Why Your Website Is Not Getting Leads
Five places a website quietly loses the people it was built to win, how to check each one yourself, and what a free two-minute scan can and cannot tell you.
If your website is not bringing in calls, the instinct is to buy more traffic. Run ads, hire someone for SEO, post more. Hold off. More visitors into a site that loses people just loses more of them, and you pay for every one.
A website fails to produce leads in a small number of predictable places. Somebody cannot find you. They find you and cannot tell what you do. They understand you and cannot reach you easily from a phone. Or they reach you, and afterwards nobody can say what brought them. That last one matters more than it sounds, because it decides whether you can fix any of the others.
I will walk through each: what it looks like, how to check it yourself, and where a machine can check it for you. The fastest way to check most of them is a free scan I built for exactly this. Run it on your own site. It takes about two minutes and asks for nothing but your web address. Then come back, because the grades only mean something once you know what they are measuring and what they are not. I ran it on this site too, and the result is further down. It is not all A's.
Are people leaving before your page even loads?
Your customer is probably on a phone, possibly on a patchy connection, and comparing you with the two businesses listed above and below you. If your page is slow to show anything, they do not wait. They tap back and pick the next one.
Google publishes the yardstick it uses, and it is more useful than any single speed score (Google, web.dev). There are three measures, and each has a line for good and a line for poor (Google, web.dev).
- Main content appearsLargest Contentful Paint
- The page answers a tapInteraction to Next Paint
- Nothing jumps as it loadsCumulative Layout Shift
Google measures all three at the 75th percentile of real visits, so a site that is fast on your office wifi and slow for more than a quarter of your customers still fails.
The usual causes are unglamorous. Photos uploaded straight off a camera at ten times the size they need to be. A heavy theme or page builder loading code for features you never used. A chat widget, two analytics tags, a review badge and a booking script all loading before your headline. An auto-rotating slider at the top of the page.
What the scan checks: it runs Google PageSpeed Insights on your page twice, once on a throttled mobile connection and once on desktop. Where Google has collected enough real Chrome visits to your site to publish the data, it reads that too, so you see what your visitors actually experienced rather than only what a lab measured.
This is where my own site loses points. In Google's throttled mobile test, the main content on the homepage took 6.6 seconds to appear, well past the 4-second line, and the scan graded speed a D.
What it does not mean: a slow site loses some people who would have called. A fast site does not make anyone call. Speed is a floor, not a sales strategy.
Can Google actually read what your page is about?
Before a page can rank for anything, a crawler has to understand it. That comes down to a handful of things in the page's code that most owners have never looked at, because they do not show up on the screen:
- A title for each page. Not "Home." The title is usually the blue link people click in search results.
- A description. Leave it out and Google writes its own from text on the page, which is sometimes your cookie banner.
- An instruction that the page may be indexed. A surprising number of sites still carry a "do not index" setting left over from when the site was being built.
- Structured data saying you are a business, with an address, hours and a phone number, in a format search engines read directly.
- A sitemap and a robots file that a crawler can actually reach.
What the scan checks: it reads what the page tells a crawler about itself, including the title and description, canonical and robots directives, headings and structured data, then reconciles all of it against Google's own SEO checks so that no single source carries a claim on its own.
What it does not mean: being technically readable is not the same as ranking. It is the entry fee. If someone promises you a ranking position in return for fixing these, read this first. For the fuller version of the technical side, I wrote a guide to technical SEO for small sites.
When someone lands, can they tell what you do?
This is the failure owners are worst at spotting on their own site, for a simple reason: they already know what they do.
Try this. Show your homepage to someone who does not know your business for about six seconds, then close it. Ask them three things. What does this business do? Where? And who is it for? If they hesitate on any of the three, so does every stranger who lands there from a search.
The usual culprit is a headline written for the owner rather than the customer.
- What?
- Cannot tell
- Where?
- Cannot tell
- For whom?
- Cannot tell
- What?
- Emergency plumbing
- Where?
- Northeast Philadelphia
- For whom?
- Someone with a leak, today
What the scan checks: its Content category looks at the structure that makes a page readable to people and to machines. Whether headings step down in order, whether images carry a text description, whether links say where they go instead of "click here," and whether the page has a title at all.
What it cannot check: whether your headline is clear to a customer. Structure can be perfect and the message can still be fog. That part needs a person reading it with your customer in mind.
Can someone who is ready contact you from a phone, without hunting?
Somebody has decided. They want to call you. Now count the obstacles.
A few common ones: a phone number set inside an image, so it cannot be tapped. A contact page reachable only from a link in the footer. A form asking for twelve things when three would do. No address or service area anywhere, so a visitor cannot tell whether you even cover their street. A "request a quote" button that opens an email draft on a phone that has no email set up.
Each of these is small. Together they are the difference between a visitor who meant to contact you and a lead you actually got.
What the scan checks: its Trust and Conversion category looks at what a visitor who is ready to act runs into. Whether there is a way to reach you without searching for it, whether the page says who you are and where, and whether the path from landing to getting in touch survives being opened on a phone.
What it cannot check: whether anyone answers. A perfect contact path that rings through to a voicemail box nobody checks until Friday is still losing the lead. No website scan can see that.
When a lead arrives, can you tell where it came from?
This is the one that decides whether you can fix the other four. If nothing on your site is counting calls and form submissions, you cannot tell which of the problems above is costing you, and you cannot tell whether a fix worked. You are left guessing, and paying for the guesses.
What the scan checks: it watches the network while your page loads and records what actually fires, looking for Google Analytics 4, Google Tag Manager, Meta Pixel, LinkedIn Insight and HubSpot. A tag it sees firing is reported with high confidence. A tag that is only mentioned in the page's code, but never seen running, is reported with low confidence and labelled that way rather than counted as working.
What it does not mean: finding Google Analytics installed is not the same as it counting your leads. A default install records page views and some automatic interactions. With one setting switched on, it can also log a generic form submission, but that records somebody pressing the button, not an inquiry arriving, and it may not see a form embedded from another service at all. A tap on your phone number is not counted as a lead unless someone sets that up. Until both are set up and tested with a real inquiry, the two moments that matter are guesses. I wrote about that exact gap in what a marketing audit actually finds, and about how to set the counting up properly in the GA4 attribution guide.
The five, side by side
What the scan measuresGoogle PageSpeed on mobile and desktop, plus real-visit data where Google has it.
What it cannot seeWhether speed is your biggest problem.
What the scan measuresTitles, descriptions, indexing, structured data, sitemap.
What it cannot seeWhere you rank, or why.
What the scan measuresHeading order, image text, link text, page title.
What it cannot seeWhether a customer understands you.
What the scan measuresWhether a contact path exists and works on a phone.
What it cannot seeWhether anyone answers.
What the scan measuresWhich tracking tags actually fire.
What it cannot seeWhether they record leads or only visits.
What a free scan cannot tell you
I would rather you hear the limits from me than discover them later.
It reads one page. The address you give it, which is usually your homepage. It does not crawl your service pages, and it does not follow your internal links, so it will not tell you that a menu item leads nowhere.
It does not see your Google Business Profile. For a local business that matters a great deal. In Whitespark's study of local searches, the "plumbers in Houston" kind, Google showed the map listing 93 percent of the time (Whitespark, 2025). For a lot of your customers the profile, not the website, is the first thing they see. A clean website grade tells you nothing about it.
It does not know your business. Your prices, your reviews, whether your offer is the right one, whether anyone calls back. Those decide more leads than anything on this list, and no machine reading a page can judge them.
A good grade is not a forecast. It will not tell you how many leads you will get, where you will rank, or how much traffic a fix will bring. Nobody honest can promise those from a scan. It tells you whether the basics a machine can measure are in order, and which two or three to fix first.
How to read your result
You get a grade for each of the five categories and two or three findings, ordered by what is worth fixing first. Here is a real result, run on this site. The address field is cropped from the screenshot below; the grades and findings are unedited.
Both findings are about mobile speed, and they are the right two to lead with. Tracking is a C because the scan did not see a Meta Pixel, which only matters if you advertise on Facebook or Instagram. Read each grade against what your business actually does.
Start with the findings, not the grades. The grades tell you where you stand. The findings tell you what to do on Monday.
If a customer who is ready cannot reach you, fix that first. A form that goes nowhere or a number that cannot be tapped is losing inquiries today.
Then, if Tracking is low, fix it next, even if it is not the lowest grade. Every other fix becomes something you can measure instead of something you hope worked.
Most findings are an afternoon of work for whoever built or maintains your site. Hand them the list as written.
On how it works: the grades and the findings both come from fixed rules, not from AI.
- It measures the pageGoogle PageSpeed on mobile and desktop, real-visit data where Google publishes it, the page's code, and every request the page makes while it loads.
- Fixed rules calculate the gradesThe same measurements always produce the same grade. No model is involved.
- Fixed rules write the findingsA finding appears only when the scan has the evidence for it, and it quotes the measured number.
- A language model may add and rephraseIt can suggest more findings and word them more clearly. If it fails, the report is complete without it.
- The scan picks the two or three worth fixing firstNear-duplicates are dropped, and it does not lead with something that is fine.
That is the order I work in on every engagement. Measure what can be measured, then have a person read the rest with your business in mind. The scan is the first half, and it costs nothing. Run it on your own site. If you want the second half, book a call and I will read it with you.
Common questions
Is the website audit really free?
Yes. There is no account to create and no card, and you do not need to give an email address to see your result. It shows on the screen, and it is saved at its own link for 90 days.
How long does it take?
About two minutes. Most of that is Google's PageSpeed test running on your page twice, once as a phone on a slow connection and once as a desktop.
Does a good score mean my website will get more leads?
No. A good score means the basics a machine can measure are in order. Leads also depend on your offer, your reviews, your Google Business Profile and whether anyone answers the phone, none of which a website scan can see.
Will it find broken links on my website?
No. It reads the one page you give it and does not follow links to other pages.
What should I fix first?
Anything that stops a ready customer reaching you, like a form that goes nowhere or a phone number that cannot be tapped. After that, if Tracking is low, fix it next so every other fix can be measured. Otherwise, work down the findings in the order the scan lists them.
Does the scan use AI?
Only to help word the findings. The grades and the findings themselves come from fixed rules applied to what the scan measured, so a score never depends on how a model chose to word something. Speed scores do move a little between runs, because Google's own test does.