A website that loses customers rarely looks broken. It stays online, with the right photos and the logo in place, while some of those who arrive leave without calling, without sending a message and without leaving a trace. No one calls to say they have given up. That's why the business owner only sees those who got in touch and concludes that the website "is working".
This checklist shows you where to look, in what order and with what tools. It can be done in an afternoon, without hiring anyone.
Why this loss goes unnoticed
The customer who calls leaves a record: a conversation, a quote, an appointment. The one who closes the tab leaves nothing behind. Without measurement, the website is judged by what the owner sees when opening the page on their own computer, with good internet and knowing exactly what to look for. The real visitor is in a different situation: on their phone, in a hurry, comparing you with two other options.
The signals below range from the simplest to the most technical. The first ones only require a phone and someone from outside the business. The last ones use tools from Google and Microsoft.
1. Is the first screen clear in 10 seconds?
An analysis by Microsoft researchers of more than 205,000 pages, discussed by Nielsen Norman Group, showed that people tend to leave a page within 10 to 20 seconds, and that pages with a clear value proposition hold attention for much longer1. NN/g's recommendation is straightforward: communicate the value of the page in the first 10 seconds1. The study is from 2011, but the logic still applies to those who make quick decisions on their cell phones.
Test it with someone outside the business. Show the first screen of the website on your cell phone for 10 seconds, close it and ask:
- What does this company do?
- Where does it operate?
- How do I reach them?
If the person hesitates at any of the three, the first screen is wasting the most valuable time of the visit. Possible culprits include an abstract slogan in place of the service, a generic photo that could be of any company, and the main service hidden within the menu.
In practice
Write on one line “what I do, for whom and where”. Example: "Residential and commercial air conditioning in Campinas and the surrounding area". Compare it with the headline on your first screen. If the sentences are very different, the headline needs to change.
2. Does your mobile website work well or just load?
According to the ICT Households 2025 survey, by Cetic.br, 65% of internet users in Brazil went online only through a cell phone2. Google also evaluates the website by its mobile version: the official documentation says that the search engine uses the mobile version of the content to index and rank pages3.
"Working on mobile" is not enough. Open your website on your phone and check:
- Is the contact button within reach of your thumb, without having to scroll the entire page?
- Can the text be read without zooming?
- Does the form open the numeric keypad when asking for a phone number?
- Do any elements cover the content, such as a cookie warning that won't close, a pop-up, or a floating button on top of the text?
- Does the mobile version have the same content as the desktop version? Google recommends that the main content be equivalent in both3.
3. Speed measured by real users
A website that loads well on the office Wi-Fi says little about the experience of people who visit it over a mobile network, on the go. Google's PageSpeed Insights tool shows two types of data. The field data come from the Chrome UX Report and reflect the experience of real users over the last 28 days. The laboratory data are a simulation, useful for finding the cause of a problem4.
The numbers that matter are the Core Web Vitals. Google considers the experience good when these thresholds are met in at least 75% of page loads, measured separately on mobile and desktop5:
| Metric | What it measures, in simple language | Good |
|---|---|---|
| LCP | How long the largest element on the screen takes to appear | up to 2.5 s |
| INP | How long the website takes to react to a touch or click | up to 200 ms |
| CLS | How much the page "jumps" while loading | up to 0.1 |
Sites with low traffic may not have enough field data. In this case, PageSpeed shows data for the whole domain or no real data at all4. Then use the lab result as guidance, without treating it as a verdict. If the result on mobile is in the red, the article on how much a slow website costs details what to do.
4. Does every contact option work?
It's a costly silent failure and, at the same time, easy to check. A form that shows "sent" and doesn't deliver anything, a WhatsApp link pointing to an old number, a phone number that doesn't open the dialer when tapped, a contact email going to spam. No analytics tool warns you about this. Only testing does.
Use the review to look at the form with fresh eyes. Nielsen Norman Group recommends short forms: cutting fields whose data can be collected later or simply skipped, keeping at most one or two optional fields, always labeled, and showing visible and specific error messages6. For a service business, name, contact and a short description of what the person needs are usually enough for the first response.
In practice
Once a month, run the 15-minute checklist. Submit the form with a test name and confirm it arrived. Tap the WhatsApp button on your phone and click it on your computer. Tap the phone number. Send an email to the address listed on the site from an external account. Write down the date of each test.
5. Does Google show your website, but no one clicks?
Google Search Console is a free service from Google14 and shows how the site appears in searches. The performance report contains four numbers: clicks (how many times someone clicked on your website from the search), impressions (how many times it appeared), CTR (clicks divided by impressions) and average position. You can filter by search term and page7.
Three patterns deserve attention:
- Many impressions and few clicks in searches about your service. The website appears, but the title and description are not convincing, or the page that appears is not the most appropriate.
- Only searches for the company name. People who already know you find the website. People looking for the service don't.
- Steady drop in clicks over several months, with no seasonal change to explain it.
None of these signals require tinkering with code to be seen. All you need to do is open the report and compare periods.
6. Do visitors take any action?
Google Analytics 4 considers a session "engaged" when it lasts more than 10 seconds, has at least two page views, or records a key event. Engagement rate is the percentage of engaged sessions. Bounce rate is the opposite: the percentage of sessions that did not engage8.
The most useful number for a service business, however, is a different one: how many people tried to contact you. In GA4, any event can be marked as a key event, the name Google now uses for a business's most important actions9. Enhanced measurement already records clicks on links that lead away from the site (the WhatsApp button, for example, generates a click event with the link address) and form interactions10. Mark these actions as key events and submit a test to confirm they appear in reports.
Then compare the pages. A service page with many visits and almost no contacts is telling you something: the offer is not clear, the price scares people off without an explanation, or the path to contact is hidden.
7. Where people get stuck
Microsoft Clarity is free (Microsoft describes the service as "free forever") and shows session recordings and heatmaps11. It also automatically flags frustration behaviors:
- Rage clicks: several quick clicks at the same point, usually where the person expected something to happen.
- Dead clicks: clicks on elements that give no response.
- Quick backs: the person goes to another page and comes back almost immediately.
- Excessive scrolling: more scrolling than average, which may indicate that the person is not finding what they are looking for12.
Watch ten mobile visit recordings on your main service page. Write down what repeats from one session to the next. Recording tools process browsing data, so disclose their use in your privacy policy: the LGPD (art. 6) requires transparency about data processing13. Microsoft itself warns that Clarity should not be used on websites aimed at children under 1811.
What to monitor each month
| Signal | Where to look | What it may indicate |
|---|---|---|
| Contacts (WhatsApp, form, telephone) | GA4, key events | Whether the site generates conversations or just visits |
| Clicks and impressions by service | Search Console | Whether people looking for the service find you |
| Core Web Vitals on mobile | PageSpeed Insights | Whether speed is hurting people on the go |
| Rage clicks and dead clicks | Clarity | Where the page confuses people |
| Manual contact testing | You, with a cell phone | Whether any path broke without warning |
This week’s checklist
- Take the 10-second test with someone outside the business.
- Open the website on your cell phone and go through the five questions in item 2.
- Run PageSpeed Insights on the home page and main service page.
- Test the form, WhatsApp, telephone and email, and note the date.
- Review the form and cut any field that is not necessary for the first reply.
- Open Search Console and compare the last three months with the same period in the previous year.
- Mark contact clicks as key events in GA4.
- Install Clarity, disclose it in the privacy policy and watch ten recordings.
Go beyond the checklist
This checklist covers what you can check on your own. It doesn't tell you which problem matters most in your case, or where to start when several appear at once. That's why Flowt's free diagnosis exists: it measures the business's digital presence with fixed criteria and shows, point by point, where it loses customers. If after this afternoon you're left with more questions than answers, it's a good next step.
