Conversion optimisation: turning visitors into enquiries

"Improve conversions" is not a task anybody can start on Monday. "Three quarters of visitors leave the services page without scrolling" is, and so is "the contact form has been failing silently since the hosting move".

The difference between those is knowing where in the journey people are being lost. This article is a way of finding that out, stage by stage, with the fixes that belong at each one and not at the others.

Before anything: are you recording it?

Start here, because a surprising number of sites in this situation are converting perfectly well and cannot see it. Phone calls, WhatsApp messages and direct emails do not appear in analytics unless somebody set them up to.

1. 1Submit your own contact form and confirm the email arrives. Contact forms fail silently more often than anyone expects, usually after a hosting or DNS change.
2. 2Check that the submission is recorded as a tracked event, not just that the email works.
3. 3Make phone numbers and email addresses real links, and track the taps. On a phone, a number that cannot be tapped will not be dialled.
4. 4Record the page the visitor first arrived on, not only the page they converted from. Informational pages rarely convert directly and frequently start the relationship.
5. 5Ask your last five customers how they found you. If several say the website and analytics shows nothing, you have a measurement problem rather than a conversion one.

Test the form before rebuilding the page around it. It is the least interesting explanation and it is true more often than it should be.

The five stages, and what failure looks like at each

Every enquiry passes through the same sequence. Losing people is normal at every step; losing almost all of them at one particular step is the thing worth finding.

Stage 1 — Visitor: the right people at all?

List your ten most common entry pages and sort them into two groups: pages for people solving a problem themselves, and pages for people choosing who to hire. If eight of ten are in the first group, you do not have a conversion problem. You have an audience that was never buying today, and no amount of button work will change it.

That is not a failure, but it is a different job: it means writing content aimed at the choosing stage, and connecting the existing content to it. Traffic but no customers is the full diagnosis, and it is worth running before anything below.

Stage 2 — Engagement: past the first screen

A visitor arriving from search gives you a few seconds to establish that they are in the right place. Three things decide it, and the first is the one most sites get wrong.

• Clarity. The first screen must say what this is, who it is for, and what to do next. Headlines like Innovative solutions for modern business describe nothing and would be equally true of every competitor.
• Speed. Aim for Largest Contentful Paint at 2.5 seconds or less at the 75th percentile of real visits, and Interaction to Next Paint at 200 milliseconds or less.
• Stability. Cumulative Layout Shift at 0.1 or less. A page that moves while loading makes people tap the wrong thing, and on a form that is fatal.

A five-second test settles the first one: show the top of the page to somebody outside your industry, then ask what the company does and who for. If they cannot answer, neither can your visitors. High bounce rate covers when short sessions are a real signal and when they are not.

Stage 3 — Intent: reaching a page where action is possible

This is the stage nobody instruments and almost everybody leaks at. An article answers a question well, the reader is satisfied, and there is no path from there to the page for people who would rather have it done.

• One relevant link from each content page into a commercial page, placed where the reader has just understood how much work is involved.
• Anchor text describing what they get rather than the page's name.
• Not in the introduction. Somebody two paragraphs in has not decided anything yet.
• Proof on the destination page: named clients, real photographs of real work, a case with a before and an after. Anything a dishonest competitor could not copy without lying.

Stage 4 — Action: starting the form or the call

The form is where the money is and it is usually the least considered element on the page. Every field is a small negotiation, and people abandon negotiations.

• Name, a way to reply, and space to describe the problem. That is a complete first enquiry for most businesses.
• Never ask for something you will not use in your reply. Budget dropdowns lose more good enquiries than they filter bad ones.
• Labels above the input, not placeholder text that vanishes the moment somebody types.
• Validate as they go rather than on submit, and never clear a form on an error.
• Say what happens next: we reply within one working day, with a name attached.
• One primary action per page. Six things to do is the same as no obvious thing to do.
• The confirmation is part of the design. A blank page after submitting reads as a failure, and people either send it again or give up.

For traffic arriving from a specific campaign, go further and remove the navigation. Somebody who clicked an advertisement for one service does not need a menu offering the other eight, which is the entire reason a dedicated landing page converts better than sending the same traffic at a homepage.

Stage 5 — Lead: arriving and being answered

Two failures live here, and both are unglamorous. The submission never reaches a human, or it reaches one slowly enough that the person has already hired somebody else.

Test the delivery path monthly rather than assuming it. Check the address the form sends to still exists, that the messages are not landing in spam, and that somebody owns replying to them. A form that works and a reply that takes four days are not the same as a working pipeline.

Working out whether it is worth it

Do the arithmetic before committing a week. Sessions, times the share who make contact, times the share you close, times what a customer is worth.

For business services sites, somewhere between one and three percent of visitors making contact is a commonly observed range rather than a law. Under one percent usually indicates a real fault. Above five percent generally means low traffic that is extremely well targeted, which is a good position rather than a suspicious one. Use your own numbers; these are only a sanity check on whether the effort pays for itself.

What matters less than people think

• The colour of the button. Contrast and placement matter; the specific hue is where redesigns go to argue.
• Carousels. Visitors interact with the first slide and almost never the rest, so the second slide is an element nobody sees.
• Page length. A long page that answers the question outperforms a short one that does not, and the reverse is equally true. Length is not the variable.
• Looking different from competitors. Familiar layouts convert well because people already know how to use them.

If the constraint turns out to be the structure of the site rather than any single page, that is a design question rather than a copy question. Designing a website that generates leads covers the structural version, and our website design work exists for the case where the current site cannot get there.

Questions people actually ask

Where should I start if everything looks weak?

Stage 5, then stage 1. Confirm enquiries are arriving and being recorded, then confirm the traffic is people who could buy. Those two take an afternoon and they decide whether the other three stages are worth touching.

How much traffic do I need before testing changes?

More than most small sites have, which is why formal split testing is usually the wrong tool here. With a few hundred sessions a month, judgement and obvious fixes beat statistics. Testing becomes worthwhile when a one percent difference is worth measuring.

Should I publish prices?

Something about price, yes. A starting figure, a typical range, or a worked example. Refusing entirely loses more good enquiries than it filters, because people assume the worst and leave without asking.

Do chat widgets help?

When somebody answers them promptly, yes. When they are an unmonitored bot that cannot answer, they add page weight and frustration. Measure whether yours produces enquiries before assuming it does, and load it on click rather than on every page view.