The website gets visits but few leads: how to find the problem

When a website gets visits but few leads, the most common conclusion is that it needs to be rebuilt. It's the most expensive conclusion, and almost never the first one to reach: a potential customer's path can break down at four different points, and only one of them is solved by rebuilding the website. This guide helps you find out which one, before you spend.

The sequence is always the same: someone searches for something, lands on a page, understands (or doesn't) whether you're the right answer, and tries (or doesn't try) to contact you. If you only look at total visits and total leads, you see the first and last link in the chain and guess at the ones in between.

First: are you counting the right thing?

Three numbers often get confused, and the confusion alone can make a working website look broken.

• Visits include people searching for something else entirely. Someone who arrives searching for a competitor's name, or a definition, isn't a missed customer: they never were one.
• Clicks on a button or a phone number aren't leads. They say someone tried, not that something arrived.
• Leads received should be counted where they actually arrive — the inbox, the business management system, the phone — not where the website says it sent them.

The four points where the path breaks down

Taken in order, from earliest to latest. The first one that doesn't hold up is the one to fix: repairing the fourth when the first is broken changes nothing.

1. The promise. Does a visitor understand within a few seconds what you do and for whom? If the page opens talking about innovation and digital transformation instead of the problem you solve, the reader has no reason yet to stay.
2. The proof. Is there anything that gives a reason to believe you — work shown with its context, verifiable numbers, checkable references — or only adjectives? Someone choosing a vendor is looking for reassurance, not enthusiasm.
3. The contact. Is it clear what happens if someone writes to you? Who responds, how quickly, and what they'll receive? A lead is a small risk: you hand over your contact details without knowing what will come back.
4. The mechanics. Does the form actually send, on every device? Is the phone number clickable from a smartphone? Does the page load under normal network conditions, not just on your office's fiber connection?

Comparing pages and sources, without being misled by small numbers

The useful comparison isn't between this month's total and last month's, but between pages and between sources: which ones bring leads and which don't. Two warnings, though, because they're the most common source of bad decisions.

• Small samples say nothing. Out of twenty visits, one more or one fewer lead changes the percentage by five points. That's not an improvement: it's noise.
• Seasonality has to be removed before you read the numbers. Many industries have slow months and busy months: if you compare August with October, you're measuring the calendar, not the website.
• Sources aren't equal. Someone who arrives by searching your name already knows you; someone who arrives from a generic search doesn't. If you mix them together, the average describes neither one.

If the numbers are too small to read — and in many companies they are — it's still possible to make the diagnosis by looking at the path instead of the data: open the website the way a customer would, from the search to the lead, and note where things stall.

An example path: the installer

A company that installs systems gets steady visits and almost no leads. Following the path, you find three different things, and only the third has to do with the website in the strict sense.

• Someone searching for “system maintenance” lands on the services page, which lists the technologies installed but doesn't say whether maintenance is also done on systems installed by others. The most interested visitor doesn't find an answer to their question.
• The portfolio page shows twelve photos without a line of context: it's not clear what problem was solved, how long it took, or for what kind of customer. The proof is there, but it doesn't prove anything.
• The form asks for eleven pieces of information, including VAT number and full address, before even knowing whether the job is doable. Someone who just wanted to ask “can you do this?” gives up.

None of the three requires a new website. The first is a line of text, the second is a caption per job, the third is removing fields.

Choosing a fix, and deciding upfront how to check it

Once you've found the weak point, change one thing at a time. Changing five things together makes it impossible to know which one worked, and you won't be able to answer the next question — “should we do the same on the other page?”

Before making a change, it's worth writing down two things: what you expect to change, and how long it will take to know. If the website gets few leads per month, “we'll see in two weeks” isn't an honest answer: it will take longer, or you'll need to look at an intermediate signal — how many people reach the form, how many start filling it in — instead of the final result.

What this guide doesn't cover

This guide takes you as far as finding the point where the path breaks down. What to ask and what not to ask in a contact form, and how to decide that, deserves a discussion of its own. The same goes for the question “should we fix the website first or invest in advertising?”, which only makes sense once you know where you're losing people.