Two weeks ago, I did something that started as curiosity and ended as mild horror.
I collected 100 small business websites from across North America. Plumbers, restaurants, dental offices, boutique retailers, law firms, yoga studios, landscapers, accounting practices. Real businesses with real customers and real revenue. Not agencies. Not tech startups. The kind of businesses that make up, let's say, 80% of the actual economy but get about 2% of the attention in SEO conversations.
I ran every single one of them through Licheo's AI-powered audit. Full analysis -- technical SEO, on-page optimization, performance, security, content quality, GEO readiness. The whole thing.
What came back was... not great. And the patterns were so consistent that I stopped being surprised by website number 40 and started being worried by website number 70.
Here is everything I found.
How I selected the 100 websites
I want to be transparent about methodology because the data only means something if the sample is honest.
I pulled from Google Maps listings in 14 different cities across the US and Canada. Not the top results -- I went to pages two and three because that is where most small businesses actually live. I targeted businesses with between 2 and 50 employees, which I estimated based on their About pages and LinkedIn profiles. I excluded any business that was clearly part of a franchise or national chain.
The industry breakdown ended up being: 22 home services (contractors, plumbers, electricians, HVAC), 19 restaurants and food businesses, 17 retail shops, 16 professional services (lawyers, accountants, consultants), 11 health and wellness (dentists, chiropractors, yoga studios), 8 auto services, and 7 that did not fit neatly into a category.
Each audit took Licheo about four hours to complete. I ran them in batches over 12 days. The total cost was something like $40 in AI processing, which tells you something about how accessible this kind of analysis has become.
The aggregate numbers are brutal
The average overall score across all 100 websites was 47 out of 100.
I sat with that number for a while. Forty-seven. If you scored 47% on an exam, you would have failed. And these are not obscure businesses running hobby websites. These are companies that depend on their online presence to attract customers, that probably paid someone to build their site, and that are competing against each other for the same local customers every single day.
The median was even lower at 43, which means a handful of decent websites pulled the average up. The highest score in the entire sample was 78. Only one. The lowest was 11 -- an HVAC company in Phoenix whose website appeared to have been built around 2014 and never touched again.
But the aggregate score is just the starting point. The specific findings are where things get truly interesting.
81% had missing or duplicate meta descriptions
This was the single most common problem across the entire sample. Eighty-one of the 100 websites had either missing meta descriptions on their main pages, duplicate descriptions copied across multiple pages, or descriptions so generic they might as well have been missing.
The truth is, I expected this number to be high, but not this high. Meta descriptions are one of the most fundamental signals you send to search engines about what a page contains. They appear directly in search results. They influence whether someone clicks on your listing or the one above it.
And yet, a roofing company in Atlanta had the exact same meta description on every single page of their 34-page website: "Welcome to [company name], your trusted roofing professionals." A restaurant in Portland had no meta descriptions at all -- not one. A law firm in Denver had meta descriptions that were clearly auto-generated by their website template, all reading "Page -- [Company Name]" with nothing else.
What makes this particularly frustrating is that meta descriptions are among the easiest things to fix. You do not need a developer. You do not need expensive tools. You need 20 minutes and someone who understands what each page is actually about.
73% had no analytics installed
This one genuinely shocked me.
Seventy-three of the 100 websites had no detectable analytics tracking. No Google Analytics. No Google Tag Manager. No Clarity. No Matomo. Nothing. These businesses have absolutely zero visibility into who visits their website, where they come from, what they look at, or whether their website is actually generating any business at all.
Imagine running a physical store where you could not see the customers walking in. You would not know if your new sign was working. You would not know if your window display was attracting people. You would not know if customers were picking up products and putting them back. That is what operating a website without analytics is like -- you are flying completely blind.
Of the 27 that did have analytics, 9 were running Universal Analytics, which Google stopped processing over two years ago. So really, only 18 out of 100 websites had functioning, current analytics. Eighteen percent.
I keep coming back to this number because it explains so much about why small businesses struggle with their websites. They literally cannot see what is happening. Every decision about their site -- whether to redesign it, whether to add a page, whether to change the homepage -- is being made based on gut feeling rather than evidence. And gut feeling, alla fine, is usually wrong.
89% had no GEO optimization at all
Generative Engine Optimization -- making your website visible to AI search tools like ChatGPT, Perplexity, Google's AI Overviews -- was essentially nonexistent in the sample. Eighty-nine websites showed zero preparation for how AI systems discover, interpret, and cite content.
Now, I understand that GEO is still relatively new and many business owners have not even heard the term. But AI search is not coming. It is here. According to recent data, about 30% of all search queries in the US now trigger some form of AI-generated response. And that number goes up every month.
What does "no GEO optimization" look like in practice? It means content that is not structured in a way AI can extract clean answers from. It means no FAQ sections that AI systems love to pull from. It means no clear entity identification -- the AI does not understand what you do, where you do it, or why someone should choose you. It means missing structured data that AI systems rely on to build their understanding of your business.
The 11 websites that showed some GEO readiness were almost all professional services firms, specifically law firms and accounting practices. My guess is that these businesses tend to hire more sophisticated marketing help, or they have content-heavy websites that naturally perform better with AI systems. But even among these 11, the optimization was partial at best.
62% had no schema markup
Schema markup is how you speak to search engines in their own language. It tells Google explicitly that this is a local business, these are the hours, this is the address, these are the services offered, and these are the reviews.
Sixty-two websites had none at all. Not a single line of structured data.
Of the 38 that had some schema, 14 had only basic website or organization schema that their template or website builder had automatically included -- the bare minimum. Only 8 of the 100 had what I would call genuinely useful schema: LocalBusiness markup with services, hours, geo-coordinates, and review aggregation.
This matters more than most people realize. When someone searches "plumber near me" and Google shows a rich result with star ratings, hours of operation, and service areas, that business has schema markup. The business below it, showing just a blue link and two lines of text, probably does not. Which one do you think gets the click?
67% had broken links
Two-thirds of the websites had at least one broken link. The average number of broken links per site was 7, but the distribution was wildly uneven. Some sites had one or two. One flooring company in Sacramento had 43 broken links, almost all pointing to product pages for materials they presumably no longer carry.
Broken links are one of those problems that accumulate silently. A business adds a page, links to it from the homepage, then later deletes the page. Nobody updates the homepage. The link breaks. Multiply this across years of incremental changes and you get a website held together with dead ends.
From a search engine's perspective, broken links signal neglect. They tell Google that this website is not being maintained, which directly affects how much trust the algorithm places in the content. From a user's perspective, clicking a link and hitting a 404 page is about the fastest way to lose someone's confidence.
44% were on shared hosting (and it showed)
I estimated hosting type based on server response headers and load times. Forty-four of the websites appeared to be on shared hosting plans, the kind where your site lives on a server with hundreds or thousands of other websites.
The performance difference was dramatic. Shared hosting sites had an average initial server response time of 1.8 seconds. Sites on dedicated or cloud hosting averaged 0.4 seconds. That is a difference your visitors feel with every single page load.
One Italian restaurant in Chicago -- and I find this particular combination almost poetic in its irony -- had a server response time of 3.7 seconds before the first byte of content even started loading. The full page took over 8 seconds. Nobody is waiting 8 seconds to see a dinner menu. They are going to the next restaurant on the list.
Google has been clear for years that page speed is a ranking factor. With Core Web Vitals now directly influencing rankings, a slow website is not just a bad user experience. It is an active handicap in search results.
Only 12% had proper mobile optimization
This finding, bisogna dire, was the one that made me want to close my laptop and go for a walk.
We are in 2026. More than 60% of web traffic comes from mobile devices. Google has been using mobile-first indexing since 2019. And still, only 12 of these 100 websites were properly optimized for mobile.
"Properly optimized" means more than just being responsive -- more than just shrinking to fit a phone screen. It means touch targets large enough to tap without accidentally hitting three other links. It means text readable without pinching and zooming. It means forms that work on a phone. It means images that do not force horizontal scrolling.
The most common problem was tap target sizing. Forty-one websites had navigation links and buttons so close together that using them on a phone was an exercise in frustration. Twenty-eight had text that was too small to read on mobile without zooming. Nineteen had forms where the input fields were practically unusable on a phone screen.
A dental practice in Austin had a "Book Appointment" button on their homepage that was literally 12 pixels tall on mobile. Twelve pixels. You would need surgical precision to tap it. For a dental office, the irony writes itself.
The industry breakdown tells its own story
The differences between industries were, in some cases, enormous.
Professional services (lawyers, accountants, consultants) had the highest average score at 54 out of 100. Still a failing grade, but comparatively the best performers. These businesses tend to have content-heavy websites, which naturally helps with SEO even when nobody is actively optimizing. They also tend to invest more in their online presence because their clients are researching them before making contact.
Restaurants came next at 49, which surprised me. I expected them to score lower, but many restaurant websites are built on platforms like Squarespace or dedicated restaurant website builders that handle some basics automatically -- mobile responsiveness, basic meta tags, decent loading speed. The restaurants that scored poorly were the ones still running custom WordPress sites from 2016 that nobody has maintained.
Retail shops averaged 45. The main problems here were product pages with thin content, images without alt text, and broken links to discontinued products. Retail is hard because the inventory changes and the website needs to change with it, but nobody budgets for ongoing website maintenance.
Home services -- the contractors, plumbers, electricians -- averaged 41. And auto services brought up the rear at 38. These industries had the highest rates of completely missing analytics (over 80% in both cases), the worst mobile optimization, and the most outdated website designs. A pattern emerged: the more physical the business, the less attention the website gets. Which is backward, because these are precisely the businesses where customers are searching online most urgently. When your pipe bursts at 11pm, you are not browsing a directory. You are Googling.
The surprising finding: expensive websites were not the best
I tried to estimate what each website cost to build, using design complexity, custom features, and platform choice as rough indicators. Some of these sites clearly cost $15,000 or more. Custom designs. Professional photography. Elaborate animations.
And some of the most expensive-looking websites had terrible scores.
A custom-built website for a landscaping company that clearly involved professional photography, custom illustration, and probably several months of design work scored 34 out of 100. Beautiful to look at. Technically a disaster. No schema markup, no meta descriptions, no analytics, no alt text on any of those gorgeous photographs, broken contact form on mobile, and server response time over 4 seconds because the images were uncompressed 8MB files.
Meanwhile, a locksmith in Toronto using a fairly basic Squarespace template scored 61. Clean, fast, mobile-friendly. Had schema markup because the platform added it automatically. Had analytics because the owner had followed the setup wizard. Nothing fancy, but fundamentally sound.
The correlation between apparent website cost and audit score was essentially zero. If anything, it was slightly negative -- the more custom and elaborate the site, the more likely it was to have technical problems buried beneath the pretty surface.
This tells me something about how small businesses buy websites. They evaluate based on how it looks. They should be evaluating based on how it works. A beautiful website that Google cannot properly index is an expensive piece of digital art. It is not a business tool.
What the top 10% had in common
Ten websites scored 65 or above. I looked at what they shared.
Every single one had current analytics installed and functioning. That is not a coincidence. Businesses that measure their website's performance tend to improve their website's performance. It is hard to fix what you cannot see.
Nine out of ten had been updated within the last 90 days. Fresh content, new blog posts, updated service pages, recent customer testimonials. These were living websites, not static brochures. The search engines noticed.
Eight out of ten had at least basic schema markup. Not always comprehensive, but present. They were speaking Google's language.
Seven out of ten loaded in under 2 seconds. Fast hosting, optimized images, clean code.
And here is the one that I found most telling: all ten of them had unique, specific meta descriptions on their primary pages. Not template-generated garbage. Not the same description copied 30 times. Actual, intentional descriptions that described what each page was about and why someone should click.
The top performers were not doing anything revolutionary. They were doing the basics competently. That was enough to put them in the top 10% because almost nobody else was doing the basics at all.
The three issues that affected almost everyone
If I had to pick the three problems that were nearly universal -- present in 70% or more of the sample -- they would be these.
Missing or broken meta data. This includes meta descriptions, title tags that just said the company name with nothing else, and Open Graph tags that were either missing or defaulted to whatever the website template shipped with. The result is that these businesses look terrible in search results and even worse when someone shares their link on social media. Your website's first impression happens before anyone visits it, in the search result itself. And almost everyone is blowing it.
No measurement of any kind. Without analytics, without Search Console, without any form of tracking, these businesses are spending money on a website they cannot evaluate. They do not know if their website generates leads. They do not know which pages people visit. They have no way to calculate whether their website is an asset or a liability. Naturally, this makes every subsequent decision about the website a guess.
Content that exists for the business, not the customer. So many of these websites read like internal documents. Pages about "Our History" that nobody asked for. Service descriptions that list what the business does but never address what the customer needs. No FAQ pages. No helpful content. No demonstration of expertise. The content exists because someone thought a website should have content, not because they considered what a potential customer would be searching for.
What to do about it
If you own a small business and you have read this far, there is a reasonable chance your website has some of these same problems. Most do. There is no shame in it. The shame would be knowing about it and doing nothing.
The good news is that most of these issues are fixable without spending thousands of dollars. Senza dubbio, the hardest part is simply understanding what is wrong in the first place.
This is precisely why we built the free SEO standing check at licheo.com/seo-standings. You put in your URL, the AI runs the same analysis I ran on these 100 websites, and you get a scored report with specific findings and recommendations. It takes just seconds and costs you nothing.
If you want to see how your website compares to competitors in your area, the SEO standings tool will show you exactly where you rank relative to other businesses targeting the same customers.
I started this experiment thinking I would find interesting patterns. I found alarming ones instead. The gap between what small business websites should be doing and what they are actually doing is enormous. And in 2026, with AI search changing how customers find businesses, that gap is getting more expensive every month.
Forty-seven out of 100. That was the average. I genuinely hope yours is better.