How to build a landing page that converts: structure, copy and proof

A landing page converts when it matches the visitor's intent and the ad's promise. It explains one coherent offer, backs key claims with proof and makes the next step work on phones.

A dedicated URL, fewer fields and a stripped header sometimes help, without guaranteeing a better business result. This guide takes you from a campaign brief to a measurable page, with the folklore removed.

What should you decide before you build a landing page?

You need one when no existing page can keep the promise the campaign makes. In analytics, every page a visitor enters on is a landing page. In marketing, the word means a page built around one campaign, one audience and one offer. That's the meaning this guide uses.

The decision is suitability, not doctrine. Take a service page that already explains the offer, shows proof and carries a working enquiry route. It can be a perfectly good destination for a search campaign. A homepage usually cannot, because it serves everyone.

Our Google Ads landing page vs homepage guide covers that decision in detail. It includes the campaign settings that decide which page an ad opens. If the homepage is the right destination, writing a homepage that converts is its own method.

Write a campaign brief before anyone opens a design tool. It forces the six decisions a page is built on.

Message match is more than repeating the keyword. Change any one of the details below and the page breaks a promise. The visitor who clicked "same day hot water repair Geelong" expects all of them to line up with the ad:

• the service

• the place

• the price conditions

• the availability

• the next step

Traffic temperature changes how much explaining the page has to do. A high intent search wants the offer and the action quickly. A colder social audience needs context or a demonstration before the ask.

The standard pattern is one main conversion goal, repeated where the reader is ready. It also includes a phone number and a written route for people who can't use the form.

Then check that the ad opens this page. Final URL expansion is on by default once a Search campaign uses AI Max. New Search campaigns use AI Max from the start.

With expansion on, Google can swap your Final URL for another page on your site. Google does that when it judges the other page more relevant to the search. If the campaign exists for this page, turn expansion off or add URL exclusions. Otherwise the click can land on your blog.

Quality Score is the other Google Ads term that comes up here. It's easy to misread. Google's page About Quality Score calls it a diagnostic tool, not an input in the ad auction. The page also says it isn't a key performance indicator to optimise.

Landing page experience is one of Quality Score's three components. So a below average rating is a prompt to look at relevance and usability. It isn't the page's success metric. Our explainer on Google Ads Quality Score covers the rest.

Decide early whether the page should be indexed. A paid page that duplicates your service page competes with it in organic search. So give the query one owner. Keep a campaign wrapper out of the index. Let a page that answers a question nothing else owns index as that query's page.

Keeping a page out is one tag in the page head.

<meta name="robots" content="noindex">

Google's documentation on blocking indexing with noindex adds the catch people miss. The page must not be blocked by robots.txt, or Googlebot never sees the tag. The tag is a Search instruction. Your ads still run and Google Ads still checks the destination.

What should a landing page include?

A landing page should include nine content roles, in an order that follows the visitor's decision:

• understand the offer

• trust the business

• recognise the situation

• learn what is included

• see why it matters

• check the proof

• act

• resolve the last doubts

• act again

They're roles, not nine mandatory full width bands. A short urgent task can combine several. A complex offer may need more room for one.

A page's utility parts sit around these roles: the logo, essential header links and the footer. The footer carries your business identity, privacy information and terms. These parts aren't a tenth section.

Hero: make the offer clear

Name the service or offer, the audience or location it is for and a credible benefit. Add a subheading that says what is included, who it suits and what happens next. Then give one specific call to action. Keep the header to utility: a logo and a phone number. Nothing in it should pull the visitor away before they've read the offer.

Pick one formula. Stacking every placeholder into one headline produces a sentence nobody says out loud. The example deliberately skips a bond back promise, a same day claim and a free quote. Add those only when the real business can honour them.

Trust cue: reduce early uncertainty

Put one or two relevant, verifiable trust signals near the first decision. Think of a review count you can show, or a licence or qualification. Recognisable work you have permission to display counts too. The mistake to avoid here is decoration.

Fake star ratings, "as seen in" logos nobody agreed to and badges that mean nothing are not proof. Under Australian Consumer Law a false impression is a problem whether or not you meant to create one.

Problem: show you understand the task

Describe the customer's practical situation in their words. Leave out fear, exaggerated pain and the assumption that the ad was right about them. This section's job is recognition: the reader should be able to tell whether the offer fits. Repelling the wrong visitor here is cheaper than a wasted enquiry.

Offer: explain what is included

Spell out the offer:

• the deliverables

• who the offer suits

• what it excludes

• the real price or the drivers of the price

• any material conditions

Say exactly what an enquiry or booking commits the visitor to. "Get started" that quietly means a 40 minute sales call is a broken promise waiting to happen.

Benefits: connect features to outcomes

For each inclusion that matters, say why it helps the customer, in plain words. Don't promise an outcome the service can't guarantee. Don't restate the features list with adjectives attached, either.

Proof: support the important claims

Place evidence next to the claim it supports. Relevant work, a substantiated case example, a genuine testimonial and a product demonstration all count. Keep measured results and example copy visibly separate. Know the source, period and definition of any number you quote, because someone will ask.

Form: make the next step practical

Ask for what you need to reply and to qualify sensibly, no more. Say what happens after submission, who replies and a realistic response expectation. A phone call or a booking link can serve the same goal. For some tasks, it serves that goal better. The section on the enquiry path below covers the mechanics.

Objections: answer the remaining questions

Address the real concerns: price, timing, suitability, process and what the visitor is committing to. Use a question and answer format only where it helps someone decide. A FAQ block stuffed with keyword variations is padding. Readers can tell.

Final CTA: repeat the relevant next step

Summarise the offer and repeat the same primary action for the reader who is now ready. The footer underneath carries the business identity, privacy information and essential terms: available, and out of the way.

Swap the wireframe's email field for a phone field when a callback is the agreed contact route. Add a qualifying field only when the answer changes what you do next.

How do you write the headline, offer and proof?

Write the headline for the intent, not for the formula. A search for a specific service in a specific place wants both in the first line. A click from an ad that led with a benefit wants that benefit confirmed. A campaign built on removing a known frustration wants the frustration named.

The hero template's two formulas cover the first two cases. For the third, put the frustration and the fix in the same line. A fourth option, campaign match, is the ad's own words plus the outcome. It's the safest choice for a tight ad group.

Then make every claim supportable. The ACCC's guidance on false or misleading claims says claims should be true and accurate. They should also be based on reasonable grounds. So "free quote", "fixed price" and "we reply within the hour" are allowed exactly as long as they're true, qualified where needed and delivered. Our guide on how to write website copy covers the full claims check.

A button should describe what happens next. "Request a quote" and "Choose an appointment" do. "Submit" describes a form, not an outcome.

Three example situations show the difference. Each is an improvement to clarity or usability. Whether it produces more enquiries is a hypothesis to measure, not a claim.

The after copy names the service, the audience and the place. It says what the business needs to quote and describes the action. Nothing in it promises a bond back, a time or a price the business hasn't committed to.

The other two situations work as corrections. So does a navigation mistake that often travels with them.

None of those rows carries a percentage, because the effect depends on your visitors and your offer. Tighter forms can reduce completions among the wrong visitors as well as the right ones. The section on measuring useful conversions tracks exactly that trade.

How do you make the enquiry path easy to use?

The enquiry path starts when the visitor decides to act. It ends when a person in your business picks up a real enquiry. Most of that stretch has nothing to do with design.

Fields come first. There's no universal four or five field optimum. Instead, remove what you don't need at this stage and keep the qualification that changes your response. Ask for the contact route you'll use to reply.

Labels and errors are where forms can quietly lose people on phones. So give every field a visible label rather than placeholder text. Mark the required ones and use the right input type and autocomplete. When a submission fails, name the field, describe the error plainly and keep the person's valid entries. Our guide to website enquiry form design covers field choices, labels and errors in full.

Confirmation and delivery are where a working page can still lose the lead. Our audit sample is 100 website audits from May 2023 to September 2026. In it, a clearly labelled test enquiry had one business hour to reach the person who answers it. It failed on about 1 in 5 sites with an enquiry form. A spam folder, a wrong notification address, a stopped integration or a queue nobody owns is enough.

So the page should confirm actual acceptance, not a click. Then someone checks that the enquiry reaches the inbox or CRM. A named person owns the follow up, with a fallback for the hours nobody is at a desk.

The measurement event is a separate piece of work, tied to the accepted state. It fires only when collection is permitted. It must not count a lead because a button was clicked. A thank you page that anyone can reload is not proof of a new enquiry either. That's why the flow ends with a person, not a page view.

Mobile and accessibility are tested together, on real devices. Walk the whole path with the keyboard and zoom the text. Check that focus is visible and touch targets are big enough to hit with a thumb. Our guide to website accessibility in Australia covers WCAG target sizes and the other checks.

Then open the on screen keyboard, the sticky CTA and the consent banner together. That's how a visitor meets them. Make sure none of them covers the active field or the action.

Keep a written route for people who can't or won't call. Sometimes a call is the right primary action. For those pages, see our guide to getting more phone calls from your website. It covers tap to call and call handling.

How do you handle speed, navigation and privacy?

Performance matters, and the folklore around it does not. Compress imagery to the size it is displayed at. Remove blocking assets and intrusive overlays. Keep the demonstration that helps the visitor decide. Judge the result in the field, not in a lab.

Google's Web Vitals guidance says lab measurement is not a substitute for field measurement. A standard lab load test, like PageSpeed Insights' Lighthouse run, cannot measure Interaction to Next Paint. That's because nothing interacts with the page during the test. Our Core Web Vitals explainer covers the thresholds and how to read field data.

A 100 lab score doesn't prove conversion quality. The three second rule was never a rule either. If a page feels slow on a phone, start with our guide on why your website is slow.

Navigation and trust are a balance, not a purge. Reduce unrelated distractions where you can justify it. Keep business identity, necessary terms, privacy information and useful help. A page that traps the visitor looks like exactly what it is. The Google Ads destination requirements already disapprove destinations that are unnecessarily difficult to navigate.

Privacy closes the loop. Explain the relevant data use close to the form and keep the full privacy information accessible. Our guide to privacy policies and website terms covers that part. Then look after the form and the data it collects:

• collect the minimum

• protect the form

• handle spam with a method that doesn't punish real people

• check every third party integration the form feeds

The OAIC published guidance on tracking pixels and privacy obligations in November 2024. It expects covered organisations to do due diligence on third party tools. It also expects them to limit collection and disclose pixel use. The guidance says sensitive information should only be collected through a pixel with express consent.

Whether the Privacy Act covers your business at all is a separate question. So is what the reforms change. Our guide to Australian privacy law changes for websites covers both.

Whatever law applies to you, Google Analytics' guidance on avoiding personally identifiable information still applies. It says Google's policies ban passing Google any data it could use or recognise as personally identifiable. So names, emails and phone numbers stay out of event parameters, page titles and URLs.

Before launch, map where each enquiry's details travel and who holds the login for each system. Personal details belong in the inbox and the CRM. GA4 gets the event, not the person.

Which platform should you use, and what will it cost?

There are three routes. The right one depends on the job rather than the vendor's pitch. A builder isn't automatically faster, cheaper or better at converting. Your existing template isn't automatically worse.

The routes overlap, because WordPress is often the existing site. The real distinction is between three options. You can adapt your existing template, commission a dedicated one or pay for a separate specialist editor.

On WordPress with a block theme, the Template Editor lets you create a custom template. You can then assign it to a single page. That's how a landing page drops the standard header and footer. The rest of the site stays untouched. Classic themes only expose the editor if the theme enables it.

Builders are a different proposition, and the plan matters. A builder may offer a controlled split you set yourself, plus automated traffic routing on some plans. The testing section below explains why the difference matters.

If the platform decision is wider than one page, settle it before anyone builds.

On price, we can only speak for ourselves. Our landing pages are priced from $800 to $2,000 excluding GST, and ongoing testing is a separate cost.

Our web design Melbourne page lists those prices. It adds that each project is scoped individually after a discovery call. Hosting, the domain and maintenance are separate. The range is a published Elev8d indication, not an Australian market benchmark. Scope moves the quote.

The same page says a landing page can be ready in about a week. A delivery date still needs its own scope check. Whoever you brief, clarify the components below, because they're where quotes quietly differ.

For the wider numbers, see our guide to how much a website costs. It covers full sites, from the build to the running costs.

How do you measure useful landing page conversions?

Define what counts before you look at a report. A landing page can look brilliant at the first stage while the business sees nothing at the last. These are six stages, not synonyms:

• a button click

• an accepted enquiry

• an answered call

• a confirmed booking

• a qualified lead

• a sale

A phone tap does not establish a call. A booking handoff does not establish an appointment. Confirm receipt in the business system, record qualification separately, and only then talk about valuable leads.

A few cautions travel with that table. The first rate is a business measure you define, not GA4's session key event rate. That rate counts sessions and the one key event you select. Qualification needs one rule, a matching attribution scope and time for outcomes to mature.

Keep the build cost out of cost per enquiry unless the calculation says it is in. Repeat contacts and reporting delays distort quick comparisons. Never read revenue as profit.

GA4 gives you two of the pieces. The Landing page report shows the first page of each session with sessions, key events and engagement. It's session scoped. So it shows which entry pages are associated with outcomes in those sessions. It won't show where a later action happened.

The Traffic acquisition report shows session key event rate by default channel. It defines that rate as the percentage of sessions where a user triggered a key event. Source, medium and device are the useful comparisons.

Then test the real form and the intended event, checking for duplicate firing. Confirm which Ads conversion is used for reporting and bidding. That way, an imported GA4 lead and a native Ads lead aren't both counted as primary outcomes.

The setup lives in our guides to GA4 for small business and Google Ads conversion tracking. Where paid calls are the goal, call tracking for Google Ads is a separate job.

A visitor's privacy choice, a browser restriction or a disconnected booking system can limit tracking. When that happens, write the gap down. Never defeat a consent choice to collect a conversion event.

As for benchmarks, a borrowed figure rarely shares your traffic, offer or definition of a conversion. It's context at best, never a target.

We treat 5% to 6% of clicks becoming enquiries as standard. With enough conversion history and a landing page built for the campaign, 8% to 9% is realistic. Below 3% is worth investigating. Those rates are for service lead generation from high intent paid traffic. Our guide to what a good conversion rate is explains them.

Even then, the baseline that decides anything is your own. It needs matching traffic intent, offer economics and lead quality rule, plus enough sample to mean something. The example below shows why volume alone cannot settle it.

Both pages receive 1,000 measured visits and $2,000 in attributed ad spend. Each enquiry is counted once. Qualified enquiries are a subset of all enquiries. So the bars are counts, not percentages or additive categories. Page A wins on raw rate and loses on everything the business is paid for.

The lesson is in the last row. The page that invites more people to enquire can cost more per real customer. That's the mistake a "conversion rate" headline hides.

How should you test and improve the page?

Fix the faults first. A broken form, a false claim or an action a keyboard can't reach gets corrected today. None of them needs an experiment. There's nothing to learn from testing a defect against itself.

Experiments are for uncertain improvements. Write a hypothesis, then choose a control and a variant. Spend the effort on what moves decisions: the offer, message match, proof, qualification and the enquiry path. Button colour is where testing budgets go to feel productive.

A controlled A/B comparison assigns eligible visitors randomly and consistently to concurrent variants. Kohavi and colleagues describe that persistent split in Online Controlled Experiments at Large Scale. Under that split, a user gets the same experience on every visit. Before the test starts, agree:

• the primary outcome

• the effect worth detecting

• the sample you need

• how you'll decide

Duration depends on traffic and how often the outcome happens. Your business cycle and how long outcomes take to mature matter too. Seven days or 100 visits isn't a stopping rule.

Say you check the result every morning and stop the first time it turns positive. That manufactures a winner that doesn't exist. Kohavi, Deng and Vermeer's A/B Testing Intuition Busters covers common testing misunderstandings. The fix is the plan you agreed before launch. Stop when it says, not when the chart looks good.

Run a business quality guardrail alongside enquiry volume. Keep targeting, offer, budget and follow up stable where you can. Record every change you can't avoid. A redesign can be tested as a bundle. But a bundle can't tell you which edit did the work.

Know which mode your tool is running. A builder's A/B mode is a controlled split. An automated mode routes each visitor to the variant the system predicts will convert. Useful as that can be, it isn't a random comparison. So don't read its reports as one.

Many small service business pages don't have the traffic for an experiment. When traffic can't support one, improve the page with evidence of a different kind. Use the questions customers ask on the phone and a usability check with someone outside the business. Confirmed technical faults count too.

Report a before and after comparison as observational. Name the other explanations, seasonality and a changed traffic mix among them. "Inconclusive" is a valid result. It's a better one than a false win you'll spend the next quarter defending. Plan the test before anyone builds the variant.

Leave the uplift blank until there's a result. The allocation and the statistics belong in the plan. The plan belongs to the person who has to defend the result.

Sometimes the evidence points past one page to the whole enquiry journey. Read why your website is not converting first. Then look at conversion optimisation across the journey. Don't keep testing headlines on a page that isn't the problem.

What should you check before launch?

Work through the checks below and treat any failure in enquiry delivery or access as a launch blocker. There is no score. A page with every cosmetic item complete and a submission path that fails on Android is not ready.

Those checks cover the page. The site around it needs its own pass before a campaign starts sending people there.

What we recommend at Elev8d

FAQ

Next steps

Written by Ajay K.Co-founder, Elev8d

Ajay is the co-founder of Elev8d and writes most of what you will read here. Psychology grad turned marketer, which is a fancy way of saying he cares about why people click, not just where things rank. Every guide is written in plain English for Australian businesses, tested on real campaigns before it earns a spot on this site.