Websites That Convert vs Websites That Look Nice: The Difference

A founder in Cork sent me her new website last month. Agency-built, EUR 9,000, beautiful typography, a gorgeous hero video of a drone flying over the Burren. "We launched six weeks ago and we have had one enquiry. Is that normal?" I opened Chrome DevTools. Largest Contentful Paint on mobile: 6.8 seconds. Hero CTA: the word "Explore" in grey text on a grey background. Above the fold on mobile: no mention of what the company actually did. She had bought a brochure, not a website. She is not alone — this is the single most common pattern I see in SMB web work, and it is almost always fixable without starting over.

The uncomfortable benchmark — average SMB conversion rate is 1.8%

A working website for a B2B SMB converts somewhere between 2% and 5% of visitors into qualified leads, depending on the vertical. The industry median sits around 1.8% (WordStream's 2024 cross-industry benchmark; our own client data tracks closely). Anything below 1% is broken. Anything above 5% is either very good or very niche.

The gap between "converts at 1%" and "converts at 4%" is not usually design quality — most sites in both buckets look fine. The gap is structural. It is the difference between a page that answers the buyer's question in the first five seconds and a page that asks them to "scroll to learn more".

Eight things every converting website does

Not five, not ten — eight, because that is what I keep finding when I audit. And I will be honest: the first three matter more than the last five combined.

Above-the-fold clarity (the 5-second test)

Show your homepage to someone who has never heard of your company. Count to five. Ask them: what does this business do, who is it for, and what should they do next? If they cannot answer all three, the page is broken. Most SMB sites fail this. A surprising number of agency-built sites fail this, because agencies love hero headlines that read as poetry ("Crafting tomorrow, today") and bury the actual value proposition under scroll.

Our own homepage opens with the literal sentence "Websites That Convert. Apps That Scale." followed by where we operate. Not clever. Deliberately not clever.

One primary CTA per page

Every page needs one obvious next step. Not three. Not "Get started", "Book a demo", "Talk to sales", "Download the guide", "Subscribe to our newsletter" all stacked together. One. The Von Restorff effect is a real thing — the brain picks out the odd one — so when every button looks equally important, none of them feel important. Pick the action that matters most for that page and make every other CTA visually subordinate.

Proof that is not just logos

A row of client logos is the weakest form of social proof still in circulation. Buyers discount them — they assume the logos are cherry-picked or fudged. What actually moves conversions: named case studies with numbers ("we rebuilt X's site, organic traffic grew 62% in six months"), testimonials from identifiable humans with titles and company names, review aggregates from a third party (Google, Clutch, G2). If all you have is a logo wall, your proof section is near-useless.

Mobile parity

This should not need saying in 2026 and somehow still does. More than 60% of SMB website traffic is mobile for most verticals. Sites still ship with desktop-first design where mobile is a pinched, laggy afterthought. Every page we build gets designed mobile-first — literally in Figma on a 375px artboard before a 1440px one. The difference in the final product is not subtle.

Page speed that does not punish the visitor

Google's Core Web Vitals thresholds: Largest Contentful Paint under 2.5 seconds, Interaction to Next Paint under 200ms, Cumulative Layout Shift under 0.1. Hit all three on mobile, on a 4G connection, and you are in the top quartile of the web. Miss any of them consistently and you are bleeding conversions you never see — Google's own data shows a rough 20% drop in conversion rate for every additional second of LCP above the threshold.

Copy that sounds like a human wrote it for another human

Buyers skim. They skip blocks that read like marketing. Short sentences. Active voice. Specific claims with numbers where possible. If your page reads like it could belong to any of your three closest competitors, it is not converting — it is noise.

A contact path that takes less than 30 seconds

The highest-intent visitor on your site is someone who has decided to reach out. Reward that intent by making it effortless. Contact form with three fields (name, email, message), phone number visible on every page, and a calendar booking link. Nothing clever. The friction people add to contact forms in the name of "qualifying leads" is almost always self-defeating — a qualified lead who has to fill in twelve fields becomes an unqualified lead who closed the tab.

A pricing page that exists and is specific

This is the single most contentious one. A lot of agencies will tell you to hide pricing to "get them on a call". In our experience, that strategy suits the agency, not the client. Prospects use your pricing page to self-qualify. If they cannot, they move on — and the ones who move on are disproportionately the good-fit ones, because the bad-fit prospects are willing to endure the discovery-call dance. Our pricing page shows concrete numbers (web and app from EUR 1,300, hosting from EUR 59/month) because we would rather lose a bad-fit lead in ten seconds on the page than in an hour on a Zoom call.

Five things that silently kill conversion

Carousels, auto-playing video, hamburger menus on desktop

Carousels: clicked by roughly 1% of visitors even when the first slide converts well. Every slide after the first converts worse than a static image would have. Stop using them.

Auto-playing video with sound: universally loathed. Auto-playing video without sound: acceptable, but only if it is under 500KB and loads lazy. Most sites ship an 8MB hero video that tanks mobile LCP.

Hamburger menus on desktop: users expect a horizontal nav. When you hide everything behind a three-line icon on a 1440px screen, you are making them do more work for no reason. Put the five or six important links in the header.

Generic stock photography

A row of smiling model-people in a meeting room. Buyers identify stock photography instantly now and it subtly signals "this company is not serious enough to commission its own photos". Either use real photography of your team, your work, your office — or use no photography at all and lean on typography and illustration. Do not use stock photos of people. Abstract stock images (landscapes, textures) are fine.

Trust signals that are not trustworthy

"100% satisfaction guarantee!" with no specifics. A padlock icon in the footer that is not a link. "Trusted by thousands of happy customers" with no named ones. These read as marketing theatre and modern buyers are allergic to them. Either have specific proof or have none — specific silence is more trustworthy than vague bragging.

How Core Web Vitals quietly compound the problem

I want to come back to performance because it is the invisible tax most SMB sites are paying. LCP on mobile is the number I look at first on any audit. If it is over four seconds, everything else the site does right is being undone. The hero image is unoptimised, the font is loading blocking, the JavaScript bundle is 2MB of unused React component library. The fix is rarely glamorous — compressing images, deferring JavaScript, preloading the hero font, moving to Hetzner or Cloudflare so the first byte arrives quickly — but the conversion lift is immediate and measurable within weeks.

One practical number: we migrated a client off shared hosting onto our Hetzner-backed managed hosting at EUR 59/month last November. LCP dropped from 4.1s to 1.6s. Form submissions the following month rose 28%. Nothing else changed on the site. That is the compounding tax, paid down.

A 45-minute self-audit you can run today

Open your homepage on your phone, on 4G (toggle off Wi-Fi). Time how long until the main content appears. Show it to a non-technical friend and run the 5-second test. Click every CTA and count how many steps to actually buy or book. Open DevTools, Lighthouse tab, run a mobile audit, write down the Performance score and the LCP number. Read your homepage copy aloud — if you stumble or feel embarrassed, your visitors are too.

That is it. Forty-five minutes of work and you will know whether you have a converting site or a nice-looking one.

When to rebuild vs when to optimise

Rebuild if: the site is over four years old, built on a no-code tool you cannot extend, has an LCP above 5 seconds that cannot be fixed without restructuring, or is missing mobile parity entirely. Optimise if: the design is fine, the copy needs work, the CTAs are unclear, the performance is fixable. Most sites I audit are in the second bucket — EUR 1,500 of targeted work, not a EUR 10,000 rebuild.

One thing I will push back on: the instinct to rebuild every three years "because it looks dated". That is usually vanity. If it converts, leave it alone and spend the money on content or ads. A site that looks 2022 and converts at 4% is worth more than a site that looks 2026 and converts at 1.5%.

Where to start

If you want an outside pair of eyes on yours, we offer a free 10-point website audit — we run the full audit above and send a PDF report within 48 hours. If the report says "rebuild", our web and app development service starts at EUR 1,300. If it says "optimise", that is usually cheaper. We will be honest about which.