Landing Page Best Practices That Survive a Real Test: Graded by Evidence, Not Repetition

Landing page best practices are the habits widely recommended for pages built to turn a visit into one specific action. Most are repeated more often than they are tested. This guide grades each one by what it rests on, so you know which to adopt, which to verify and which to ignore until a test says otherwise.

The grades are simple. Tested means someone measured it and published a method we could read; it may be a controlled experiment or an observational study, and we say which. Consensus covers written standards, platform documentation and credible, widely shared advice, where we could not open a test of the specific claim. Folklore means the advice is repeated everywhere and we found no source we could open that tests it. Folklore is not a verdict that the advice is wrong. It means the claim has not earned your trust yet.

The table below is the working core of this guide. Read the third column before the second: it says what the grade rests on. Where a practice leans on a single study, the study's scope matters more than its headline.

As closely as the person's question demands, and no closer. Google evaluates ads on three components that make up Quality Score: expected click-through rate, ad relevance and landing page experience. It describes landing page experience as how relevant and useful your landing page is to people who click your ad, and it rates each component as above average, average or below average against other advertisers whose ads showed for the same search over the last 90 days. Google also says Quality Score is a diagnostic tool rather than a key performance indicator, and that it is not an input to the ad auction.

That makes message match a Consensus practice, not a tested one. Google tells you relevance is assessed and does not publish the mechanism, so do not tune a page to a score. Tune it to the person. The headline should echo the job they searched for, the offer should be the one the ad described, any price or qualifier in the ad should appear on the page, and the page should speak to the audience the ad was aimed at. Our guide Keep the Promise Intact From Ad Click to Landing Page covers how to run message match across intent, claim, audience, offer, evidence, qualifiers and call to action without cloning a page for every keyword.

Start with usability, not length. A CHI paper by Seckler and colleagues, as reported by Nielsen Norman Group, found that forms following basic usability guidelines took less time to complete and were submitted without errors on the first try 78% of the time, against 42% for forms that violated them. That is the best-evidenced finding in this guide, and it says something specific: the way a form is built matters at least as much as how many fields it has.

On length, Nielsen Norman Group advises cutting any field whose answer can be derived another way, collected later or left out, and says each removed field raises conversion. It also reports reducing one of its own forms from six fields to two. Take that as strong advice from a credible source, not as a measured law, because the article gives no test of the per-field claim. There is also a trade-off the advice does not cover: fields that help you route or qualify a lead, such as service type or location, may cost conversions and still earn their keep.

So judge a form change on two numbers: submissions, and the share of those submissions that become real conversations or opportunities. A shorter form that doubles junk submissions has not improved anything. Our guide to designing lead-generation pages people can trust covers the exchange the form asks for, and why people hesitate to make it.

Google's published Core Web Vitals give three thresholds, assessed at the 75th percentile of page loads across mobile and desktop: Largest Contentful Paint within 2.5 seconds, Interaction to Next Paint of 200 milliseconds or less, and Cumulative Layout Shift of 0.1 or less. These are the targets you can measure and be held to.

The conversion case rests on a Deloitte study published in March 2020. It analyzed four weeks of mobile site data from retail, travel, luxury and lead-generation brands across Europe and the US, and reported that a 0.1-second improvement in mobile site speed went with retail conversions up 8.4% and travel conversions up 10.1%. For lead-generation information pages, it reported the bounce rate improving by 8.3%. The study describes a correlation between speed and funnel progress, observed on naturally occurring improvements, not an experiment. Do not copy the percentages into a forecast for your own page; use the study as support for the direction and then measure your own.

Measure with field data from real visits, then ask two questions of your slowest pages: what delays the largest visible element, and what blocks the page from responding to a tap. Those are the two things the thresholds measure, and the answers tell you what to fix first.

Show proof you can stand behind, or say less. This is not a conversion claim; it is a publishing rule. A testimonial you cannot attribute, a logo you do not have permission to show, or a statistic with no source weakens a page the moment a reader checks it, and unsupported endorsements can carry legal risk in some markets, so check the rules where you sell. If you have little proof, use what is real and checkable: your process, your terms, how long a reply takes, what a first conversation involves. Say plainly what you cannot yet show.

Accessibility is the one area where the rules are written down and testable. WCAG 2.2 success criterion 3.3.2, a Level A requirement, says labels or instructions must be provided when content requires user input, so that people know what information to enter. Success criterion 1.4.3 requires a contrast ratio of at least 4.5:1 for text, or 3:1 for large-scale text, with exceptions for logos and decoration. A button with pale text on a colored fill or a form that relies on placeholder text alone breaks these and is also harder for everyone to use on a bright screen.

We rate these Consensus because they are written standards, not because anyone disputes them. We treat them as non-negotiable for a simple reason: a form a person cannot complete produces no leads.

The page does not end at the submit button. Show a confirmation that says what happens next and when, in plain words: who will reply, by which channel and within what time. Then keep that promise. A confirmation that says 'we will be in touch soon' sets no expectation anyone can hold you to, and a stated window gives your own team a number to meet and your reports a number to check.

Count the conversion at the confirmation state, not at the click on the submit button, so rejected and failed submissions do not inflate the total. Then follow the lead: the page's real result is a conversation, and the page cannot control the speed at which it starts. Our guide Speed to Lead Without the Five-Minute Theater covers how to measure inquiry age and staff a realistic first response.

In a Harvard Business Review article, Ron Kohavi and Stefan Thomke tell how a Microsoft employee's idea to change how Bing displayed ad headlines was rated low priority and left for more than six months. When an engineer finally ran an A/B test, the change raised revenue by 12%, which they say would come to more than $100 million a year in the United States, with no harm to key user-experience metrics. The lesson is about ranking ideas: nobody in that building could tell which idea was good. It is one case, not a base rate, and it does not tell you most tests win.

Run your own tests with the decision written down first. Our guide A Landing-Page Test Is Useful Only If Someone Knows What Happens Next sets out the decision, contrast, assignment, outcome, guardrails and stopping rule to define before traffic enters an experiment. The rules below are our judgment on top of it, not a research finding.

Use the list below before a page takes paid traffic. It is built from the practices that graded Tested or Consensus above, in the order the visitor meets them, and it leaves out the folklore on purpose.