Your site has visitors. The analytics say there are more of them than last spring. The inbox is just as quiet as it was.
The enquiry is not lost where a person arrives on the site. It is lost in the last thirty seconds: after they have already decided they want to talk to you, and are looking for a way to do it. Almost everything that goes wrong in those thirty seconds is small, boring and fixable in an afternoon.
The number that lies
At this point owners usually reach one of two conclusions, and both cost money: "we need more visitors" and "we need a new website". The first means an advertising budget, the second a project of several months, and the visitor count is also the least likely place for the fault to be.
Analytics can tell you how many people came, where they came from and which page they landed on. It cannot tell you why anybody left.
There is no line reading "four people looked for a way to make contact and did not find one", and no line reading "two messages never arrived". Those people look exactly like everybody else in the statistics: arrived, looked, left.
Buying more traffic multiplies the same result by a bigger number.
A new design has a fault of its own: it moves everything at once. If enquiries do start arriving after a rebuild, nobody can say whether it was the colours, the words, or the fact that the phone number finally ended up in the header. And if they do not, three months have gone into arriving back where you started. Both are worth putting behind the things you can check for yourself first.
What a person needs to know before writing to a stranger
Writing to a company is a small discomfort. You hand a stranger your name and your phone number, you have to put your problem into words, and you do not know what happens next. People will put up with that only when they are sure enough of what they are getting.
In the reader's head it runs roughly like this:
- Do you do the thing I need? Not your line of work, the specific job. "Construction services" does not answer the question of whether you replace roofs. A reader does not translate the general into the specific; they leave and ask somewhere else.
- Do you do it where I am? Region decides more here than most owners expect. A site with no place name anywhere on it forces the reader to guess, and a guess always leans towards "probably not".
- Are you a real company? Is there a name, an address, a registry code and a photograph of work that was genuinely done here.
- Roughly what does this cost? Not an exact price. A rough ballpark, so a person knows whether they are even in the right place.
- Will a human answer me? Is there somebody behind that form who reads it, or does it disappear.
And now the important part. These five questions are not weighted. A site that answers four of them is not four fifths finished; it is a site with one unanswered question, and that one question is what keeps the person from writing.
In our experience the one left unanswered is most often the price. Silence feels safe, because the price depends on the job, but for the reader it means not knowing whether they are asking about ten euros or ten thousand.
A range, a starting figure, or one honest sentence about what moves the price turns that into an answered question.
Where the enquiry quietly disappears

Now those last thirty seconds. Nothing below is anybody being stupid: each of these arrived for a reason that made perfect sense at the time.
Contact details only in the footer. The footer is where you scroll when you are already looking for something. A person still being persuaded is not looking for anything yet, they are reading. If the phone number appears only after a whole page of scrolling, it is available only to the people who had already made up their minds. A number in the header and one clear contact block at the end of every content page settles it.
A phone number that is a picture. On older sites the number is often drawn into the design or written onto the header image. On screen it looks exactly like a number. On a phone it cannot be tapped, on a computer it cannot be copied, and a screen reader speaks only whatever alt text somebody remembered to write - which on a header image is almost never the number. The same goes for an email address. This is one of the places where accessibility and sales are precisely the same thing.
How many fields a form should have
A form with eleven fields. Every field went on at somebody's reasonable request: sales wanted the budget, production wanted the deadline, somebody wanted to know where the client came from. Together they are a questionnaire. A first contact needs three things: who you are, how to reply to you, and what the trouble is. Everything else can be asked in the reply, once the person has started talking.
Nothing happens after sending. The page reloads, the form is empty, and there is not a word anywhere. One sentence saying the message arrived, somewhere visible rather than at the very bottom of the page, costs a couple of minutes of work.
Nobody says when a reply is coming. Somebody who has sent a message is left waiting with no idea whether to expect an hour or a week. In our experience, a confirmation that says when you normally reply is what stops people writing to a competitor on Wednesday because the silence got uncomfortable. The only condition is that the promise has to be true.
And then there is the worst case of all: a form that says thank you and sends nothing.
The form that never delivered

A website form does not send mail from your inbox. It sends through the site, and it has to satisfy the same checks as any other message.
The usual fault is that the form puts the visitor's address on the sender line, so your server sends a message claiming to come from a stranger and fails those checks in the clearest possible way. The fix is a one-line change: send from an address on your own domain and put the visitor's address on the reply line.
Usually it does not bounce and you see no error at all: it is accepted and then filed as spam, or refused so quietly that only the site's own mail log knows.
The important part is that the failure is invisible from both ends. The visitor saw "thank you, we will be in touch" and left satisfied. You saw nothing at all, because a message that never arrived is not visible anywhere. Two people, both certain the thing was done.
This is not rare, and it does not only happen to old sites. The form was tested on the day the site launched, and it worked then. After that the domain moved to different hosting, the mail provider changed, a plugin updated, or somebody added security settings to the domain that no longer permit the old way of sending. None of those changes touched the form. Any of them can lose the form's messages.
Why that happens, and what else can be wrong in the domain settings, is a subject of its own and has its own article. What matters here is only that you should not assume the form works. You should know it does.
One more thing is worth arranging: the form should save every enquiry inside the site as well, not only send it. That way the enquiry survives even when the mail does not, and a glance at the list tells you immediately whether any came in. It is also the only way the silence becomes something you can investigate: if there are saved enquiries but no emails, you know exactly where to look.
Trust is specific, not general
Most websites write exactly the same things about themselves. Professional. Reliable. Long experience. An individual approach. Because everybody writes that way, the reader no longer reads those words as words. They are noise, and the eye slides over them.
Trust is built only out of specific things, and they are usually the ones that feel too small to be worth the trouble:
- Real names and real faces, not a stock photo team in a glass-walled office.
- Photographs of the actual premises, even when the premises are not pretty.
- A few named jobs you have actually done, rather than "over a hundred satisfied clients".
Customer testimonials belong on that list too, but only when they carry a name, a company and one concrete thing that changed. "Very happy, highly recommend!" with no name is read as a sentence you wrote yourself, and it loses its effect exactly the way "professional and reliable" does. What makes a testimonial believable has an article of its own.
An honest note here: some of this work is what we sell, because sorting out the words and the photographs is one of our services. But the biggest piece of it you do yourself, for nothing. A real name, a real address and a few photographs you take yourself outweigh every word somebody else writes on your behalf.
The test you can run today

The most honest measurement here does not come from software. It is your own phone and half an hour.
- Take your phone and turn the wifi off. Your customer is not on your office network, and their phone has never loaded your site before, so nothing on it is stored and ready.
- Type the address by hand. Not from a bookmark, and not from a search where your own site comes first anyway. You have to arrive the way a stranger arrives.
- Pick yourself a problem. One real thing a customer would ask about. You are that customer now, and you want a quote for it.
- Look for a way to make contact, and start a stopwatch. How long until you find a number you can tap, or a form you can fill in.
- Actually send the enquiry. Not "test test", but the way a customer would write it, in your own words.
- Check that the message arrived. Including the spam folder. This is the most important step in the test, and the one people skip.
- Write down where you stalled. Where you had to scroll back, what had no answer, which field you did not know how to fill in.
Those two or three lines are what you came for. Not the stopwatch, the places where you had to think. They tell you more than a month of watching statistics, because they are the only places where you saw your own site through somebody else's eyes.
If you can borrow twenty minutes from somebody, have the test done by a person who has never seen your site, and watch without helping them. Those twenty minutes usually make everything in this article obvious.
What order to repair it in
Cheapest and most certain first, more expensive and less certain last:
- Make contact possible from every page. An hour of work, and it needs nobody's decision.
- Cut the form to three fields.
- Make the form confirm. A couple of minutes, and it keeps working on the days you do not.
- Make sure the messages arrive. Test it yourself, check the spam folder, and have the form's sender address checked first, then the domain's mail settings. This is the only item on the list that usually needs somebody else.
- Only then go near the words. Answer the five questions above, put the price out at least as a range, and take down "professional and reliable".
None of the first four touches your design or your message: they change the path, not the pitch. That is exactly why they are worth doing first: if the enquiries still do not come afterwards, at least you know for certain that the path was not the problem.
Some sites do not have an enquiry problem
This section argues against our own interest, but without it the rest would be dishonest.
Some sites do not lose enquiries on the way. They lose them earlier: the offer itself is not the thing people want, or it is worded so that nobody can tell what they would be buying. A roofer whose page sells "complete building solutions" is not losing enquiries to the form - somebody searching for a roof repair simply does not recognise themselves there. In that case a perfect contact page is a smooth road to an empty place, and the work starts with the content, not with the form.
And some sites get the wrong visitors. A page that ranks first for "what does a new roof cost" collects readers who were checking a price list, not shopping for a roof. Those people were never going to buy, and raising that number produces nothing except a bigger number.
No button repairs either of those. We would rather say so than sell a fix that cannot work.
Where to start
If you want a second pair of eyes, our free search engine audit checks one part of this from the outside: whether your contact details are readable as text on the page at all. It cannot fill in your form, and it cannot tell you whether the message arrived. That half is the test above.
And if the test produced a list you would rather not work through yourself, ask for a quote and put those stalling points in the message. Work can start from those. It cannot start from "nobody gets in touch".
So if you read this far and do one thing, take your phone off wifi tonight and send yourself an enquiry.
Frequently asked questions
The enquiry is usually lost at the very end of a visit, not at the start: after the person has decided they want to talk to you and is looking for a way to do it. Analytics does not show this, because there is no line reading “looked for contact details and found none” - those people look exactly like everybody else in the statistics. The fault itself is usually small and boring: contact details only in the footer, a phone number that is a picture, a form with eleven fields, or no confirmation anywhere after sending.
Five things, in roughly this order: do you do the specific thing I need, do you do it where I am, are you a real company with a name and an address, roughly what does it cost, and will a human answer me. These five are not weighted. A site that answers four of them is not four fifths finished; it is a site with one unanswered question, and that one is what stops the person writing. The one most often left unanswered is the price, even though a range or a starting figure would be enough.
A first contact needs three things: who you are, how to reply to you, and what the trouble is. Everything else can be asked in the reply, once the person has started talking. Long forms grow one reasonable request at a time - sales wants the budget, production wants the deadline, somebody wants to know where the client came from - and together they become a questionnaire nobody starts filling in for a stranger. What happens after sending matters just as much: a visible confirmation that the message arrived, and a promise about when you reply.
Not first. If the site is not producing enquiries from the visitors it has, it will not produce them from twice as many: buying more traffic multiplies the same result by a bigger number. A rebuild has a fault of its own - it moves everything at once, so afterwards nobody can say whether it was the colours, the words, or the phone number finally reaching the header. Both are worth putting behind the things you can check yourself, and that part is an hour of work rather than a budget.
Take your phone, turn the wifi off and type the address by hand rather than using a bookmark. Pick one real problem a customer would ask about, then look for a way to make contact and time how long it takes. Send the enquiry for real, in your own words, and then check that it arrived - including the spam folder. That last step is the one people always skip. Write down where you stalled. Better still, have the same test done by somebody who has never seen your site, and watch without helping.
Read next

The first screen says "Welcome to our website", and below it "Quality and reliability since 2011". One test you can run today: if a sentence stays true with a competitor's name in it, delete it.