Publish pricing or how they charge | 7 /10 | 14 /38 |
| Show Cyber Essentials | 7 /10 | 19 /38 |
| Visible on Google | 7 /10 | 18 /38 |
These traits go together with higher scores; with 48 websites we don't claim they cause them.
Enter your website. NetNav scans it with the same checks and shows your score against the benchmark, pillar by pillar.
25 that rank on Google's first page for IT support in their town and 25 that don't, 5 from each of 10 UK towns, drawn at random from Google search and Maps results.
Google, Business Profile, AI, security, speed, accessibility and cookie checks. All passive: public pages only, one page a second, nothing submitted.
One AI model read each site's screenshots and pages for pricing, proof, positioning and calls to action, quoting its evidence. We re-checked every borderline answer against that evidence and corrected 25.
Comments on what the findings mean are labelled as views. They sit beside the data and never change it.
Limits. 50 sites can't describe every UK MSP. We report counts, use medians, avoid small sub-groups and don't claim cause and effect. Data collected 9 November 2026. Two Belfast sites showed a maintenance page or blocked our crawler, so website counts are out of 48.
What we think the findings mean for an MSP planning its website. Marketers at MSPs outside the sample will add their views here.
“Explaining how you charge is enough for most buyers. Silence on price sends them to the firm that answers.”
“AI answers change from run to run. A clear website and accurate directory listings are what you can control.”
“The firms that stand out say who they serve. That one line does more than any slogan.”
“Your own website is the one piece of your security a buyer can see. It should pass.”
Each piece below answers one question MSP owners and marketers ask, using this data as evidence.
NetNav Research (2026). The MSP Website Benchmark: 50 MSP websites in 10 UK towns. netnav.io/research/msp-websites
Charts may be reused with a link to this page. Press and partners: ask us for the chart pack or the anonymised dataset.