Skip to main content
Canvas resource

The Evidence-Based Landing Page Guide

A practical, evidence-informed guide to planning, writing, designing and testing landing pages—with the source, scope and uncertainty behind each recommendation kept visible.

Editorial note: This is a supplied guide, not original research. Its studies mix peer-reviewed, industry and observational evidence, and the recommendations are not universal causal guarantees. Test changes in your own context and evaluate them honestly.

Download the free PDF

Ungated download · No form or email required

Share the guide

Share it because it is useful to your audience. Attribution and reciprocal links are optional, not required.

LinkedInXEmail

How to use this resource

Use the guide to make a hypothesis, choose one meaningful change, and measure the outcome against a suitable comparison. Evidence quality and context matter: a result from one audience, channel or study does not guarantee the same result for another.

About this guide

Read the evidence in context

This is an evidence-led working guide to landing-page decisions, adapted from the supplied CMS Canvas draft. It is not original research, a conversion guarantee or a universal formula. The draft combines peer-reviewed studies and academic working papers with large observational analyses and industry-sponsored research: the Qubit meta-analysis of 6,700 online experiments was produced by a vendor and says its methodology was assured by PwC; the Deloitte/Fifty-five/Google speed study was commissioned industry research. Those sources can be useful, but they are not the same kind of evidence as an independently replicated randomized experiment. Four principles organize the page. First, the top of the page earns disproportionate attention, so the first two screenfuls must make the offer, mechanism and next action clear. Second, design is a threshold rather than a reliable conversion lever: a page must look credible and usable, but a zero average effect does not mean zero effect for every audience, context or treatment. Third, behavioural content and a credible explanation generally deserve more attention than cosmetic changes. Fourth, choose one coherent strategy—quality-led or promotion-led—and do not make a premium promise while surrounding it with contradictory discount cues. The Qubit/PwC-assured sample reported scarcity (+2.9%), social proof (+2.3%), urgency (+1.5%) and abandonment recovery (+1.1%) as positive categories while cosmetic changes averaged around zero; these vendor, sample-specific averages are not a universal ranking or guarantee, so use them as hypotheses rather than permission to manufacture scarcity or assume attention becomes conversion. Read the evidence with care. Attention is not conversion; a correlation is not a causal result; an average effect can conceal heterogeneous winners and losers; and an observation in one market, device, funnel or sample is not a universal rule. There is no research-backed universal word count, sentence count, form-field count, button colour, CTA wording, video lift, logo-bar lift or FAQ lift. Numbers in this guide stay scoped to the source that reported them. Where a supplied claim could not be independently checked here, it is labelled as an attributed draft estimate or a lead for verification rather than presented as settled fact. The assembled-page blueprint is: hero; causal explanation; behavioural social proof and a few honest testimonials; identity and customer treatment; depth content; offer details and risk reversal; then the form and final CTA. Universal rules run through every section: make the page fast, write plainly, match customer language, choose one strategy, segment when the evidence and operations support it, and measure qualified outcomes rather than attention alone. The final section also records what this guide deliberately does not tell you and how to test responsibly.

Section

The hero (first screen)

The first screen is the page's orientation point. Put the offer, audience, one next action and a relevant representation of the thing there. Attention findings justify priority, not a claim that the first screen alone causes conversion.

Say what this is, plainly

Use one sentence in the visitor's vocabulary to name the offer and who it serves. A description is more useful than a slogan, adjective stack or unsupported superlative.

Suggested steps

  1. Write a plain proposition that answers what it is, who it is for and what decision the page supports.
  2. Replace slogans, hype and vague benefit claims with concrete nouns, verbs, constraints and a meaningful next step.
  3. Read the sentence aloud with a customer or colleague who was not in the room; revise any phrase they would not naturally use.
  4. Keep the claim accurate for the audience and traffic source. A clear sentence is not permission to imply an outcome the evidence cannot support.
Caution: The cited Nielsen Norman Group eye-tracking study is about attention, not conversion. Its supplied 57% first-screen and 74% first-two-screen figures should remain attributed study observations, not universal page laws.

Give the first screen one unambiguous action

Put one primary CTA in the first screen for visitors who are ready. CTA wording and colour do not have a research-backed universal winner; presence, relevance and an understandable action matter more than decoration.

Suggested steps

  1. Choose the one action that matches the page's job: enquire, buy, book, start, download or another clearly defined step.
  2. Make the CTA visually discoverable without surrounding it with competing primary buttons or contradictory destinations.
  3. Repeat the same action later only when it follows the reader's decision path; do not turn every block into a new conversion demand.
  4. Instrument the action and its qualified downstream outcome separately so clicks are not mistaken for completed conversions.
Caution: The supplied CTA-copy meta-analysis estimate of −0.3% average effect is a scoped draft attribution to the 6,700-experiment Qubit analysis. Treat the average as context, not as evidence that wording never matters.

Show the offer and clear the design threshold

If a price-based offer is central to a promotion-led strategy, state it early. Pair it with one authentic image or short video of the product, result or people who deliver the service, then use legible type, whitespace and a restrained palette.

Suggested steps

  1. State the qualifying price, free tier, discount or introductory condition near the proposition when it genuinely defines the chosen strategy.
  2. Use a real, relevant image or video; explain what it shows and provide an accessible alternative for people who cannot view it.
  3. Check readability, contrast, hierarchy, responsive behaviour and load weight before polishing decorative details.
  4. Stop at a professional, trustworthy threshold and redirect effort to clearer evidence, useful content and customer experience.
Caution: McDowell et al. and Hylewski et al. are observational/associational sources, while the Airbnb image analysis is a different demand context. A reported zero average for cosmetic A/B changes is not zero effect for every subgroup, and none of these studies supplies a universal design formula.

Section

How it works (the causal explanation)

Immediately below the hero, explain the mechanism: why the product or service should produce the stated result. This is the section the supplied draft says many pages skip when they substitute statistics or testimonials.

Explain the mechanism, not only the outcome

A causal explanation has a because-X, therefore-Y shape. It helps a reader judge whether the offer can plausibly work for their situation rather than asking them to borrow confidence from a number.

Suggested steps

  1. Name the input, process and observable result: what happens first, what the service or product changes and what the customer can expect to notice.
  2. Show the steps in the real order, including prerequisites, handoffs, timing and limits.
  3. Support material claims with primary evidence, methodology, examples or a transparent explanation of what is known and unknown.
  4. Ask a new reader to restate the mechanism; ambiguity signals that the page needs a diagram, example or plainer language.
Caution: Haans, Raassens and van Hout found causal evidence outperformed other evidence types for conversion in their search-advertising setting. That peer-reviewed result supports testing explanations; it does not establish a universal causal effect for every landing page.

Write readable, varied body copy

The supplied draft reports an approximately 10–11-word sentence optimum, about 22 sentences for a whole page, a Fog index around 8 and a positive association between unique-word percentage and conversion (r = 0.34). Treat these as model-specific observational guidance, never as quotas.

Suggested steps

  1. Prefer familiar words, active verbs and one idea per sentence; remove jargon unless the audience uses it precisely.
  2. Vary sentence length and paragraph shape instead of forcing every sentence to a target count.
  3. Use headings, bullets, examples and diagrams to reduce cognitive load while retaining the detail a careful buyer needs.
  4. Read for comprehension with representative users and compare qualified outcomes; readability scores are screening tools, not a substitute for user research.
Caution: Korniichuk & Boryczka and the cited Marketing Letters work report relationships/predictions in particular samples. Correlation is not causation, and there is no universal word count, sentence count or reading-level target.

Section

Social proof

Show evidence of what real people do, then add a small number of specific testimonials in the customer's own language. Recognition can help orientation, but logos and quotes cannot replace a credible offer and mechanism.

Lead with behavioural proof

Recent purchases, current sign-ups, active-customer counts or genuinely popular choices show behaviour rather than merely asserting approval. Keep each signal dated, defined and representative.

Suggested steps

  1. Choose a behavioural measure that is real, current and relevant to the decision: what others actually selected, used or completed.
  2. Define the time window, population and denominator wherever a number could be misunderstood.
  3. Update or retire the signal when it becomes stale; do not create fake activity, hidden scarcity or a misleading popularity ranking.
  4. Measure whether the proof helps qualified visitors make a better decision, not only whether it attracts clicks.
Caution: The supplied +2.3% social-proof estimate and 82% positive-test figure come from the vendor-produced Qubit/PwC-assured meta-analysis. Preserve the attribution and scope; an average across tests is not a promise for this page.

Use a few specific customer-voice testimonials

Two or three concrete testimonials can add texture when they sound like the customer, name the situation and avoid claims the page cannot verify. Authenticity of voice matters more than polished marketing language.

Suggested steps

  1. Request permission and preserve the customer's meaning, relevant context and any material qualification; do not ghostwrite a certainty they did not express.
  2. Select different but representative experiences, including limits or a less-than-perfect fit when that helps honest choice.
  3. Pair each quote with verifiable context such as role, use case, date or approved organisation—not invented detail.
  4. Review testimonials for privacy, disclosure, accessibility and continued accuracy; remove or update them when circumstances change.
Caution: Ludwig et al. studied review language and conversion associations. A matching dialect or positive tone is not a fixed testimonial lift, and one glowing quote is not independent proof of every claim.

Use names and affiliations with restraint

Recognisable client names, press mentions and legitimate affiliations can help a visitor identify the organisation. They are orientation cues, not substitutes for customer treatment or an evidence-backed claim.

Suggested steps

  1. Use a name, logo or affiliation only with permission and a clear relationship; link to the original source where that helps verification.
  2. Explain what the relationship means and when it was current. Do not imply endorsement, certification or partnership where none exists.
  3. Place logos near the relevant proof rather than using a wall of unrelated marks as borrowed authority.
  4. Keep privacy-policy and legal links available without pretending their presence alone establishes trust.
Caution: Fogg et al. reported name recognition/reputation in credibility judgments and much lower mention of seals. Credibility judgments are not conversion tests; no research-grade universal trust-badge lift was found in the supplied bibliography.

Section

Who you are and how you treat customers

Commercial pages need an accountable identity and a plain explanation of customer treatment. Tell people who operates the offer, why the business does it and what happens when something goes wrong.

Make the organisation accountable

A real company name, people, location and contact path let a visitor understand who stands behind the page. State the motive plainly: what the customer gets and what the business gets.

Suggested steps

  1. Name the legal or trading entity, responsible people, operating location and a usable contact method appropriate to the offer.
  2. Explain why the business exists and how the offer is funded without hiding commercial incentives behind a faux-independent voice.
  3. Connect claims to evidence: credentials, methodology, dates, customers or primary documents that a reader can inspect.
  4. Keep ownership, authorship and updates visible when the page is revised or assisted by outside contributors.
Caution: The supplied Fogg/Consumer WebWatch findings concern perceived credibility and suspicion on commercial sites. They do not prove that adding an identity block causes a fixed conversion increase.

Show how problems are handled

Describe returns, refunds, support, response times, cancellation, delivery and fees with the specificity a cautious buyer needs. Clear policies are customer treatment, not decorative trust copy.

Suggested steps

  1. List the material terms before commitment: eligibility, cost, timing, support channel, cancellation and what happens if the expected outcome is not delivered.
  2. Use concrete examples and plain boundaries instead of vague assurances such as risk-free or guaranteed.
  3. Make legal, privacy and accessibility information findable while keeping the main page focused on the decision.
  4. Give the team an owner and a review date for each promise so operations can deliver what the page says.
Caution: The supplied Fogg report says customer service and policies were mentioned more often for commercial sites; that is evidence about credibility judgments, not a universal conversion coefficient. Privacy links remain necessary for legal and ethical reasons even when rarely mentioned in judgments.

Choose quality-led or promotion-led

A page should make one coherent argument. A premium, quality-led promise and a discount-led urgency frame can contradict each other when mixed without a clear reason.

Suggested steps

  1. Choose whether the page primarily sells quality, fit, service and proof or a defined promotion with transparent terms.
  2. Align proposition, imagery, testimonials, pricing, CTA and risk reversal with that strategy.
  3. If audiences or intents differ materially, segment the experience rather than asking one page to make incompatible arguments.
  4. Review the assembled page as a first-time visitor: can they explain why this offer is worth choosing and what happens next?
Caution: Di Fatta, Patton & Viglia provide a context-specific SME e-commerce basis for coherent positioning. The result does not mean every page must avoid all price information or that mixed strategies can never work.

Section

Depth content (FAQ, specifications and details)

Below the second screenful, give careful readers the substance needed to verify fit. Most people may not read everything, but an organised depth layer can signal authority and prevent a headline-only page from feeling thin.

Answer the questions that block a decision

Use direct FAQs, specifications, dimensions, ingredients, methodology, comparisons and clear edge cases—including who the offer is not for. Make the detailed version easy to scan.

Suggested steps

  1. Collect questions from sales calls, support, search queries, returns and customer research; group them by decision stage.
  2. Answer each question directly before adding background. Include units, definitions, dates, assumptions and what would change the answer.
  3. Show meaningful alternatives and trade-offs when a comparison helps the reader; do not turn a self-authored comparison into a fake independent review.
  4. State who is not a fit and what the offer cannot do. Honest boundaries reduce avoidable mismatch and support a better decision.
Caution: Fogg's information-focus and information-structure findings support making depth available and organised. They do not establish an optimal FAQ length or prove that any FAQ block causes conversion.

Add depth without a wall of text

Use collapsible FAQs, tables, labelled specifications, diagrams and progressive disclosure when they improve scanning without hiding material terms or harming accessibility.

Suggested steps

  1. Keep headings, labels and table cells meaningful when read out of visual order or by assistive technology.
  2. Put critical limitations and decision terms in the initial page content; do not hide them behind an interaction.
  3. Link detailed methods, policies and primary evidence from the relevant claim so the reader can verify rather than hunt.
  4. Test the structure on mobile, keyboard navigation, slow connections and real comprehension—not only on a design mock-up.
Caution: Information depth can communicate authority while attention remains uneven. Do not infer that being present means being read, or that being read means being believed or converted.

Section

Offer details and risk reversal

Near the commitment point, resolve delivery, fulfilment, onboarding, guarantee and return questions. Use scarcity or urgency only when it reflects a real constraint and explain the boundary.

Make fulfilment and guarantees concrete

People need logistics when they decide. Say what happens after sign-up or purchase, how long it takes, what it costs, what is covered and what the customer must do.

Suggested steps

  1. State delivery, shipping, onboarding or access terms in the same vocabulary used by the checkout, sales team and support team.
  2. Explain a guarantee, return or cancellation policy with eligibility, time limits, exclusions and the actual process.
  3. Place the answer beside the commitment point, then link to full terms without replacing the plain-language summary.
  4. Review operations and analytics after launch: a promise that increases action but creates avoidable disputes is not a successful page.
Caution: Miller & Hosanagar's meta-analysis reports stage-specific associations in a sample of e-commerce tests. It is a reason to test logistics messaging late in the funnel, not proof that every visitor weighs it only at that stage.

Use scarcity only when it is true

Limited inventory, seats or time can help a decision when the limit is real, visible and not manufactured. Urgency can also backfire, especially when it looks like manipulation.

Suggested steps

  1. Define the real constraint, who it applies to, the time window and what happens when it ends.
  2. Show a current, auditable status rather than a resettable countdown or invented stock message.
  3. Use calm language that helps planning; do not turn uncertainty, a normal sales deadline or an always-on promotion into fake scarcity.
  4. Monitor complaints, refunds, trust signals and downstream quality, not only the immediate click or order.
Caution: The supplied +2.9% scarcity, +1.5% urgency and positive-test figures are attributed to the Qubit/PwC-assured analysis. The draft notes that fake-scarcity trust costs were not measured; assume honesty is the minimum, not a tested optimisation.

Section

The form and final call to action

End the natural reading path with one clear action that matches the hero. Keep the form no longer than necessary, but do not treat minimal field count as a proven law. Then apply the universal rules and audit the assembled page.

Match the action and respect the friction question

Use one final CTA, a form that asks only for information the next step needs and a placement that follows the reader's decision path. For left-to-right reading, bottom-right terminal placement is a useful hypothesis, not a law.

Suggested steps

  1. Use the same action label and promise as the hero unless the decision has legitimately changed.
  2. Ask for only information with a clear operational purpose; explain why a sensitive or unusual field is needed.
  3. Make errors, validation, privacy, accessibility, success state and follow-up expectations visible before submission.
  4. Test short and longer forms against qualified completion, sales quality, retention and support load—not registrations alone.
Caution: Hernandez & Resnick is a small eye-tracking study. Huang et al.'s randomized field experiment found an early registration request increased registrations by 58% in its context; it does not identify a universal optimal field count or prove that earlier friction improves every funnel.

Apply the universal rules without overclaiming

Speed, tone, customer language, coherent strategy and responsible segmentation are cross-page disciplines. They improve the conditions for a decision, but no one rule guarantees conversion.

Suggested steps

  1. Speed: treat performance as part of the page. The Deloitte/Fifty-five/Google study associated a 0.1-second mobile improvement with 8.4% higher retail conversions, 10.1% higher travel conversions and 8.3% more lead-generation form submissions; these are industry-study associations, not universal causal effects.
  2. Tone: be straightforward and friendly. Avoid slang, profanity, sensationalism and obvious spin; credible communication can understate rather than overclaim.
  3. Language: use customer vocabulary, varied words and a level representative readers can follow. Do not use readability targets as a substitute for listening.
  4. Segmentation: if operations support it, adapt by source, device, intent or audience. The supplied Qubit analysis reports targeted treatments at nearly three times expected uplift; preserve the source and test the local effect.
Caution: The same visitor can be affected by many interacting factors. Measure page speed, comprehension, action, qualified outcome and customer experience together instead of claiming one universal lever.

Audit the assembled page

Use the sequence below as a practical blueprint, then check whether each claim has evidence and each promise can be fulfilled. It is an organising model, not a validated seven-step causal funnel.

Suggested steps

  1. 1 — Hero, screen 1: plain statement, single CTA, offer and real image. Basis: NN/g attention; Fogg credibility; Miller & Hosanagar promotion placement; Zhang et al. image analysis.
  2. 2 — How it works, screen 2: causal explanation and readable copy. Basis: Haans et al. evidence types; Korniichuk & Boryczka copy analysis.
  3. 3 — Social proof, screens 2–3: behavioural proof first, customer-voice testimonials, then names. Basis: Browne & Jones; Ludwig et al.; Fogg name recognition.
  4. 4 — Who you are and customer treatment, screen 3. Basis: Fogg commercial-site credibility findings.
  5. 5 — Depth content, screen 4+: FAQ, specifications, comparisons and edge cases. Basis: Fogg information focus/structure; Korniichuk & Boryczka guidance.
  6. 6 — Offer details beside the final CTA: delivery, guarantee and genuine scarcity. Basis: Miller & Hosanagar funnel analysis; Browne & Jones scarcity categories.
  7. 7 — Form and final CTA, at the bottom: one clear action, with friction tested in context. Basis: Hernandez & Resnick; Huang et al.
Caution: The order prioritises orientation and decision support. It does not imply that everyone reads in this order, that a section guarantees an uplift or that the first two screens are more important than the offer's actual fit.

Know what the evidence does not tell you

Common landing-page advice can be useful hypotheses, but the supplied bibliography did not establish research-grade universal effects for headline formulas, specific CTA wording, button colour, optimal field count, trust seals, testimonial/logo/FAQ effect sizes, above-the-fold lift claims or video lift percentages.

Suggested steps

  1. Write a testable hypothesis with audience, context, intervention, primary outcome, guardrails, sample or decision rule and stopping criteria.
  2. Record source, date, device, traffic mix, page version, definition of conversion and important exclusions before interpreting a result.
  3. Separate attention, click-through, registration, purchase, qualified lead, retention and customer satisfaction; a gain at one stage can be neutral or harmful downstream.
  4. Report uncertainty and heterogeneity. A zero average is not zero effect, a correlation is not causation and a result that worked once is not automatically portable.
  5. Reject any specific number that arrives without a source, scope and method. The supplied rule of thumb is that many precise claims for untested elements originate in vendor benchmarks, not controlled studies.
Caution: This guide preserves the supplied source numbers with attribution and context where feasible. It does not claim that every bibliography item has been independently reanalysed, that all research has been verified, or that any result is universal.

Sources to consult

References & further reading

These references are supplied starting points for checking the evidence and context behind the guide. Peer-reviewed, industry and observational sources are not interchangeable, and none is presented as proof of a universal causal effect.

Cite or share the resource

If you refer to this guide, here is optional citation text you can adapt. There is no required attribution, license or reciprocal-link arrangement.

Optional HTML attribution link

Use this only if it suits your page. A reciprocal link is not required.

Keep the guide handy

Download the ungated PDF for your team or return to the copyable web edition whenever you need it.

Download PDF
Click here if you have any questions!