The Landing Page Blueprint
A good landing page asks for one action and earns it in a fixed order: a headline that names the outcome, a short line saying who it is for, a visual that shows the product, one call to action, proof, benefits, answers to objections, and the same call to action again. Keep it fast, keep the form short, and change one element at a time.
Everything below expands that sequence into decisions a designer can make on an actual page. Where the guide gives a number (a performance threshold, a target size, a sample size), it comes from a primary source or from arithmetic with stated inputs, listed at the end. Where something depends on your offer and audience, the guide says so rather than quoting an industry average. Conversion rates in particular vary too much by offer, traffic source and market for a generic benchmark to tell you if your page is good, so you will not find one here.
Start with the brief, not the layout
Before any wireframe, write four lines:
- The single action. Start a trial, book a call, download a file, join a waitlist. If two actions feel equally important, you have two pages.
- The visitor. Where they arrive from (an ad, a search result, an email) and what they already believe when they land.
- The promise. The outcome they get, in their words.
- The main doubt. The reason a qualified visitor would still say no: price, effort, trust, timing.
Those four lines decide most of the page. The source of traffic matters more than designers expect. A visitor from a search ad for "invoice template for freelancers" already knows what they want and needs a fast route to it. A visitor from a cold social post needs more context before any form makes sense. Same product, different page length, different proof.
The clarity check: can a stranger say what it is?
The first screen has to answer three questions without scrolling: what is this, who is it for, and what should I do now. A simple way to test it uses no tools at all. Show the first screen to someone who has never seen the product, for about five seconds, then hide it and ask them those three questions. If they cannot answer, fix the hero before anything else, because no later section can rescue a visitor who left during the first screen.
The check works best with people close to your real audience, and it is quick enough to run on every headline draft. Write down the answers word for word. When a tester describes the product in phrases you did not use, those phrases are often better headline material than yours.
Section by section
| Order | Section | Its job | What it needs | Typical failure |
|---|---|---|---|---|
| 1 | Headline | Name the outcome | A specific result, plainly worded | A product name or slogan with no meaning on its own |
| 2 | Subheadline | Say who it is for and how | One or two sentences | Repeating the headline in other words |
| 3 | Hero visual | Show the product or result | Real screenshot, product photo or output sample | Abstract stock imagery |
| 4 | Primary call to action | Offer the next step | One button, outcome-worded | Two buttons of equal weight |
| 5 | Proof | Make the promise believable | Named testimonials, customer logos, verifiable numbers | Anonymous praise |
| 6 | Benefits | Turn features into results | Three to six items | Feature lists with no "so what" |
| 7 | How it works | Show effort is low | Three steps | A wall of setup detail |
| 8 | Objections | Remove the main doubt | Pricing, FAQ, guarantee or trial terms | Questions nobody actually asks |
| 9 | Final call to action | Catch convinced readers | The same button and wording as the first | A different offer at the bottom |
| 10 | Footer | Legal and contact basics | Privacy, terms, contact | Full site navigation that leads visitors away |
Headline and subheadline
A headline that names an outcome usually beats one that names a product, because the visitor does not know the product yet. Compare these rewrites for a fictional invoicing tool:
| Draft | Problem | Rewrite |
|---|---|---|
| "Introducing InvoiceFlow 2.0" | Says nothing to a newcomer | "Send a client-ready invoice in under a minute" |
| "The future of billing" | Unfalsifiable, could be anything | "Invoices, reminders and receipts for freelancers" |
| "Powerful invoicing for everyone" | No audience, no result | "Get paid faster without chasing clients by email" |
The rewrites are short, but length is not the goal. A clear 14-word headline is better than a vague 6-word one. The subheadline then answers whatever the headline leaves open, usually who the product is for and the mechanism ("Create an invoice from a saved template, and InvoiceFlow sends polite reminders until it is paid.").
Promises in a headline are claims. "Under a minute" should be true for a typical user, and if it is not, choose a claim you can support.
Hero visual
Show what the visitor will get: the interface, the physical product, or a sample of the output. A screenshot of a real dashboard communicates more than an illustration of people pointing at a laptop. If you use product mockups, the mockup best practices handbook covers framing and device choice, and if you need supporting photography, check its license first; the guide to free stock photos for commercial use explains which licenses allow use in ads and landing pages.
The hero image is also the most likely Largest Contentful Paint element, which ties it to performance. That is covered below.
The call to action
The button label should say what happens next in the visitor's terms: "Start my free trial", "Get the template", "Book a 20-minute demo". "Submit" and "Learn more" make the visitor guess.
Design rules that come from accessibility requirements, not taste:
- The label is text, so WCAG 2.2 success criterion 1.4.3 asks for 4.5:1 contrast between label and button fill (3:1 if the label is large-scale text, roughly 24px regular or 18.5px bold).
- The button needs a visible boundary of 3:1 against the page when its shape is what identifies it, under 1.4.11.
- The target must be big enough. Criterion 2.5.8 Target Size (Minimum), Level AA, sets 24 by 24 CSS pixels as the floor for pointer targets, with exceptions for adequately spaced small targets and for links inside sentences. Primary buttons are usually far larger, but secondary links such as "See pricing" under the button can fall short.
The accessible color systems guide has a tested set of button color pairs if you need a starting point.
Repeat the same call to action, with the same wording and destination, after the proof and at the end of the page. A different offer at the bottom ("Or subscribe to our newsletter") splits attention between two goals.
Proof that holds up
Proof works when a visitor can tell it is real. Useful forms, roughly in order of credibility: a named customer with role and company, a recognizable customer logo used with permission, a specific result tied to a named customer, a count you can verify from your own records, and an average rating with its source and number of reviews.
There are legal limits as well as persuasive ones. The US Federal Trade Commission's guidance on its Endorsement Guides says an endorsement must reflect the honest opinion of the endorser; that a material connection (payment, free product, employment, family ties) should be disclosed clearly and conspicuously when the audience would not expect it; and that when a testimonial shows results that are not typical, the advertiser must make clear what results people can generally expect. Rules differ by country, so check the ones that apply to your market. In practice:
- never write a testimonial yourself or edit one until it says something the customer did not;
- keep a record of permission for every quote and logo;
- add a short note when a featured customer received something for the review;
- do not show a standout result as if it were normal.
Placement matters too. Put the strongest proof right after the hero, where visitors who will never scroll still meet it, and put specific proof next to the final call to action, where doubt peaks.
Benefits and how it works
Write each feature as the result it produces. "Automatic reminders" is a feature; "Late invoices follow up on themselves" is the result. Three to six benefits is enough for most offers. Each needs a short sentence of explanation, ideally with a concrete detail ("Reminders go out 3, 7 and 14 days after the due date, and you can change the schedule").
The "how it works" block exists to show the effort is small. Three numbered steps from sign-up to first result fits most products. If setup genuinely takes longer, say so and say what help is available. Hiding effort produces sign-ups that never become customers.
Objections
List the reasons a qualified visitor would not act, then answer them on the page. Collect the reasons from real sources: sales call notes, support tickets, chat transcripts, refund requests, reviews of competing products. The most common categories are price, effort to switch, data security, contract length and whether the product fits the visitor's specific case.
Pricing deserves direct treatment. If pricing is published, show it or link to it plainly. If it is custom, say why and say what the first conversation involves. An FAQ of four to six real questions works well here, followed by any guarantee or trial terms in plain words.
Forms: ask only for what the next step needs
Every field is a reason to stop. Decide what the immediate next step truly requires (an email address for a download, name and email for a trial) and collect the rest later.
Build the fields correctly so browsers and assistive technology can help:
| Practice | Why | Source |
|---|---|---|
| A visible label for every field | Labels or instructions must be provided for user input (3.3.2, Level A), and a label is something presented to all users | W3C Understanding 3.3.2 |
| Valid autocomplete tokens (email, given-name, family-name, tel, organization) | Lets browsers fill fields; valid tokens satisfy 1.3.5 Identify Input Purpose (Level AA) | MDN autocomplete reference |
| Errors described in text, next to the field | An automatically detected error must be identified and described in text (3.3.1, Level A) | W3C Understanding 3.3.1 |
| type="email" or type="tel" for those fields | MDN advises these types for inputs that require that data; inputmode alone is only a keyboard hint and enforces nothing | MDN inputmode reference |
| A single column | Keeps the tab order and the reading order the same | Usability practice |
| An outcome-worded submit button | "Send me the checklist" sets expectations | Usability practice |
Placeholder text that disappears on typing is a weak substitute for a label. Users who come back to a half-filled form cannot see what each field was for. Keep instructions in a visible label or helper text.
Avoid autocomplete="off" unless there is a specific reason, such as a one-time code field. MDN notes it should be used only where genuinely necessary, and on a sign-up form it mostly slows people down.
Speed and stability: the page has to arrive
A landing page that loads slowly loses the people the ad paid to bring. Google's web.dev documentation defines three Core Web Vitals and their "good" thresholds, measured at the 75th percentile of page loads, separately for mobile and desktop:
| Metric | Measures | Good threshold |
|---|---|---|
| Largest Contentful Paint (LCP) | Loading: when the main content appears | 2.5 seconds or less |
| Interaction to Next Paint (INP) | Responsiveness to taps, clicks and keys | 200 milliseconds or less |
| Cumulative Layout Shift (CLS) | Visual stability | 0.1 or less |
INP replaced First Input Delay as a stable Core Web Vital in 2024. Google Search Central says Core Web Vitals are used by its ranking systems, and also that good scores do not guarantee top rankings: relevance comes first. So treat speed as a conversion and quality issue first and a search issue second.
For a landing page, most of the work is in the hero:
- Do not lazy-load the hero image. web.dev's LCP guide is blunt that the LCP image should never be lazy-loaded, because that always adds load delay.
- Make the hero image discoverable in the HTML. Images injected by JavaScript or set as CSS backgrounds cannot be found by the browser's preload scanner. Use an img element in the initial HTML, or preload the image if it must come from CSS.
- Mark it as high priority. fetchpriority="high" on the expected LCP image helps, but the same guide warns that applying it to more than one or two images makes it unhelpful.
- Serve a sensible format and size. The image format decision guide compares AVIF, WebP, JPEG and PNG, and the responsive images guide to srcset and sizes explains how to stop phones downloading desktop-sized files. The hero image file size and LCP budget guide works out how many KiB the hero can weigh at each width.
- Reserve space for everything. web.dev lists images without dimensions, late-loading embeds and ads, and web fonts among the common causes of layout shift. Give every image width and height attributes (or a CSS aspect-ratio), and reserve the height of any embedded video or review widget before it loads. Scored examples for an unsized image, a cookie banner and a late ad slot are in our CLS worked examples.
- Handle fonts deliberately. Font swaps can shift layout. web.dev suggests matching fallback font metrics, using font-display: optional, or preloading critical fonts; the troubleshooting guide for web fonts that fail to load covers the common causes when fonts misbehave.
- Keep animation cheap. Scroll-triggered effects and heavy hero animations can hurt INP. The web animation performance guide explains which properties animate without layout work.
web.dev also breaks LCP into four parts: time to first byte, resource load delay, resource load duration and element render delay. Its guideline is roughly 40% each for server response and download, with the two delay parts kept under 10% each. If your hero image takes most of its LCP time waiting to start downloading, the fix is discoverability and priority, not compression.
Pop-ups and overlays
Full-screen overlays on arrival push the offer out of view at the moment the visitor is deciding whether to stay. Google's Search Central guidance on interstitials advises against obscuring the entire page and suggests banners that take up a small fraction of the screen instead, while treating legally required gates such as age verification as exceptions. For a landing page with one goal, a pop-up promoting a second goal competes with the first. If you want an exit-intent offer, test it as a separate variant rather than adding it to every visit.
Copy frameworks as a first draft
Two classic structures help when the copy will not start.
Problem, Agitate, Solve (PAS). Name the visitor's problem, make its cost concrete, then present the product as the fix. For a time-tracking product: the problem is unbilled hours, the cost is invoices that undercount work, the fix is a timer that feeds the invoice directly. On the page, the headline can carry the solved state, the benefits section carries the cost, and the hero and how-it-works blocks show the fix.
Attention, Interest, Desire, Action (AIDA). The hero draws attention, the benefits build interest, proof and objection handling build desire, and the button asks for the action. The section order in the table above follows the same arc, which is one reason that order is so common.
Both frameworks break in the same way: they produce invented pain when the writer does not know the customer. The agitation step needs real language from interviews, support tickets and reviews. If a sentence in the "problem" block would not appear in anything a customer has actually said, cut it.
Testing: one change, enough traffic, a written log
Improve a page by testing changes against the current version, not by redesigning on opinion. A routine that works for small and large teams:
- Fix clarity first. Run the stranger check above. Testing button colors on a page whose headline confuses people measures noise.
- Write the hypothesis. "Visitors from the freelancer ad do not see that reminders are automatic; adding that to the subheadline will increase trial starts."
- Change one element. A sensible order to work through, starting with the elements every visitor sees, is headline and offer, then hero and proof, then form length, then call-to-action wording, then visual details.
- Measure the outcome that matters. A change that raises button clicks but lowers completed sign-ups has moved the drop-off further along the funnel. Track visit, click, completed form and the first meaningful use of the product.
- Run full weeks. Weekday and weekend traffic often behave differently, so stopping a test mid-week mixes the two.
- Log every result, including losses, with the dates, traffic and screenshots of both versions.
How much traffic does a test need? A worked illustration
Sample size depends on your current conversion rate and the smallest change you care about detecting. Using the standard two-proportion formula with a two-sided significance level of 0.05 (z = 1.96) and 80% power (z = 0.84):
n per variant = (1.96 + 0.84)² x [p1(1 - p1) + p2(1 - p2)] / (p2 - p1)²
| Baseline rate (p1) | Rate you want to detect (p2) | Relative lift | Visitors per variant | Total | Days at 1,000 visitors/day |
|---|---|---|---|---|---|
| 3.0% | 3.6% | +20% | about 13,900 | about 27,800 | about 28 |
| 3.0% | 4.5% | +50% | about 2,500 | about 5,000 | about 6 |
| 3.0% | 3.15% | +5% | about 208,000 | about 416,000 | over a year |
These inputs are illustrations, not benchmarks; plug in your own baseline. The pattern is what matters: small effects need enormous traffic. A page with a few hundred visitors a day cannot detect a 5% lift in any reasonable time, so it should test bold changes (a new headline and hero, a different offer, a shorter form) and rely more on qualitative research: watching session recordings, reading support questions, and talking to a handful of recent customers and people who left.
Stopping a test early because one variant "looks like it is winning" undoes the arithmetic. Decide the sample size before starting and wait for it.
A decision table for test results
| Result after the planned sample | What to do |
|---|---|
| Clear win on the main outcome, no loss further down the funnel | Keep the variant, log it, test the next element |
| Clear win on clicks, loss on completed sign-ups | Revert; the change attracted the wrong clicks |
| No meaningful difference | Keep whichever is simpler to maintain; try a bolder change |
| Clear loss | Revert, and record why you think it lost |
Page length
Length should match how hard the decision is. A free download or a waitlist can convert from a page that is mostly hero and form. A high-priced service, a product that replaces an existing tool, or anything that needs approval from someone else needs more proof, more objection handling and usually a longer page. The test is whether each section answers a question a real visitor has. If not, cut it, regardless of how long the page is.
Mistakes that cost the most
| Mistake | What it does | Fix |
|---|---|---|
| Two or more competing calls to action | Splits attention; visitors pick neither | One action, repeated |
| Headline names the product, not the result | Newcomers cannot tell what it is | Lead with the outcome |
| Generic stock imagery in the hero | Says nothing about the product | Real screenshots or product photos |
| Unverifiable or invented testimonials | Erodes trust and can break advertising rules | Named, permissioned, honest quotes |
| Lazy-loaded or CSS-background hero image | Delays LCP | img in the HTML, not lazy-loaded, high fetch priority |
| Images and embeds without reserved space | Layout shifts push the button around | Width, height or aspect-ratio on everything |
| Placeholder-only form labels | Users lose track of fields | Visible labels and helper text |
| Full site navigation in the header | Gives visitors many exits | A logo and, at most, one or two essential links |
| Testing tiny changes on low traffic | Tests never reach a result | Bigger changes, plus qualitative research |
FAQ
Should a landing page have a navigation menu?
Usually a minimal one at most. A logo that links home and possibly a pricing or contact link is enough on most pages. The footer can carry privacy, terms and contact. Pages that serve existing customers, such as feature announcements, are the exception, because those visitors may need the rest of the site.
How many calls to action can one page have?
One action, shown as many times as the page length needs. A short page might show the button once in the hero and once after proof; a long page might show it four or five times. What matters is that every button leads to the same next step with the same wording.
Are video testimonials better than written ones?
They do different jobs. Written quotes with a name, role and photo can be read in a few seconds and stay scannable. Video carries tone and conviction but needs the visitor to press play, adds weight to the page, and needs captions. Many pages use written quotes in the main flow and link to a video for visitors who want more. Reserve the video player's height in the layout so it does not shift content when it loads.
Is "above the fold" still a useful idea?
Yes, as long as you treat it as a range rather than a line. The first screen differs by device, so check the hero at a small phone width, a common laptop size and a large desktop size. The promise and the call to action should be visible on all three without scrolling. Visitors do scroll when the first screen gives them a reason.
What should a landing page link to?
As little as possible during the conversion flow. Links that support the decision (pricing details, a security page, a case study) are useful if they open content the visitor needs. Links that lead to unrelated parts of the site, a blog index, or social profiles tend to pull people away before they act.
Primary sources for this guide
- LCP, INP and CLS thresholds and the 75th percentile rule (web.dev Web Vitals overview), https://web.dev/articles/vitals. Checked October 22, 2026.
- Hero image loading, fetch priority and the four LCP sub-parts (web.dev LCP guide), https://web.dev/articles/optimize-lcp. Checked October 22, 2026.
- Causes of layout shift and reserving space for media and fonts (web.dev CLS guide), https://web.dev/articles/optimize-cls. Checked October 22, 2026.
- How Google Search uses Core Web Vitals and what they do not guarantee (Search Central page experience documentation), https://developers.google.com/search/docs/appearance/page-experience. Checked October 22, 2026.
- Full-page overlays versus small banners (Search Central interstitials guidance), https://developers.google.com/search/docs/appearance/avoid-intrusive-interstitials. Checked October 22, 2026.
- The 24 by 24 CSS pixel pointer target floor (W3C explainer for 2.5.8), https://www.w3.org/WAI/WCAG22/Understanding/target-size-minimum.html. Checked October 22, 2026.
- Button label contrast and the large-text pixel sizes (W3C explainer for 1.4.3), https://www.w3.org/WAI/WCAG22/Understanding/contrast-minimum.html. Checked October 22, 2026.
- Visible labels for form input (W3C explainer for 3.3.2), https://www.w3.org/WAI/WCAG22/Understanding/labels-or-instructions.html. Checked October 22, 2026.
- Describing form errors in text (W3C explainer for 3.3.1), https://www.w3.org/WAI/WCAG22/Understanding/error-identification.html. Checked October 22, 2026.
- Autocomplete tokens and their link to 1.3.5 (MDN autocomplete reference), https://developer.mozilla.org/en-US/docs/Web/HTML/Reference/Attributes/autocomplete. Checked October 22, 2026.
- Email and telephone input types versus the inputmode hint (MDN inputmode reference), https://developer.mozilla.org/en-US/docs/Web/HTML/Reference/Global_attributes/inputmode. Checked October 22, 2026.
- Honest endorsements, material connections and typical results (FTC staff Q&A on the Endorsement Guides), https://www.ftc.gov/business-guidance/resources/ftcs-endorsement-guides-what-people-are-asking. Checked October 22, 2026.
The sample-size figures were calculated by the Lucky Graphics editorial team from the formula and inputs shown; they are illustrations, not industry data.
Before you build the next page
The hero headline and the type scale around it decide how much fits in the first screen at phone width. The typography scale system guide shows how to size the headline, subheadline and button text as one set so the promise and the call to action both stay visible without scrolling.