Landing Page Conversion Design - System Reference¶
A working reference for generating high-converting landing pages. Every rule below is concrete enough to be applied without further interpretation. Sources are cited inline and listed at the bottom.
Core principles (the non-negotiables)¶
- One page, one goal. Conversion-Centred Design principle #1: create focus. Remove every link, nav item, or CTA that does not serve the single conversion goal. (Unbounce / Oli Gardner)
- Clarity beats persuasion. A visitor must be able to answer "what is this, who is it for, what do I get" within 5 seconds of landing. If they can't, the page has failed the 5-second test before any other optimisation matters. (CXL / Peep Laja)
- Match the source. Headline, hero image and offer must mirror the ad, email or link that brought the visitor. Message-match drops bounce rate and lifts conversion more reliably than any colour or copy tweak. (Unbounce CCD principle #3)
- Reduce friction at every step. Every field, click and decision is a tax. Default position: cut it unless it earns its place.
- Speed is a conversion lever, not a tech concern. LCP under 2.5s is the floor. Each 0.1s of load time improvement is worth roughly 8% in conversion in retail benchmarks. (web.dev, Google)
- Stable usability principles age slowly. NN/g's 25-year position: human visual scanning, reading patterns and trust signals barely move year to year. Treat shiny trends with suspicion. (NN/g)
- Test, don't decree. Every "best practice" below is a strong default, not a universal truth. Ship the default, then test the top three highest-traffic elements (headline, hero, CTA copy) first. (CXL "What to Test First")
Hero / above-the-fold¶
Must contain, in order of priority:
- Headline (the value proposition in plain language).
- Subhead (one sentence expanding the headline with specifics).
- Primary CTA (button or single-field form).
- Hero visual (product screenshot, demo, or relevant photo - not stock).
- One trust signal (logo bar, review count, or single high-credibility quote).
Rules: - Fold target: assume 1366x768 desktop and 390x844 mobile. The CTA must be visible without scrolling on both. - Hero copy must pass the 5-second test: a stranger reading for 5 seconds must be able to state what you sell, who it's for, and the next action. (CXL) - No carousel. Erik Runyon's Notre Dame data showed sliders get a ~1% click-through rate and 84-90% of those clicks land on the first slide - everything after slide 1 is invisible. Use a single hero. (Runyon, 2013) - LCP element is almost always the hero image or headline. Pre-load it, serve next-gen format (AVIF/WebP), inline critical CSS. Target LCP < 2.5s on mobile 4G. (web.dev)
Headlines that convert¶
Default formulas (pick one, fill in specifics):
- Outcome + timeframe: "Ship your landing page in 20 minutes."
- [Target] who [pain] get [outcome]: "Founders who hate writing copy get high-converting pages without hiring a copywriter."
- Problem -> promise: "Slow WordPress site? Get to A-grade PageSpeed in one week."
- Specific number + benefit: "Cut your checkout abandonment by 35%."
Rules: - Length: 6-12 words for the headline. Subhead 10-25 words. - Lead with the customer's outcome, not your product's feature. Features go in the body. (CXL value-proposition canon) - Avoid jargon, brand-speak, and self-referential phrasing ("We are a leading provider of..."). - The headline must not require the visual to make sense. If the image fails to load, the offer must still be clear. - Numbers in headlines outperform vague claims in CXL and WiderFunnel test logs. "39% lift" beats "huge lift".
CTAs¶
Copy - Use first-person, action-led: "Get my free audit" beats "Get your free audit" by ~90% in Michael Aagaard's Content Verve test. "Start my trial" beats "Start your trial". - Specify the reward, not the action: "Show me the pricing" beats "Submit". "Download the checklist" beats "Click here". - 2-5 words. If you need more, it's two CTAs.
Colour and contrast - There is no universally winning colour. The HubSpot "red beats green 21%" test only held because red had higher contrast against that specific page. (HubSpot, 2011; CXL replication) - Rule: the primary CTA must be the highest-contrast element on the page. Achieve this with a colour not used elsewhere in the hero. (EyeQuant) - Secondary CTAs (if any) must be visually subordinate: ghost button, text link, or muted fill.
Size and placement - Minimum tap target: 44x44 CSS points (Apple HIG) / 48x48 dp (Material) / 24x24 px (WCAG 2.5.8 AA, 44x44 AAA). - Above the fold: one primary CTA, always. - Long pages: repeat the CTA every 1.5-2 screens, and after every major proof block (testimonial, case study, pricing). - Sticky CTA on mobile pages over 2 screens.
Forms¶
Field count - Each removed field above 4 typically adds conversion. Imaginary Landscape's classic test: 11 fields -> 4 fields lifted conversion 120%. (Imaginary Landscape / Imagescape) - For top-of-funnel lead capture, default to 3-4 fields (name, email, company, one qualifier). - For e-commerce checkout, Baymard's benchmark is 12.8 fields average; optimal is 6-8. Completion drops 4-6% per field above 8. (Baymard) - Strip optional fields. Expedia recovered $12M/yr by removing one optional "Company" field that was confusing users into entering bank addresses. (Duncan Jones / UX Movement)
Layout - Single column. Multi-column forms slow completion and break scan rhythm. (Baymard, NN/g) - Labels above the field. Inside-the-field placeholder-only labels fail accessibility and disappear as soon as the user types. - Group logically and split long forms into steps with a progress indicator - Baymard's finding is that perceived field count beats actual; a 15-field form in 3 logical steps outperforms a 10-field single-page form. - Show inline validation as the user moves off a field, not on submit. Green check for valid, red message with specific fix for invalid. - Required-field indicator should be the exception (mark optional fields), since most are required.
Microcopy - Submit button reflects the reward, not the action (see CTA section). - Privacy reassurance under email field: "We will not share your email" or "No spam, unsubscribe in one click."
Social proof¶
Order by credibility (use whichever you have, in this order):
- Quantified results from named customers. "Stormline lifted conversion 39.6% in 6 weeks - Tom, CEO" with photo and logo. Specific number + named person + face beats every other format.
- Recognisable logo bar. 5-8 logos, monochrome, single row. Caption "Trusted by" or "Used by teams at". Logos should be readable; if your customers aren't recognisable, drop the bar.
- Aggregate review count. "4.8 stars from 1,247 reviews" with the platform name (G2, Trustpilot, Capterra). Numbers must be real.
- Testimonials with face + name + role + company. Avoid first-name-only or initials - reads as fake.
- Case-study cards linking to the full story with headline metric on the card.
Rules: - Place one trust signal in the hero, the main testimonial block 1-2 screens down, and a final proof block adjacent to the CTA. - Specificity > volume. One detailed quote with a number beats ten vague "Great service!" lines. - Never invent or paraphrase a quote. Use the customer's words verbatim.
Trust signals¶
Evidence-based (use these): - Money-back guarantee with terms stated plainly. - Real founder photo and name, ideally with a short signed message for high-trust offers (consulting, finance, health). - Visible security context for payment forms: padlock icon, "Powered by Stripe", card logos. Trust seals (Norton, McAfee) have weak modern evidence; payment-processor branding is stronger. (Baymard) - Specific company details in the footer: legal name, registered address, VAT/company number. - Press mentions only if from publications the target audience recognises.
Cargo-cult (skip or downgrade): - Generic "100% secure" badges with no link or verification. - Stock photography of "diverse team smiling at laptop" - reduces trust, signals template. Use real product screenshots, real customer photos, or commissioned illustration. - "As seen in" rows of logos with no link to the actual mention.
Visual hierarchy¶
What the research actually says:
- F-pattern is a failure state, not a target. NN/g's own follow-up to the 2006 study: the F-pattern emerges when content lacks structure (weak headings, dense paragraphs). Strong H2/H3s, short paragraphs and bullets break the F and produce better comprehension. (NN/g "F-Shaped Pattern... Misunderstood")
- Z-pattern is a layout heuristic for sparse pages - landing pages with little text and a single CTA. Works as a default for hero sections: logo top-left, secondary nav/CTA top-right, headline/sub mid-page, primary CTA bottom-right of the hero block.
- Visual weight directs attention more than position. Size, colour contrast, whitespace, and human gaze (a photo of a person looking at the CTA pulls eyes to the CTA - EyeQuant heatmaps).
- Limit competing focal points to one per screen. Two equally weighted CTAs split attention and depress both.
Copy hierarchy¶
Order on the page, top to bottom:
- Headline (6-12 words, outcome-led).
- Subhead (10-25 words, adds specifics).
- Hero CTA (2-5 words, first-person reward).
- Proof strip (logos or review count, 1 line).
- Problem section (3-5 sentences naming the pain in customer words).
- Solution / how-it-works (3 steps, each step: icon + 3-word title + 1 sentence).
- Benefits over features for short pages; feature detail blocks for considered B2B purchases.
- Testimonial / case-study block.
- Objection handling / FAQ (5-8 questions in the customer's voice).
- Final CTA block (restate headline + button, no new info).
Word count: short B2C/lead-gen pages 300-600 words; considered-purchase B2B SaaS 1,200-2,500 words. Longer pages do not hurt conversion if every section earns its place - MarketingSherpa and Unbounce benchmark data both show this.
Benefits vs features: lead with benefits in the hero and problem sections; feature detail is only needed for high-consideration purchases where buyers comparison-shop on capability (B2B SaaS, dev tools, hardware).
Mobile-specific¶
- Mobile-first, not responsive-as-an-afterthought. Design the 390px width first, then expand.
- Tap targets: 44x44pt minimum (Apple HIG); spacing 8px minimum between adjacent targets to prevent mis-taps.
- Thumb zone: place primary CTA in the bottom 1/3 of the screen or as a sticky bottom bar. Avoid top-edge primary actions on mobile.
- Single-column everything. No side-by-side columns under 768px.
- Forms: numeric inputmode for phone/postcode/card;
autocompleteattributes on every field;type="email"to trigger the email keyboard. - Hide non-essential nav behind a hamburger only if the page is truly nav-light; otherwise let it scroll off.
- No hover-only interactions. Every action must work with a tap.
- Mobile LCP target 2.5s on a simulated Moto G4 / Slow 4G - this is the threshold Google scores against.
Performance thresholds¶
Core Web Vitals (Google, web.dev) - these are the operational floors:
- LCP (Largest Contentful Paint): < 2.5s good, 2.5-4s needs improvement, > 4s poor.
- INP (Interaction to Next Paint): < 200ms good, 200-500ms needs improvement.
- CLS (Cumulative Layout Shift): < 0.1 good, 0.1-0.25 needs improvement.
Conversion impact (real case data): - Pinterest: improving LCP from 2.5s -> 1.5s lifted sign-up conversion 15% and SEO traffic 15%. - Vodafone: same 1s LCP improvement, 8% sales lift. - Renault: 13% conversion lift on LCP improvement. - Mobile bounce: 53% of mobile users abandon at 3s+ load (Google). - General benchmark: each 0.1s of load improvement is worth ~8% in conversion.
Practical defaults: inline critical CSS, lazy-load below-fold images, preload the LCP image, defer non-critical JavaScript, serve images in AVIF/WebP with explicit width/height to prevent CLS.
Anti-patterns - never do these¶
- Carousels in the hero. ~1% CTR, 84% of clicks on slide 1 - everything else is invisible. (Runyon, NN/g)
- Modal pop-ups within 5 seconds of landing. Among the most-hated web elements per NN/g; tanks first-impression trust. Exit-intent or scroll-triggered after 50%+ scroll only.
- Autoplaying video with sound. Universally penalised by users and browsers.
- Stock photos of "diverse smiling team at laptop". Signals template, reduces trust.
- Multiple primary CTAs competing in the hero. Picks one or none.
- Jargon-heavy headlines. "Synergistic AI-powered solutions" tells the visitor nothing.
- Nav bar full of links on a paid-traffic landing page. Strips focus. Logo + one CTA in the nav is the maximum.
- Placeholder-only form labels. Fail accessibility and disappear on type.
- Fake scarcity / fake countdown timers. Detected, mocked, trust-destroying.
- "Click here" / "Submit" / "Learn more" as CTA copy. Replace with reward-led copy.
- Optional form fields that aren't earning their place. The Expedia $12M field.
- Animations that delay text rendering. NN/g found scroll-triggered text animations delay users; treat as friction.
- Long unbroken paragraphs. Forces the F-pattern; reduces comprehension.
Industry adjustments¶
B2B SaaS - Long-form (1,500-2,500 words) acceptable for considered purchases. - Pricing transparency wins; "Contact sales" is a step backwards for self-serve tiers. - Feature-comparison tables, integration logos, security/compliance badges (SOC 2, GDPR) are first-class trust signals. - Hero should include a product screenshot or short looped demo (no sound, < 10s). - Two CTAs allowed: "Start free trial" (primary) and "Book a demo" (secondary, visually muted).
E-commerce / DTC - Shorter copy, stronger imagery. Product photography on plain background + lifestyle context. - Trust signals: review count + average star rating, return policy, shipping speed, payment-processor logos. - Above-fold price, add-to-cart, and 1 trust line. - Baymard's checkout rules are the canonical reference: guest checkout, 6-8 fields, address autocomplete, pre-fill where possible.
Lead generation (single-form pages) - Single goal: form submit. Strip nav entirely. - 3-4 form fields maximum. - One testimonial, one logo bar, one founder face above the fold or beside the form. - Hero CTA = the form itself.
Course creators / info products - Long-form sales pages (2,000-5,000 words) are the format. - Sections: hook, problem agitation, story, solution, what's included, who it's for, who it's not for, instructor credibility, testimonials with specific outcomes, pricing, guarantee, FAQ, final CTA. - Money-back guarantee is near-mandatory. - Video sales letter optional but tested high in this category.
Local services - Phone number top-right, click-to-call on mobile, sticky on scroll. - Service area named explicitly ("Plumbers in Valencia"). - Reviews from Google with the Google logo carry more weight than any other testimonial format here. - Photos of the actual team and vans/premises, not stock.
Sources¶
- Unbounce - The 7 Principles of Conversion-Centred Design (Oli Gardner)
- Nielsen Norman Group - F-Shaped Pattern of Reading on the Web: Misunderstood, But Still Relevant
- Nielsen Norman Group - Homepage Design: 5 Fundamental Principles
- Nielsen Norman Group - Top 10 Guidelines for Homepage Usability
- CXL - Anatomy of a High-Converting Landing Page (Peep Laja)
- CXL - First Impressions: the 5-Second Test
- CXL - Unique Value Proposition: How to Create a UVP
- CXL - Which CTA Button Colour Converts Best
- web.dev - Largest Contentful Paint (LCP)
- Baymard Institute - Checkout Usability Research
- Baymard - Checkout Optimisation: From 16 Form Fields to 8
- Imaginary Landscape - Fewer Fields Case Study (11 -> 4, +120%)
- Duncan Jones - Expedia $12M Form Field Case Study
- UX Movement - The $12 Million Optional Form Field
- Erik Runyon - Carousel Interaction Stats (Notre Dame data)
- HubSpot - The Button Colour A/B Test: Red Beats Green
- WordStream - 17 Best Practices for Call-to-Action Buttons (Aagaard first-person test)
- Apple Human Interface Guidelines - Buttons
- TestParty - WCAG 2.5.8 Target Size Guide
- EyeQuant - What is the Best Colour for CTA Buttons
Supporting canon (books that hold up): - Steve Krug, Don't Make Me Think - scannability, self-evident design. - Robert Cialdini, Influence - the six principles (reciprocity, commitment, social proof, authority, liking, scarcity) underpin almost every trust-signal pattern above. - Susan Weinschenk, 100 Things Every Designer Needs to Know About People - the cognitive-load basis for short forms, single CTAs and chunked content.