What 11,432 Small Business Websites Reveal About Mobile Performance
Website performance advice usually starts with one site.
Run a speed test. Find the slow template. Compress the images. Configure caching. Remove a plugin or two. Test again.
That process makes sense when fixing an individual website. It tells us less about how common the underlying problems are.
Routeless tested 11,432 unique small business homepages across 327 U.S. cities for its State of Small Business Websites 2026 report. Of the 11,184 sites that completed a mobile PageSpeed test, the median Largest Contentful Paint was 7.9 seconds.
Google considers an LCP of 2.5 seconds or less good.
In the Routeless sample, 95.4% of completed tests came in above that threshold.
These are Lighthouse lab measurements, not field data from actual visitors, so they should not be read as a claim that 95.4% of American businesses fail Core Web Vitals. They do show something useful: under the same mobile testing methodology, slow initial rendering was extremely common.
That changes how we should think about WordPress performance. A slow website is rarely the result of one setting. It is usually the accumulated cost of decisions made across design, development, hosting, plugins, images and third-party software.
Start with what the visitor is waiting for
Largest Contentful Paint measures when the largest visible content element finishes rendering.
On a small business homepage, that element is often obvious: a hero image, banner, heading or another prominent block near the top of the page.
Finding that element should come before blindly enabling optimization settings.
If the LCP element is an image, look at its dimensions and file size. A phone should not need to download an image prepared for a 2,500-pixel desktop display. Check the format, responsive image configuration and how early the browser discovers the asset.
If the LCP element is text, the delay may come from a web font, CSS or JavaScript that prevents the browser from painting it.
A performance score tells you that something is wrong. Looking at the actual rendering path tells you what to fix.
Caching helps, but it cannot fix the page by itself
WordPress has work to do before it can return a page.
Depending on the site, PHP may execute, the database may be queried, plugins may run and the final HTML may have to be assembled before it reaches the visitor.
Page caching can skip much of that repeated work by serving a prepared version of an eligible page. Browser caching can keep returning visitors from downloading unchanged static resources again.
That can make a large difference.
But caching does not determine what happens after the browser receives the page.
A 1.5 MB hero image is still a 1.5 MB hero image. Several font families still have to load. Advertising scripts, analytics tags, review widgets, chat tools and embedded videos still add work.
This is why a caching plugin can improve a site substantially without turning a poorly constructed page into a fast one.
Caching improves delivery. The page itself still needs to be economical.
Small business websites accumulate scripts quickly
A typical business website starts fairly simple.
Then someone installs analytics. Advertising adds another tag. The company starts using call tracking. Reviews are embedded on the homepage. A scheduling tool gets added. Then live chat. A CRM form. Maybe a map, video and consent manager.
None of those tools is necessarily a mistake.
The problem is that every vendor tends to be evaluated independently. The visitor’s browser gets the combined bill.
Third-party JavaScript can require additional DNS lookups, network connections, downloads and execution time. Some scripts also trigger more scripts after they load.
A useful audit asks whether each tool needs to run during the first few seconds of the visit.
A booking system may be essential to the business without needing to block the homepage’s initial render. A video embed below the fold does not need to behave as if the visitor is already watching it. A chat widget may be able to wait until the main page is visible.
The goal is not to remove business functionality. It is to decide when that functionality needs to load.
Performance varies by market, but the pattern is widespread
The original national study used 11,432 deduplicated domains across 327 cities. Routeless has continued the local research separately and now publishes website benchmarks across 336 U.S. cities, covering 13,521 local website observations.
Those two counts measure different things. The national report deduplicates domains. The local dataset counts observations within individual markets, so the same domain can appear in more than one city.
The city data is useful because national medians can hide substantial local variation.
Fresno, California, for example, had a median LCP of 8.4 seconds among the 111 sampled sites that completed a performance test. Turlock’s local sample had a median of 7.2 seconds. Other cities performed differently.
The specific number changes from market to market. Slow mobile rendering does not disappear.
That matters for local businesses because website competition is local too. A contractor in Fresno does not need to outrun every contractor website in the country. It needs to provide a better experience than the alternatives its prospective customers actually encounter.
Page weight deserves more attention
One useful part of a performance investigation is simply looking at how much data the homepage sends to a phone.
Images are an obvious source of weight, particularly on businesses that rely heavily on photography. Restaurants, contractors, dentists, hotels and real estate companies all have legitimate reasons to use strong visuals.
The original file should not automatically become the file delivered to every device.
Modern image formats help. Responsive image sizing helps. Compression helps. So does questioning whether six large images need to appear before a visitor has scrolled.
The same principle applies outside images.
A library added for one interface effect still becomes part of the page. A WordPress plugin that loads assets site-wide can affect pages that never use its visible feature. A marketing tool may add network requests to every visit even when only one campaign needs it.
None of these decisions looks dramatic on its own.
Performance problems tend to arrive through accumulation.
Speed is not useful if the website stops doing its job
There is another finding in the Routeless data worth considering.
Among the 11,432 homepages in the national sample, 63.1% had no detectable contact form. Only 8.4% had detectable online booking, and 29.6% had no analytics tag the audit could identify.
Automated detection has limits. A tool loaded through an unfamiliar implementation may not be detected. Still, the numbers point to a mistake that performance work can create if it goes too far.
The purpose of a business website is not to produce the highest possible test score.
Someone still has to be able to call, submit a form, request an estimate, schedule an appointment or make a purchase.
Analytics also has a cost, but a business needs enough measurement to know whether the website produces anything.
The better approach is to make useful functionality efficient.
Remove duplicate tags. Avoid loading the same service through several plugins. Delay widgets that are not needed immediately. Keep forms as light as their requirements allow.
A fast page that cannot convert a visitor solves the wrong problem.
Performance decisions belong in the design process
Many speed problems are cheaper to prevent than repair.
Consider image handling.
A website can be built so uploaded images are resized into appropriate variants and delivered according to the visitor’s screen. Or the site can depend on every future editor remembering how large an image should be.
The same choice exists with fonts, page builders, animation libraries, tracking tools and plugins.
Performance therefore starts before anyone installs a cache plugin.
It starts with the architecture of the page.
A good custom web design and development process should account for performance when templates, components, integrations and content systems are chosen. Caching then improves a page that was already designed to behave efficiently instead of being asked to compensate for every decision made before it.
That is particularly important for WordPress because the platform makes adding functionality so easy. The barrier to installing one more plugin is low. The cumulative cost is less obvious.
A sensible order for WordPress performance work
Changing ten optimization settings at once makes it hard to know what helped and what broke.
A better process starts with measurement.
Test the homepage and the landing pages that receive meaningful search, advertising or referral traffic. Use mobile testing rather than assuming desktop performance represents the experience most customers receive.
Then identify the main rendering bottleneck. Look at the LCP element, images, fonts, CSS, JavaScript and server response.
After that, review plugins and third-party scripts. Remove what the site no longer needs and delay resources that do not need to participate in the initial page load.
Configure caching and compression around the remaining site.
Then test again.
Forms, navigation, checkout, booking, analytics and other important functions should also be checked after performance changes. A faster page with a broken conversion path is a regression.
Working in that order also creates a baseline. If a future plugin update, redesign or new marketing service slows the site, there is something concrete to compare it with.
Do not optimize for a perfect number
Performance tools are diagnostic tools.
They give developers a repeatable way to find expensive parts of a page and measure whether a change helped.
The score itself is not the business objective.
A useful small business website should show its important content quickly, respond when someone interacts with it and remain visually stable while loading. It also needs clear information, working conversion paths and enough measurement for the owner to understand what happens after visitors arrive.
Caching is an important part of that system on WordPress.
So are the decisions made before the cache ever sees the page.
The 11,432-domain national sample shows how much room remains for improvement. The most effective performance work treats speed as part of how the website is built, not as a repair applied after everything else is finished.
