I Audited 24 Small Business Websites. Their SEO Was Fine — Their Speed Was Not.

I kept telling clients that small business websites are slow. Then I realised I was repeating something I’d absorbed rather than something I’d measured — so I went and measured it.

I audited 24 real small business websites: dental practices, plumbers, hair salons, family law firms, independent bakeries and auto repair garages, across the US, UK, Canada and Australia. Every number below comes from those scans. Nothing is estimated, and nothing is rounded up to make a point.

The result surprised me, and it changed what I tell people to fix first.

How I did it (and where this study is weak)

I found businesses through ordinary Google searches — the kind a customer would actually do — in a handful of cities and trades. I kept any business with its own domain and skipped directories, aggregators and national chains. Then I ran each homepage through my free Website Report Card, which uses Google’s PageSpeed Insights data, on a simulated mid-range mobile phone.

25 sites went in; 24 returned complete results. One failed to return data from Google’s API across repeated attempts and was dropped rather than guessed at.

The honest limitations: 24 is a small sample, and it isn’t random — these are businesses that rank well enough to appear in search results, which if anything biases the sample toward better-maintained websites. Six trades across four countries isn’t the world. Treat this as a well-documented snapshot, not a census. I’m publishing the method so you can disagree with it.

I’m also not naming any of the businesses. They didn’t volunteer for this, and a public list of failing scores attached to real small companies would be a mean thing to do for content. The patterns are the point.

The headline: SEO fine, speed broken

Here’s the median score in each category across all 24 sites:

Category	Median score
SEO	92
Security & best practices	83
Mobile-friendliness	70
Performance (speed)	62

I expected SEO to be the disaster. It wasn’t even close to one. The lowest SEO score in the entire set was 82. Twenty-three of 24 sites had a proper meta description. Almost all of them had titles, crawlable links and a sensible page structure.

Speed was a different story:

• 22 of 23 sites (96%) failed “content appears quickly” — Google’s Largest Contentful Paint check.
• 18 of 24 (75%) were too slow on mobile.
• 9 of 24 (38%) scored under 50 out of 100 on performance.
• Exactly one site out of 24 scored 90+ on performance.

The slowest site took 35 seconds to show its main content on a mid-range phone. Second slowest: 28.5 seconds. Third: 20.6 seconds. These are not typos, and they are not obscure businesses — they’re normal companies with normal websites that someone was presumably paid to build.

Why the split makes complete sense

Once I stopped being surprised, the reason was obvious.

SEO basics are a checklist, and checklists get done. Title tag, meta description, alt text, sitemap — these appear in every “SEO for small business” article, every website builder nags you about them, and every agency’s onboarding doc includes them. They’re visible, they’re one-time, and someone ticks them off.

Speed is not a checklist. It’s a consequence. It’s the accumulated result of the theme you chose, the plugins you installed, the 4MB hero image nobody compressed, the three tracking scripts marketing added in 2023, and the page builder rendering it all. No single person decided to make the site slow. It happened to them, gradually, and nothing in their workflow ever measured it.

That’s also why it doesn’t get fixed: nobody owns it, and on the owner’s own laptop — on fast office broadband, with the site already cached — it feels fine.

An uncomfortable note about my own scoring

The median overall score was 74. That’s a “solid foundation, room to grow” result, and it’s noticeably kinder than the underlying speed numbers deserve.

The site with the 35-second load still came out at 52 overall, because near-perfect SEO and security scores pulled the average up. My scoring weights performance at 35%, and blending a catastrophe with three passes produces a mediocre number rather than an alarming one.

I’m pointing this out because it applies to every scoring tool, including mine: the composite number is the least useful thing on the report. The category breakdown is where the truth lives. If you run your own site, read the four scores separately and ignore the big number at the top. (I’ve written a fuller guide to what actually counts as a good website score, including which numbers you can safely ignore.)

The other things that were quietly missing

Beyond speed, a few gaps showed up often enough to mention:

• 14 of 24 (58%) didn’t send a basic X-Content-Type-Options security header. Low stakes, but it’s a one-line server config that signals nobody has looked at the hosting setup.
• 9 of 24 (38%) had no structured data. This is what lets Google show star ratings, opening hours and prices directly in results — a real, visible loss for local businesses.
• 9 of 24 (38%) had no Open Graph tags, meaning every time someone shares the business on Facebook, WhatsApp or LinkedIn, it appears as a bare grey link instead of a proper preview card.
• 7 of 24 (29%) had no sitemap.
• 3 of 24 (13%) had no responsive viewport tag at all — on a phone, those sites render as a shrunken desktop page you have to pinch to read.
• 2 of 24 (8%) were still served over plain HTTP. Chrome marks those “Not secure” in the address bar, next to the name of a business asking people to book appointments.

The Open Graph one is my pick for “worst value-to-effort ratio being ignored.” It’s about fifteen minutes of work, and it affects every single share of that business, forever.

If your site is in this group, here’s the order I’d fix things

Based on what actually went wrong across these 24 sites:

1. Start with your images. My audit measured how long content took to appear, not which file caused it — so I won’t pretend this data identifies the culprit. But in the sites I’ve actually opened up and fixed over the years, oversized hero images are the most common single cause, and modern formats (WebP/AVIF) at correct dimensions routinely cut load time substantially with zero design change. It’s the cheapest thing to check first.
2. Audit your plugins and scripts. Every plugin is code that runs before your customer sees anything. If you can’t name what a script does, it’s a candidate for removal.
3. Add Open Graph tags and structured data. Cheap, fast, and immediately visible in search results and social shares.
4. Get on HTTPS if you somehow aren’t. It’s free. There is no remaining argument.
5. Only then consider a rebuild. Plenty of slow sites are fixable. Don’t let anyone — including me — sell you a rebuild before they’ve shown you what the current site would score after a proper cleanup.

The part that actually matters

None of these businesses have a bad website because they’re careless. They have a bad website because nobody ever told them, and because the failure is invisible from the inside. Your site feels fast to you. It feels fast to the person who built it. It only feels slow to the customer on a phone, on mobile data, in a car park — who leaves, and never tells you why.

That’s the whole reason I built the free tool: not because the world needed another grader, but because “your site is slow” is a claim, and a number from Google is evidence.

If you want to know where your own site sits against these 24, it takes about a minute and no email address: run your free Website Report Card.

And if the result is ugly — genuinely, don’t panic. Every single issue in this article is fixable, most of them in an afternoon. If you’d rather someone just did it, message me on Fiverr and I’ll tell you honestly what your site needs, including when the answer is “less than you think.”

---

Methodology notes: all scans run on Google PageSpeed Insights data, mobile strategy, on 21 July 2026. Scores reflect a single point in time and will shift as sites change. If you’re one of the businesses in this set and want to know which line is yours, email me — I’ll tell you privately, and I’ll tell you how to fix it for free.