You have a website. When did you last really look at it?
A site that opens is not yet a site that works. My order for reviewing one: from the visitor's task and the phone to forms, content and what Google sees.
The website opens. The logo is where it should be, the photos load and the menu still says Contact. At first glance everything is fine.
But what we've checked with that is mainly whether the site is reachable. We haven't checked whether it helps a person understand the offer, find the right information and take the next step. There can be quite a gap between those two things.
If I were handed a small company's website to review today, I wouldn't start with a list of technical terms. I'd open it first as someone who doesn't know the company. Without the owner explaining what a particular sentence was meant to say, or why a certain link is hidden exactly there.
For me that is a useful starting point. A site has to tell a visitor enough even when nobody is standing next to them to explain it.
First I'd look for the reason for the visit
I'd ask myself what the person came for. Do they want to call? Check the offer? Order a service? Find out whether the company even works in their area?
Then I'd try to complete that task. Not just look at the first screen. If I'm looking for a specific service, I'd go to its page. If I need a quote, I'd try to reach the form. If I want to visit the premises, I'd look for the address and opening hours.
Here I'd pay attention to the moments where I have to guess. A "More" button can help me if the surroundings make clear what I'll get. If there are three similar buttons on the screen, I don't know which one will lead me to the answer. A pretty menu doesn't help much either if the sections are named after the company's internal structure, which the visitor doesn't know.
At the end of every important page I'd check what I'm supposed to do now. Not every paragraph has to sell. But a sensible next step should be visible enough. A link to details, a phone number or a form often does more than one more generic sentence about quality.
I'd pick up the phone fairly early
I'd check the mobile view on an actual phone, not just by shrinking the browser window. I'm interested in how the site works with a finger, with the keyboard open, on a screen where there isn't much room.
Can I read the intro without zooming? Can I open and close the menu without hunting for the right spot? Does the cookie notice cover the essential part of the page? Do I press a button by mistake while scrolling? Things like this are easier to understand when you use the site than when you only look at a screenshot.
I'd take a particular look at forms. The wrong keyboard type, a tiny field, or an error message shown far away from the input can turn a perfectly ordinary task into needless fiddling with the site. A table that is clear on a desktop may push an important figure off the screen on a phone.
I'd also check keyboard use on a computer. I want to see which element I'm on and reach the menu and the form without a mouse. That helps me notice obstacles I easily miss when clicking around quickly.
With speed, what interests me is the waiting
A number from a speed-testing tool is useful, but on its own it doesn't convince me that a site is good or bad. First I'd look at what the user is actually waiting for.
Does a blank area appear first? Does a photo shift the text just as I want to click a link? Do I press a button and for a while not know whether anything happened at all? Each of these calls for a slightly different kind of check.
Core Web Vitals cover the loading of the main content, responsiveness and visual stability. I'd look at these measurements for orientation, and then at the differences between data from real visitors and a lab test. One test run is not a description of every visit.
If I found a problem, I'd look for the cause: an oversized image, an unnecessary script, a slow server response or something else. I wouldn't start by chasing a perfect score. I'd first remove the waiting from the tasks that bring people to the site in the first place.
I'd actually submit the form
A form that looks correct is not a tested form. In an agreed test I'd send a message and check the whole path, including whether the company actually receives it.
Before sending, I'd ask who receives the messages and whether a submission triggers anything else. With an order, a payment or a system that creates a sales order, I wouldn't press buttons blindly on production. The test has to be recognisable and agreed, so it doesn't create work for somebody else.
I'd also want to know what the visitor sees after submitting. A clear confirmation matters because it tells you whether the action succeeded. I'd check the required fields too, whether the errors are understandable, and whether data already entered survives an error.
The same goes for e-mail and phone links. Do they open the right address or number? Is the information still current? These aren't especially exciting technical topics, but they are very concrete things that can make a site lose its whole purpose.
I'd compare the content with the company as it is today
A site can work without a single technical fault and still present the company as it was several years ago. The offer changes, employees leave, the location moves, and ordering works differently now.
I'd read the service pages, the company introduction and the contact page. I'd look for information where the owner would have to stop and say: we don't actually do that anymore. I'd check downloadable documents, price lists and promises about response times too. Outdated information is no less misleading just because it sits in a PDF.
With the text, I'd look at whether it is concrete enough. "Complete solutions for all your needs" doesn't tell me whether the company solves my problem. A clear description of the service, its scope and how working together goes helps me more. Where an answer isn't possible without a conversation, that can be said plainly too.
I wouldn't ask for longer text just for the sake of length. Sometimes the best fix is deleting a paragraph that promises a lot and explains nothing. Other times what's missing is a purely practical piece of information that the staff give out on the phone every day.
Then I'd look at what Google sees
Only now would I open the search data. I'd use Google Search Console to check indexing, queries, impressions, clicks and any problems. First I'd need proper access to the right property, not a screenshot of a chart without context.
I'd pick a few important URLs and check them with the URL inspection tool. I want to know whether the page is in the index, which canonical was chosen, and whether there is a problem with access or rendering. I'd distinguish between data about the indexed version and a live test of the current URL. That isn't always the same state of the page.
If I typed site: and my domain into Google, I'd take the result as orientation. Google explains that this kind of search is not a complete list of indexed pages. So I wouldn't turn the number of results shown into a precise indexing report.
I'd also check whether the main content is actually accessible to the search engine. An important description that exists only inside an image is not equivalent to normal text. With a dynamic site I'd look at the rendered content and wouldn't settle for the URL returning a successful response.
Titles and descriptions should say where a link leads
For the important pages I'd compare the visible heading, the SEO title and the content. If every subpage carries the same company name as its title, a visitor already struggles among the results to tell where to click.
I'd try to write a good title so that it says what is on that specific page. I'd use the meta description as a short, meaningful summary. Not as a place to list every service and every town you'd like to appear for.
I'd also keep in mind that Google assembles the title link from several sources and that it may pick the snippet from the page content, not necessarily from the meta description you entered. A filled-in field in the CMS doesn't yet mean the result will always look exactly like that.
If Google showed a confusing title, I'd check the whole page and its naming. Manually fixing one field makes sense, but it has to match what a person reads after clicking.
Technical clarity without magic add-ons
I'd check HTTPS on the important paths, not just the home page. I'd watch for browser warnings, wrong redirects and forms that lead to an unexpected address. An encrypted connection is basic protection for the transfer; on its own it doesn't confirm that the company's content is trustworthy.
I'd check structured data where it describes something real: the company, a product, an article or other supported content. The markup has to match the visible content. I wouldn't add ratings, questions or business details just because some tool knows how to generate them.
I look at readability for AI systems the same way. First I'd sort out clear naming of the company, services and author, sensible headings and accessible text. These are things that help a person too. An extra technical file can't fix a site from which you can't tell what the company offers.
For a local business I'd compare the address, phone, service area and opening hours on the site with the business profile. I'd check the directions as well. The visitor doesn't care which system the discrepancy came from; they care about when and where they can turn up.
From the review I'd make an order of work
In the end I wouldn't want just a long list of remarks. I'd sort them by what they obstruct and how quickly the problem can be fixed.
A broken form or a wrong phone number would come before a minor design improvement. An important service that can't be found would come before one more generic article. For a more demanding change I'd want to know how we'll check whether the fix actually helped.
A review like this is useful to me if it makes the company's next decision easier. The whole site doesn't have to be replaced right away. Sometimes small fixes are needed; other times the review shows that the offer has changed so much that the old design is no longer enough.
Reviewing, building and optimising websites is also part of what I do at GE-KO, my web development and SEO business. But my first suggestion to an owner stays fairly simple: pick up your phone, open your site and try to do what you expect a visitor to do. All the way. Through the form, the confirmation and the message that arrives.