The five-second test is not a design test. It is a comprehension test. Show the hero to a participant for five seconds. Hide it. Ask four questions. If the participant cannot answer "what does this product do" from memory, the hero has failed — not the product, not the offer, not the price. The copy. This is the long version of PageLint check CL-3: Hero passes 'what is this?' test.
What CXL actually measures
CXL Institute's formalized version of the five-second test uses a four-question framework developed by Angie Schottmuller. The questions are asked immediately after the five-second exposure, with the page hidden:
- What does this company or product do?
- Who is it for?
- What is the primary call to action?
- What is the main benefit or outcome?
The five-second test is not about whether participants like the design. It measures whether the page communicates its core value proposition fast enough to survive the real-world attention environment — where visitors make stay-or-go decisions in under eight seconds.
A hero passes CL-3 under the following conditions: question 1 — "what does this product do?" — is answerable from the H1 alone, in plain language. Questions 2 through 4 are answerable from H1 plus subhead plus CTA combined. If the H1 cannot carry question 1 unassisted, the hero has a structural comprehension problem that no design improvement will fix.
The underlying reason these four questions matter: they map to the four decisions a visitor makes before converting. The product decision (do I understand this?), the fit decision (is this for me?), the action decision (do I know what to do next?), and the value decision (is this worth my time?). The five-second test forces the copy to answer all four before the visitor reaches the cognitive stage where objections form.
The Schwartz connection
The verbalization concept in Schwartz's Breakthrough Advertising (1966) is the theoretical foundation for what CXL operationalized:
Copy cannot create desire for a product. It can only take the hopes, dreams, fears and desires that already exist in the hearts of millions of people, and focus those already-existing desires onto a particular product. This is the copywriter's task: not to create this mass desire but to channel and direct it.
Channeling desire requires a match between what the visitor already wants and what the copy immediately offers. That match depends on the visitor being able to verbalize what they're being offered — to form a clean mental model of the product — before they will take action. If the mental model is unclear, action is blocked regardless of how strong the desire is. The five-second test measures the speed and accuracy with which that mental model forms.
CXL's research, applied across thousands of test sessions, consistently confirms: pages where participants answer question 1 incorrectly or with "I don't know" convert at dramatically lower rates than pages where the answer is immediate and specific. The correlation holds even when the incorrect mental model is close to accurate. A visitor who thinks "this is a landing page template builder" when it's actually a landing page auditor is not in the right frame to convert — they'll evaluate based on the wrong criteria.
What the check tests
CL-3 is an LLM check, not a programmatic one. PageLint passes the H1 and subhead text to a zero-context prompt:
You are a first-time visitor. You saw this page for 5 seconds and it was hidden. Answer these questions from memory only: (1) What does this product do? (2) Who is it for?
If the LLM response is accurate and specific relative to the actual product description, the check passes. If the response hallucinates a different product category, describes the product in generic terms ("some kind of analytics tool"), or returns uncertainty, the check fails.
The zero-context constraint is critical. The LLM is not told what the product is. It is given only what a first-time visitor would see. Accuracy is measured by comparing the response against the product category and audience declared in SC-1 (the setup check). A response that correctly identifies the product category and broadly correct audience passes. A response that cannot name a category fails.
Confidence for CL-3 is High for text-dominant heroes. It drops to Medium when the hero relies on image-embedded text — OCR is imperfect and the extracted text may be incomplete or out of order.
The seven patterns that pass
These hero patterns produce reliable five-second comprehension across audience types. They are not templates — they are structural approaches.
1. Category + audience
State the product category and who it is for. No metaphor. No abstraction.
2. Verb + outcome + timeframe
The outcome is specific. The timeframe is specific. The verb is active.
3. Problem + solution
State the problem the visitor already experiences, then name the solution. No preamble.
"Stop losing leads to slow follow-up. Automated outreach for solo sales reps." The first sentence lands on a known pain. The second delivers the category and audience simultaneously.
4. Comparison anchor
"Like Notion, but for engineering runbooks." The comparison anchor piggybacks on an existing mental model. The visitor knows what Notion is; they immediately understand the category and use case. The risk: the comparison anchor is only useful if the reference product is universally known to the target audience.
5. Named role + named outcome
"For designers who need client feedback without back-and-forth email." Role (designers), problem (back-and-forth email), outcome (resolved feedback). All three present. Participant can answer all four CXL questions from this headline plus a short subhead.
6. Specific number claim
"Cut invoice-to-payment time from 14 days to 3." The before and after state are both specific. The participant remembers 14 and 3. They know what the product does even if they can't name the category — it makes invoices get paid faster. That mental model is enough to proceed.
7. Job-to-be-done
"Track what's breaking in production before your users report it." The job is named. The timing is specific ("before your users report it"). The visitor who has experienced the job knows immediately this product is for them.
The three patterns that fail
1. Metaphor lead
"Beyond the inbox." Requires the subhead to decode. In a five-second exposure, participants cannot hold the metaphor and translate it. When the page hides, all they retain is "something about an inbox." Category: unknown.
2. Adjective stack
"The intelligent, seamless, next-generation platform." Four adjectives. Zero nouns that could be a product category. Participants remember the adjective-heavy phrasing as a signal of marketing copy and retain nothing specific. Response to CXL question 1: "I have no idea."
3. Category-less statement
"The future of how teams work together." No category. No audience. No mechanism. "Teams" is too broad to constitute an audience signal. "Work together" names a use case so generic it encompasses every productivity tool ever made. Five-second comprehension rate: near zero.
Common mistakes when CL-3 fails
The subhead dependency trap. The H1 is deliberately cryptic, meant to intrigue, and the subhead carries all the actual information. "Beyond the inbox" / "Customer support software for Shopify stores." The five-second test treats H1 and subhead as a unit — but in real visitor behavior, many visitors never reach the subhead. Their eye fixes on the H1, makes a category assessment, and either proceeds or leaves. A cryptic H1 produces wrong category assessments. The subhead arrives too late.
The audience hedge. The hero names a category but refuses to name a specific audience, to avoid excluding anyone. "Landing page optimization for businesses" attempts to be inclusive and ends up being addressable to no one specifically. The visitor for whom the product is perfect cannot recognize themselves in the copy. The five-second response to "who is it for?" returns "businesses, I think." That is not a passing answer.
The feature-first hero. The H1 leads with a feature rather than a use case or outcome. "Real-time AI-powered copy analysis with 20-principle scoring." The participant after five seconds knows there is AI and there is scoring. They do not know what the product is for. The feature is not the product — the outcome is.
How PageLint detects this
LLM check. Confidence: High for pages with textual heroes where H1 and subhead are rendered as real HTML text.
Confidence drops to Medium when:
- The hero uses text embedded in images (requires OCR)
- The H1 is a single proper noun without a descriptor (brand-name-only H1; see CL-1 edge case)
- The page uses animated headline rotators where multiple headlines cycle through a single H1 slot
When confidence is Medium, the check result is shown with a flag indicating lower certainty. Manual verification is recommended.
CL-3 depends on CL-1 passing first. A page with zero H1s or multiple competing H1s will fail CL-1; CL-3 will be marked as blocked until CL-1 is resolved. The structural check must pass before the comprehension check is meaningful.
Related checks
- CL-1: Single dominant H1 in hero — verifies structural dominance before CL-3 runs the comprehension test.
- CL-2: H1 names product category in plain language — tests one dimension of the CL-3 CXL question 1.
- CL-4: Hero answers "who is it for?" — tests CXL question 2 at the structural level.
Related principles
- CL-1 — Single dominant H1 in hero
- CL-2 — H1 names product category in plain language
- CL-4 — Hero answers "who is it for?"
- SC-1 — Awareness state declaration
Sources cited
Primary:
- CXL Institute. Five-Second Test methodology. Schottmuller, Angie. "The 5-Second Test: Measuring Your Site's Content Pages." 2013.
- Schwartz, Eugene M. Breakthrough Advertising. Prentice-Hall, 1966. Ch. 3. (Verbalization concept; theoretical basis for why five-second comprehension predicts conversion.)
Secondary:
- Nielsen, Jakob. "How Long Do Users Stay on Web Pages?" Nielsen Norman Group, 2011. (Eight-second attention window; corroborates CXL's framing of the five-second test as survival-condition copy.)
- Krug, Steve. Don't Make Me Think. New Riders, 2000. Ch. 2. (Self-evident design; the "trunk test" as a parallel comprehension framework for navigation.)
Further reading:
- Schwartz awareness states explained — awareness state determines what "accurate comprehension" means; a hero that correctly describes the product for Solution-Aware visitors may confuse Unaware visitors.
- Single dominant H1 — prerequisite check; CL-3 depends on exactly one H1 being present.
- PageLint principles — full check framework reference.