How to Build a Website That Converts: A Practical Guide

We get asked to redesign sites that don't convert, and about half the time the design isn't the problem. The site looks fine. A visitor lands, reads a headline that could belong to any company in the sector, can't tell what happens next, and leaves. That's a clarity failure, and no amount of new typography fixes it.

Here's the sequence we actually follow, in order, with what we measure at each step.

Give the page one job

A page with four calls to action has none. Ask what a visitor should do next and write down a single answer. Book a call. Start a trial. Add to cart. Request a quote.

Everything else on the page is either supporting that action or competing with it. Newsletter signup boxes, chat widgets that open on their own, "follow us" icons above the fold — each one is a fork in the road at the moment you wanted a straight line.

This gets harder for a homepage serving three audiences. The answer isn't to serve all three equally; it's to pick the most valuable one for the primary path and give the other two a clearly signposted route to their own page. A site that tries to convert everyone converts nobody in particular.

Audit the objections before you write anything

Sit with whoever answers the phone at your company and ask what people ask before they buy. You'll get a list of six to ten questions. That list is your page structure.

For most of the businesses we work with the list looks something like this:

• What does this cost, roughly?
• How long does it take?
• Have you done this for someone like me?
• What happens if it goes wrong?
• Who exactly will I be dealing with?
• What do you need from me?

Pages that answer those questions convert better than pages that describe features, because those are the questions standing between a visitor and a decision. If your pricing genuinely can't be published, publish a range and the variables that move it. "Most projects land between $4,000 and $12,500 depending on page count and integrations" is infinitely more useful than "contact us for a quote", and it filters out the enquiries you'd have declined anyway.

Structure the page as an argument

Read a page that converts well and you'll find a sequence rather than a collection of sections. It usually runs:

1. What this is, in one sentence a stranger understands
2. Who it's for, specifically enough that the wrong person self-selects out
3. Proof that it works — a number, a name, a case
4. How it works, in three or four steps
5. The two or three objections that stop people, answered directly
6. What happens when you get in touch

That's the argument. The design's job is to make each step legible at a glance and to keep the reader moving down it.

One structural detail worth more than it sounds: put a call to action after the proof section, not only at the bottom. Some visitors are convinced by the third scroll and shouldn't have to hunt for the button.

Write the headline last, and write it about the reader

Headlines written first tend to describe the company. Headlines written after the page exists tend to describe the outcome, because by then you know what the page is actually promising.

Two versions of the same headline from a project last year:

Before: "Innovative solutions for modern retail."

After: "Your stock count is wrong in two places at once. We fix that."

The second one loses everybody who doesn't have that problem, which is the point. Enquiries dropped about 15% and qualified enquiries roughly doubled.

Use proof that survives scrutiny

Logo walls do very little on their own. A visitor can't tell whether you built a client's whole platform or fixed their favicon. What works is specific: a number, the context around it, and ideally a name attached.

"Checkout completion up 31% in ninety days" is proof. "Trusted by industry leaders" is decoration. If you don't have numbers yet, go and get them — email three past clients and ask what changed. It's a week of chasing and it's the highest-value week in most redesigns.

Where a client won't share numbers, describe the situation and the work honestly without inventing an outcome. An empty claim costs you more credibility than a missing one.

Design the form like it costs money

Because it does. Every field is a small tax on completion, and most forms are collecting things the sales team could ask on the call.

What we do:

• Cut every field that isn't needed to route or price the enquiry. Company size, job title, and "how did you hear about us" are usually the first three to go.
• Label fields above the input, not inside it. Placeholder-only labels disappear the moment someone starts typing and are a genuine accessibility failure.
• Validate on blur, not on submit, and write errors that say what to do — "Enter an email address like name@company.com", not "Invalid input".
• Tell people what happens next. "We reply within one working day, usually with two or three questions" sets an expectation and reduces the number of people who submit twice.

There's one exception to the shorter-is-better rule. For high-value, low-volume services — legal work, complex builds, anything with a five-figure minimum — a longer form filters. If someone won't spend four minutes describing their matter, the call was unlikely to go anywhere. We've seen in-scope enquiry rates go from a fifth to seven in ten on the back of a longer, better-structured intake form.

Fix speed before you fix anything visual

Conversion work on a slow site is painting over damp. On mobile, over real network conditions, the difference between a 1.5-second and a 5-second largest contentful paint is a large share of your traffic never seeing the page you optimised.

The usual culprits, in the order we find them:

• Unoptimised hero images — a 2MB JPEG where a 90KB AVIF would do
• Third-party scripts, especially two analytics tools loading the same library
• Web fonts loading without font-display: swap, blocking text render
• Layout shift from images and embeds without reserved dimensions

None of that is glamorous and all of it moves the number. On one store we removed four apps and took mobile LCP from 5.4 seconds to 1.9, which produced more conversion lift than any design change in the same project.

Instrument it, then leave it alone for a month

Set up events for the actions that matter — form submits, calls, add-to-cart, trial starts — and verify they fire correctly before launch. Half the analytics setups we inherit are quietly broken, usually double-counting or missing the mobile path entirely.

Then resist the urge to change things weekly. Traffic is noisy, and a fortnight of data on a mid-sized site tells you almost nothing. Give it four weeks, look at where people leave, and fix that one step.

Two numbers worth watching beyond conversion rate:

• Scroll depth on the primary page. If most people stop at 40%, your argument breaks at 40%.
• Enquiry quality. Track how many enquiries are in scope, not just how many arrive. Volume can rise while revenue falls.

The honest part

A conversion-focused rebuild takes six to ten weeks for a typical business site, and most of the first three weeks is research and writing rather than design. Anyone quoting two weeks is skipping the objection audit, and that's the part that does the work.

You also won't get one clean before-and-after number, because you'll change several things at once. That's a reasonable trade — sequential testing on a site with 20,000 monthly sessions would take a year to reach significance on a single element. Change the whole page, measure the whole page, and be honest about the attribution.

If you want a second opinion on where your current site is losing people, send us the URL. We'll tell you what we'd change first, and it's frequently cheaper than a rebuild.