Website Speed for Local Business: What Moves the Needle and What Doesn’t
You’ve probably got a speed report sitting in your inbox with a row of red bars on it. Nothing in it tells you whether those bars are costing you phone calls, which makes it hard to know what to do next.
So the site ends up with three caching plugins, somebody loses a Saturday to chasing a green score, and August looks a lot like July. Website speed for a local business really only needs a short list of changes, most of which you can handle in an afternoon, and the rest of that report can wait until you’ve done them.
The three numbers that matter
Google’s Core Web Vitals are the short list, and they’re the same three whether you run a plumbing company or a dental office. Pages should reach Largest Contentful Paint within 2.5 seconds of when loading starts, and Interaction to Next Paint should stay at 200 milliseconds or less (web.dev). Cumulative Layout Shift should come in at 0.1 or less (web.dev on CLS). Each one tracks something a visitor actually notices, like how long they stare at a blank space, whether a tap does anything right away, and whether the button moves before their thumb lands on it.
| Metric | What it measures | Good |
|---|---|---|
| LCP | Main content painted | 2.5 seconds or less |
| INP | Response to a tap | 200 ms or less |
| CLS | Layout jumping | 0.1 or less |
Sources for the table: web.dev on Core Web Vitals, web.dev on INP, and web.dev on CLS.
The detail almost everyone misses is how those targets get measured. Each one is judged at the 75th percentile of page loads, split between mobile and desktop (web.dev), which means your own phone on office wifi tells you very little. The customer you care about is on a three-year-old Android in a truck with two bars, and that load is the one being scored.
Lab scores and real visitors are different data
PageSpeed Insights hands you two kinds of data on the same screen, and most arguments about speed happen because people are reading the wrong one. Lab data comes from Lighthouse, which simulates a page load in a controlled setting and exists for debugging. Field data comes from the Chrome User Experience Report, which captures what real visitors experienced across a narrower set of metrics (Google).
A lab score can swing eight points between two runs while nobody visiting your site notices a change in load time. Watch the field data month over month and treat the lab number as a diagnostic tool. Field data also reflects the people you actually want, loading your service pages on their own phones in your service area.
Google’s own wording about rankings is careful, and yours should be too. Its core ranking systems look to reward content that provides a good page experience, and Core Web Vitals sit among the questions it lists for assessing that (Google). Speed helps. It won’t lift your SEO rankings past a competitor with more reviews and more relevant pages, so keep it in proportion with the rest of your technical SEO work.
Four fixes that improve website speed for a local business
None of these require a rebuild, and on a typical service site they’ll take a developer less than a day.
Serve right-sized modern images. WebP gives you a modern lossy and lossless format built for the web (web.dev), and switching to it usually cuts image weight substantially on a photo-heavy page. A 2,400 pixel hero photo that the browser shrinks to fit a phone screen wastes your visitor’s data and your LCP at the same time.
Lazy-load what sits below the fold. Browser-level lazy loading works through the loading="lazy" attribute with no library involved (web.dev). Leave the hero image out of it, because that’s the one image you want the browser to fetch immediately.
Cut the third-party scripts you can’t name. A chat widget, a heat map tool, two analytics tags, a review carousel, and an ad pixel each add work on the main thread. Most small business sites carry at least one that nobody has looked at in two years. Every script you remove is a permanent win rather than a temporary one.
Ship less JavaScript. Code splitting breaks a large bundle into pieces the browser loads only when it needs them (web.dev). On a brochure site the practical version of this fix is deleting the page builder add-ons that got installed for one section you later removed.
Two supporting items are worth a look once those four are done. Time to First Byte feeds directly into LCP, and a slow TTFB often traces back to server redirects, distant servers, or poor network conditions (web.dev). Web fonts are also a routine source of both delay and layout shift, so if your headings pop in late or the text reflows as you read, start with font loading.
What doesn’t move the needle
Some of the most commonly sold speed work produces nothing a customer would ever feel.
| Common advice | Verdict | Why |
|---|---|---|
| Chasing a 100 lab score | Skip | Lab is a debug tool |
| Stacking three cache plugins | Skip | They fight each other |
| Minifying a 4 KB CSS file | Skip | Rounding error |
| Deferring the hero image | Skip | Hurts LCP directly |
| Swapping hosts before an audit | Wait | Fix payload first |
| Removing unused scripts | Do it | Permanent, measurable |
Taking bytes and requests off the page shortens the wait for the person tapping your number, which is the whole point. So when someone quotes you for speed optimization, ask which bytes are leaving the page, what the load time was before they started, and how they plan to measure the result. Vague answers there usually mean the score goes up and the site feels the same.
A one-hour audit you can run yourself
Run your three most valuable pages through PageSpeed Insights: the homepage, your busiest service page, and the contact page. Read the field data section first, then look at the lab diagnostics for clues about what’s causing it.
Then answer four questions and write down the answers with the date:
- How large is the hero image, in kilobytes?
- How many third-party scripts load, and can you name what each one does?
- Does anything jump around while the page loads?
- Does tapping the phone number respond within about a second?
Re-run the same three pages after each change, one change at a time, so you learn which fix moved the load times instead of guessing. That habit also protects you from paying twice for the same result, which happens more often than it should when a site gets rebuilt without anyone measuring the old one first.
Frequently asked questions
How fast should a local business website load?
Aim for the Core Web Vitals targets: LCP within 2.5 seconds, INP at 200 milliseconds or less, and CLS at 0.1 or less, measured at the 75th percentile (web.dev).
Does website speed affect my search rankings?
It contributes. Google says its core ranking systems look to reward content that provides a good page experience, and Core Web Vitals are part of that picture (Google).
Why does my PageSpeed Insights score keep changing?
You’re likely reading lab data, which is a simulated load from Lighthouse used for debugging (Google). Field data from the Chrome User Experience Report is the steadier read.
What is the single biggest win on most small sites?
Images. Serving right-sized WebP files and lazy-loading what sits below the fold removes real bytes (web.dev).
Should I lazy-load every image?
No. Use loading="lazy" for offscreen images and leave your hero out of it (web.dev).
Will a faster host fix everything?
It can help time to first byte, which feeds LCP (web.dev). It won’t fix a 3 MB hero image or a stack of third-party scripts.
Where to start this week
Pick the three pages that earn your calls, fix the images, drop the scripts you can’t account for, then measure again before touching anything else. If you’d rather not spend the afternoon in PageSpeed Insights, we can run the audit on your site and hand back the short list of what’s worth fixing. Take a look at our website work or get in touch and we’ll take it from there.
[INFOGRAPHIC HERE · file: images/speed_03_infographic.png · alt text: “J&A Digital Solutions infographic titled The one-hour speed audit, listing the three pages to test and the four questions to answer about hero image size, third-party scripts, layout shift, and tap response”]