Is SEO Worth It for a Small Business?
Leads from social is a good place to be. You just do not own that platform. Score your channels against four questions and see which one you cannot afford to lose.
A good website loads fast on a phone, shows up on Google and in AI answers, gets every lead to your inbox, reads clearly for everyone, says only what's true, and keeps its design on every screen. After more than five years of building websites, I check them with 98 automated checks, and most of those exist because something went wrong on a real site once.
A good website is one that does six jobs for a business every day: it's fast on a phone, it gets found on Google and in AI answers, every lead reaches you, anyone can read and use it, everything it says is true, and its design holds up on every screen.
Website design gets most of the attention, and it should get some, because a visually appealing page is what keeps a visitor reading long enough to trust you. I've also watched sites that look great lose the inquiry anyway, because the form saved nothing or the main photo took 8 seconds to show up on a phone.
The page your customer waits for shows up quickly on a phone, on a normal connection.
Search engines and AI assistants can read every page, tell what it is about, and quote it.
Every form, call and quiz lands in your inbox with the page it came from.
Text is easy to read, links look like links, and every button works where it is drawn.
Every promise, price and review on the site is true and can be checked.
The layout you approved looks the same on every page and every screen size.
You can test your own website against the six jobs with nothing more than your phone and a browser, one short check per job. Count your no answers as you go.
Open pagespeed.web.dev, paste your homepage address and look at the Mobile result. Find the line called Largest Contentful Paint.
If it's a no: Over 2.5 seconds means a customer on a phone is looking at a half-built page. The main photo is usually what's holding it back.
Search Google for your main service and your town, the way a customer would. Then ask ChatGPT the same question.
If it's a no: If your business shows up in neither, search engines and AI assistants can't tell what your pages are about or where you work.
Fill out your own contact form with the word TEST in the message. Then check your inbox, your spam folder and anywhere else leads are supposed to land.
If it's a no: If it didn't show up within a few minutes, real inquiries could be going missing the same way.
Take your phone outside, open your website, read one full paragraph and tap every button on the homepage.
If it's a no: If the text was faint in daylight or a button needed a second tap, some visitors are giving up before they reach you.
Read your homepage and service pages for promises like same-day service, 24/7, guaranteed, a number of years in business or a review count.
If it's a no: If there's one you couldn't back up when a customer asks, it's costing you trust, and AI tools may repeat it as fact.
Open your homepage and two other pages on your phone, one after the other. Compare the logo, the buttons, the spacing and the photos.
If it's a no: If one page looks like a different website, or something sits off to one side, visitors notice even when they can't say why.
Two or more no answers is the point where I'd want to look at the site with you. Bring your results to a free call, and we'll start with whichever job is costing you the most.
It's all right here, nothing hidden. If you'd rather run it later on your phone, I'll send you the list.
Sent. It should land in the next couple of minutes.
More than five years of building websites taught me most of these lessons the expensive way. In 2026 I started writing each one down as a check that runs on its own, so a mistake I'd already made once couldn't reach another client's site. I build in code with Claude Code, and the checks are small programs that read the finished site the way a browser and a search engine would.
Most of them run every time I change a site. Some run only where they apply, like on a blog or a page that shows a review count, and a few run before or after launch, like the redirect map when a site moves over from a platform like Squarespace.
From 1 check in late June 2026 to 98 by early October. Most of the climb came in September, the month I ran Google's speed test on every live site I manage and most of the client problems below turned into checks.
Design has the most checks, 24, because it has the most ways to quietly go wrong, one card or one button at a time. The seventh row isn't one of the six jobs, because its 14 checks are there to make sure the site that goes live is the one you approved.
A website is fast on a phone when its main content shows up within 2.5 seconds on a normal mobile connection, which is the mark Google's own guidance calls good, and the main photo is usually what decides it. So being mobile friendly takes more than a layout that fits a small screen, and the photo itself has to be light enough to reach mobile devices quickly.
I ran Google's PageSpeed test on the homepages of the 11 live custom sites I manage in October 2026. Laptops scored close to perfect and phones trailed, which is true of almost every website and the reason phone speed gets its own 8 checks.
Desktop, median of 11 sites
99
the middle score on a laptop
Phone, median of 11 sites
73
the middle score on a phone
Phone, best site
95
the fastest of the 11 on a phone
Google PageSpeed scores out of 100, homepages of 11 live sites I manage, October 2026.
The slowest phone pages all shared one thing: a main photo saved as a heavy JPEG, even where the photo itself was a reasonable size. One site's main image was a 60 KB WebP and showed up in 2.4 seconds, while another's was a 289 KB JPEG and took 8 seconds.
The image a phone waits for has to be a light format like WebP, loaded first and never lazily. Converting one site's main images cut their total weight by 47%, and they stayed just as high quality on screen.
A settlement company's site asked Google Fonts for its typeface with one small mistake in how the request listed the font's styles. Google refused it, the browser quietly fell back to Georgia, and every page showed the wrong type for four days, through several reviews, because nothing on screen looked broken. Now a check confirms that every font a page asks for actually loads.
Google and AI assistants can find a website when each page carries labels in its code that say what the page is about, links to the pages related to it, and passages that answer a question on their own. None of that shows on screen, which is exactly why it gets skipped, and a page search engines can't read has a hard time showing up in search results. Getting found is 16 checks of that plumbing.
Two pages for a painting company went live with no structured data at all, the hidden labels that tell Google what a page is. Every check passed, because none of them had ever asked whether the labels existed, so now one asks on every page.
A link sweep on my own site found four pages with two or fewer links pointing at them, one of them a brand new service page. Search engines read a page with few links pointing at it as a minor page, so the whole site's link map gets checked every time the site changes.
When I moved my own site onto this setup, one page went live as a redirect pointing at itself, past the eighteen checks I had then, because redirects only run on the live server. Now every redirect rule gets tested before anything goes live.
For AI answers, every FAQ answer on my sites has to make sense with no question printed above it, because an assistant lifts the answer on its own and leaves the heading behind. I also keep a plain list of every page for AI tools to read, called llms.txt, and a check fails when a page is missing from it. If your site isn't getting found at all, my post on why your website is not showing up on Google goes through the causes one by one.
Not on every site, and you can't tell by looking, because a contact form can look perfect, show a thank you message and still never deliver a thing. Nothing shows an error when a record is simply never written, so nobody notices. Lead delivery has 15 checks, more than any job except design and getting found.
A therapy practice's contact forms sent each inquiry by email, but the backup copy every lead is supposed to leave behind was never saved, so for most leads that one email was the only record. Now every form has to prove its backup stores every field before the site can go live.
A tutoring company kept getting spam after its contact form was fixed, because a second form on the site used its own path and skipped the protection. The anti-spam check now looks at every form, and it has to prove it lets awkward but real customers through, like a one-word message or a name that isn't in English.
Confirmation emails that invite a reply have to come from an address that can receive mail, or a customer's reply bounces and nobody ever sees it. Every lead also has to carry the page and the channel it came from, whether that's Google, an ad or social media, all the way to the inbox and the spreadsheet, so an owner can see which pages bring in business.
I wrote more on what to do once the lead lands in my lead follow up post.
Everyone can read and use a site when it's easy to navigate for someone in bright sunlight, someone using a screen reader and someone who just wants to tap a button and have it work the first time. That's what user friendly means in practice, and on a small business site it's most of what a good user experience comes down to. I'd rather measure it than assume it, so these 10 checks run in a real browser at phone and desktop sizes, where the problems actually show up.
Everything on a great website is true, which means it only promises what the business can back up. If the owner hasn't confirmed same-day service, a response time or a guarantee, it doesn't go on the page, and a check scans every page for those promises. Honesty has 11 checks, and my own site has needed the same fixes.
My site said custom sites like mine test in the 90s on Google's PageSpeed tool. That was true on desktop and not on phones, and the sentence didn't say which. I changed it to name the device and the date, and removed a comparison to other people's sites I had never measured.
A quote on the page has to match the customer's real review exactly, typos included, with their name beside it.
Every result shown on a site carries the date it was pulled and how long it can stay up, and it has to be refreshed or taken down once it goes stale.
A price on a service card has to match the one price list the business keeps, so an old number can't hang on in a corner of the site nobody looked at.
The writing gets checked too. One check asks whether a sentence could move to another business's website unchanged, and flags the page when too many could, because generic copy is the quickest way to sound like everyone else.
What makes a good website design is consistency more than any single flourish, so the layout you approved looks the same on every page and at every screen size. Responsive design means the layout works at every width, and the 24 design checks measure the finished page in a real browser at phone and desktop widths to prove it.
One of them came from my own review of seven new pages. Six of my 23 notes were the same problem in different places: a section centered everything except one row of buttons or one card, which sat stuck to the left, and every check had passed.
The website that goes live is the one you approved only when the settings files the host reads get checked too, because some problems show up at the moment a site goes live and nowhere before it. These 14 checks make sure the site that reaches your customers is the one you signed off on.
An auto body shop's settings file had a quote mark in the wrong place. The site built and every check passed, but the host would have dropped the rule that lets the spam filter load, and every lead would have been held for review, so now a check proves every settings file reads cleanly before anything goes live.
A site update can finish with zero pages in it, and most checks pass a site with no pages because there's nothing to fail, so one check exists only to stop everything when the finished site is empty.
Automated website checks can't judge whether a page sounds like the business, speaks to its target audience or simply looks right, which is why every page also gets four reviews by judgment before it goes live. Sixteen checks once passed on a page I rejected the moment I opened it.
Does it sound like the owner talking to a customer, or like any website in the industry?
Does it cover what people searching for it need, and add something the pages ranking above it don't?
Does it look right on a phone and a laptop, the way a customer would see it?
Is there a clear call to action where a reader is ready, and is the next step easy to take?
Whoever wrote the copy is never the only one who judges it. A separate review reads it fresh, and it's allowed to say a page passes every rule and still looks wrong.
When website checks get written faster than they reach every site, the newest sites get the newest protection and the older ones fall behind. In early October 2026 I found I'd been doing exactly that. Each new check went to a client site only once that site already passed it, so a check that would have flagged a problem got parked instead. Some client sites ended up between 12 and 56 checks behind.
I caught those sites up and changed how a check reaches them, so it can't pile up again.
A well built website costs more because building it this way takes a while, and that time is the reason it works. Each change runs every check that site carries, a failure gets fixed instead of waved through, and every page still gets the four reviews by judgment on top.
Most of what you pay for in a well built website is work you never see. Every main photo gets converted so the page stays fast on a phone, my monitoring sends test leads through the live forms on most of the sites I manage so I know they still deliver, and anything the owner can't confirm stays off the site.
I build the website inside a monthly plan, so the checks keep running after launch instead of stopping on the day the invoice is paid.
If you want the numbers, my breakdown of how much a website costs for a small business lays them out, and my small business website design page shows what goes into a site I build. If you already have a site, my post on website redesign cost tells the full story of the redirect loop above.
On your website I'd check three things first: whether a test lead actually reaches you, how fast your homepage shows up on a phone, and whether Google can tell what each page is about. Bring your website and the last couple of inquiries you got, and we'll go through it together on the call.
You've probably built some of it yourself, and that's how I started too. Owners usually bring me in when time becomes the bottleneck, not know-how.

Put your web address in the message with what you want it to do better, and I'll tell you the first thing I'd check.
Last 90 days, across the sites and Search Console/GA4 accounts Another Monday manages. Figures as of August 2026.
A good website loads fast on a phone, shows up on Google and in AI answers, gets every lead to the business, is easy for anyone to read and use, says only what is true, and keeps its design on every screen. Looks matter, but a website is good when it does those six jobs for the business every day.
A successful website helps website visitors find what they need before they lose patience, with calls to action that say what happens next and landing pages that match the search that brought them in. The best practices behind user experience (UX) are mostly plain: fast pages, clear headings, a phone number you can tap, and nothing to dig through. To optimize your website for potential customers, open it on a phone and read it the way a stranger would.
The most important things on a website are a clear statement of what the business does and where, a fast page on mobile devices, an easy way to contact the business, proof from real customers, and pages that search engines can understand. For a small business, the contact path matters most, because a visitor who cannot reach you is a lost customer.
A website launch checklist should cover phone speed, working forms that reach the right inbox, redirects from every old page address, a sitemap search engines can read, structured data on each page, readable contrast, and analytics that only count the live site. Each of those can fail without anything looking broken, which is why they get checked rather than eyeballed.
Testing a website before launch means running it in a real browser at phone and desktop sizes, sending a test lead through every form, checking every link and redirect, and measuring how fast the main image loads on a phone. Automated checks catch the repeatable problems, and a person still needs to look at every page for the ones a script cannot see.
Common website red flags for a business are a contact form that never confirms anything was sent, pages that take several seconds to show anything on a phone, claims with no proof behind them, and buttons or links that do nothing when tapped. Each one quietly costs inquiries without the owner ever seeing an error.
A custom website is better than a template when the website has to bring in customers from search, because every page can be built and tested around what people actually search for. A template is a fair starting point for a site nobody needs to find, like a portfolio you send people to directly.
A good small business website tells a local customer what you do, where you do it and how to reach you within a few seconds of landing, and then backs that up with real reviews and a form or phone number that works on the first tap. It also needs its own page for each service, so search engines can match each one to the people searching for it.
A website should show its main content on a phone in 2.5 seconds or less, which is the mark Google's own guidance calls good for Largest Contentful Paint. The biggest image on the page is usually what decides it, so its file size and format matter more than almost anything else.
The quickest way to make a website load faster on a phone is to fix its main image: save it as WebP or AVIF instead of JPEG, keep it under about 150 KB, load it first instead of lazily, and size it for the space it fills. On one site, converting the main images that way cut their total weight by 47% with no visible loss in quality.
A website looks good when its design is consistent: every page lines up on one column with the same spacing and type, a short last row of cards sits in the middle instead of hugging one edge, and the layout holds on a phone the way it did in the approved design. Real photos of the business help more than any effect, as long as the same photo never shows up twice on one page.
Getting a website into AI answers starts with pages that answer one question clearly near the top, FAQ answers that make sense on their own, the business name, services and service area written in plain text on the page, and an llms.txt file that lists the pages worth reading. AI assistants quote passages, so a passage that only makes sense with the paragraph before it rarely gets used.
A well built website costs more because the testing, the fixes and the second look at every page take real time, and that time is what keeps forms working, pages fast and claims true after launch. The price difference is mostly the work nobody sees until something breaks.
The checks keep running after a website launches, because every change to the site has to pass that site's checks again before it goes live, and the live sites I manage are watched for uptime, with test leads sent through the forms on most of them. A new check also gets tried on at least three client sites before it becomes part of the standard.
A free 30-minute strategy session with the person who'd be building it. Bring your website and we'll run through the six jobs together.
¡También Hablamos Español!