The question a visitor asks in the first five seconds is not "should I buy this?" It is "is this for me?" If the answer is unclear, they leave. Clarity is the precondition for every other lever — trust, motivation, social proof — because none of those levers operate on a visitor who doesn't yet understand what the product is. PageLint runs the Clarity lens first because a page that fails clarity doesn't need better social proof. It needs a rewritten hero. This article explains what the eight Clarity checks test and why each one exists.
Why clarity is the first lens
Ogilvy's arithmetic: five times as many people read the headline as read the body copy. The implication is not to write more headlines — it is to write one better one.
On the average, five times as many people read the headline as read the body copy. When you have written your headline, you have spent eighty cents out of your dollar.
CXL Institute's five-second test formalizes what Ogilvy described as reader behavior: scan the dominant visual element, make a stay-or-go decision, then read or leave. The test asks four questions answerable in five seconds: what does this do, who is it for, what's the CTA, what's the benefit? When participants can't answer those questions from the hero, they leave — and in practice, they do leave. The five-second decision is real.
The Clarity lens operationalizes this sequence into eight checkable conditions. They run in order because the output of earlier checks is a precondition for later ones. CL-1 verifies that a dominant H1 exists. CL-2 verifies that the headings beneath it nest logically. CL-3 verifies that the full hero is comprehensible. The remaining five checks address specific failure modes that survive that initial sequence.
CL-1: Single dominant H1 in hero
One H1, visually dominant, semantically marked, in the hero region. Two H1s create a competition for attention that neither wins — eye-tracking splits fixation between them, and the visitor anchors on neither. Zero H1s mean the semantic anchor is missing entirely.
The check is programmatic. It parses the page DOM and counts <h1> elements. Copies with identical text count once. Exactly one is a pass. Zero or more than one is a fail. The most common failure mode is not an absence of headlines — it is an abundance of them. Eyebrow, H1, styled sub-head at H2 weight, a second statement at H1 size, and a CTA with a sentence in it. Each element individually looks reasonable. Collectively they eliminate the hierarchy that makes any one of them the primary signal.
The full argument for CL-1 is in the dedicated article, including the three most common patterns and how to diagnose them from the DOM rather than the visual design.
CL-2: Heading hierarchy is logical
Headings should nest: an H1 before any H2 or H3, and no skipped levels. A page with H2s and H3s but no H1 is flagged medium. A page that jumps from H1 straight to H3 is flagged low. The check is programmatic: it reads the heading tags in the DOM.
This looks like an accessibility nicety, and it is one. It is also the outline a scanning reader uses. Screen readers announce it, search engines read it, and a visitor skimming the page reads the H2s as the table of contents. When the hierarchy is broken, the page has no outline to skim.
CL-3: Hero passes the 5-second test (what + who)
The CXL test applied to the full hero block: a first-time visitor should understand WHAT the product does and WHO it is for within five seconds. The engine sends the H1, subhead, CTA labels and the opening of the body text to a language model, with no description of the product, and asks whether both are clear. An unclear audience is flagged low, an unclear product is flagged medium, and both together are flagged high.
The "what" half usually fails on metaphor. "Reimagine the future of customer connections" contains no category word, so a visitor can't say what the product does. "CRM for indie SaaS founders" carries two signals: the product type (CRM) and the audience (indie SaaS founders).
The "who" half fails on "for everyone," which is an audience of no one. "For solo founders launching their first SaaS product" lets the right reader feel addressed and the wrong reader self-select out. Both outcomes are correct. Strong implication also counts: a hero full of developer terminology implies a developer audience without stating it.
If the engine can't answer "what is this?" from five seconds of hero text, neither can the visitor.
The check fails most often on pages with strong visual design but weak copy — the product screenshot carries the explanatory load, but the text doesn't. A visitor on a slow connection, or using a screen reader, or just reading fast, will miss what only the image was saying.
CL-4: Headline specificity
CL-4 only counts when CL-3 passes. If CL-3 fails, CL-4 is marked passing, so the same failure isn't counted twice. Once the visitor can tell what the product is, the question becomes whether the headline makes a concrete claim (an outcome, a number, a point of difference) or one that could describe many unrelated products. "Work smarter, together" fits a calendar, a CRM, and a chat app. "Cut sprint planning from two hours to twenty minutes" fits one product.
The check is an LLM evaluation. A headline that names the category but promises nothing specific is the most common CL-4 failure.
CL-5: Jargon and insider terminology
Vocabulary should match one clear target audience. CL-5 flags jargon in the hero with no plain-language gloss, clusters of unexplained technical terms anywhere on the page, and audience tension, where the page targets two audiences but uses only one audience's vocabulary. A hero that reads like an internal Slack thread is flagged high.
Words like leverage, synergize, seamless, intelligent, robust, world-class, and next-gen are the usual offenders. Each one substitutes for a specific claim without making one. "Intelligent CRM" says less than "CRM that routes leads based on the topic of your last conversation with each contact." The full list and a replacement for each word is in the jargon article.
The check is an LLM evaluation, not a word filter. A term your audience really uses, glossed nearby, passes.
CL-6: Single message focus
The page should push one primary CTA or benefit, not split attention across competing offers. Two CTAs aimed at the same goal are flagged low. Three or more CTAs pointing at different products or offers are flagged medium. A visitor forced into a meta-decision ("which of these buttons is for me?") is flagged high.
Ogilvy's one-promise rule applies to the page as a whole: "Start free trial," "Book a demo," and "Download the whitepaper" as equal choices are three promises. Pick the one that matters and let the others recede.
CL-7: H1 doesn't use negation as primary frame
"Don't miss out." "No more losing leads." "Stop wasting money on traffic that doesn't convert." Negation in headlines reads as loss-aversion framing, and loss-aversion framing works — but it works better in body copy than in headlines, for a specific reason.
Avoid negatives in your headlines. Headlines are typically read at speed. The 'don't' drops out.
Reading speed strips the modifier. "Don't miss out" scanned quickly leaves "miss out" as the semantic residue. "No more losing leads" leaves "losing leads." The page has primed the reader for the problem, not the solution. The negative was intended to motivate action through loss framing; what it actually does is anchor the reader on the undesired state.
The prescription is not to avoid loss framing — it is to apply it in body copy, where reading speed is lower and the "don't" has time to register. In the headline, positive framing is more reliable because the full sentence gets processed, not just the content words.
CL-7 is a programmatic check. It checks the first two words of the H1 for a negation word: "don't," "do not," "no more," "stop," "never," "cannot," "can't," "won't," "isn't," "aren't," "doesn't." A match is flagged low. Negation that leads the headline has the highest rate of modifier-drop at scanning speed. Negation later in the headline ("The CRM that doesn't require a six-month implementation") is not flagged — the structure is different and the modifier is less likely to drop.
CL-8: Mean sentence length below 22 words
Ogilvy's prescription in Ogilvy on Advertising (1983) is direct: "Use short sentences." CXL's readability research puts the comfortable-parsing upper bound for web reading at approximately 22 words per sentence. Above 35 words, comprehension drops measurably. The check is computed across the page's main text, with navigation, header and footer removed.
The check reports the mean sentence length. A mean of 22 words or less passes. Above 22 is flagged low, and above 30 is flagged medium. Pages with fewer than five sentences pass, because there is too little text to judge. The fix is almost always structural: one long sentence contains two or three ideas that each deserve their own sentence. Splitting them doesn't lose information — it makes the same information easier to process.
This is a programmatic check with no LLM component. Word count across sentence boundaries is deterministic. A line break also ends a sentence, so headings and bullets count as their own short sentences. Fragments of one word are dropped.
The sequence matters
The engine runs all eight checks on every audit, and none waits for another. They still read in order, because earlier checks are structural prerequisites for later ones.
CL-1 and CL-2 are structural: one H1, and headings that nest beneath it. CL-3 is the comprehension test, and CL-4 only counts once CL-3 passes. There is no point grading a headline's specificity when the visitor can't yet tell what the product is.
CL-5, CL-6, CL-7, and CL-8 target specific failure modes and can fail on a page that passes everything else. A hero can pass CL-1 through CL-4 and fail CL-5 because of one unexplained term. CL-8 can fail on a page where the H1 is perfect and the subhead is good but the third sentence of the body paragraph runs to 47 words.
What clarity unlocks
A page that passes all eight Clarity checks has a hero that any first-time visitor can decode in five seconds. That is a necessary condition, not a sufficient one. Passing clarity doesn't guarantee conversion — it means the visitor now has enough information to make an initial evaluation. The Trust lens, the Motivation lens, and the Relevance lens operate on that foundation.
Before that foundation exists, they don't. Social proof doesn't convert a confused visitor. A compelling offer doesn't close someone who doesn't yet understand what they're being offered. The Clarity lens is not the most exciting part of copywriting — Cialdini's trust mechanisms and Schwartz's awareness staging have more surface-area for discussion. But clarity is the part that makes everything else work.
PageLint runs every lens even when a page fails CL-1 through CL-3. Fix those findings first anyway. Not because the other checks don't matter. Because fixing clarity first creates the page context that makes the other checks meaningful to evaluate.
Sources cited
Primary:
- Ogilvy, David. Confessions of an Advertising Man. Atheneum, 1963. Ch. VI: "How to Write Potent Copy." (The 80-cents-out-of-a-dollar arithmetic and the negation-in-headlines warning.)
- Ogilvy, David. Ogilvy on Advertising. Crown, 1983. Ch. 7: "Wanted: a renaissance in print advertising." (The sentence-length and readability prescriptions.)
Secondary and corroborating:
- CXL Institute. Five-Second Test methodology and four-question framework. (Operationalization of the five-second visitor decision; basis for CL-3.)
- Hopkins, Claude. Scientific Advertising. Lord & Thomas, 1923. Ch. 6: "Being Specific." (The original argument for specificity over category claims; required reading per Ogilvy; foundational to CL-4 and CL-5.)
Further reading:
- CL-1: Single dominant H1 in hero — full treatment of the structural headline check.
- Hero five-second test — deep dive into CL-3 and the CXL four-question framework.
- Schwartz awareness states — the awareness-state model that gives context to what "clarity" means for different traffic segments.
- PageLint principles — complete framework reference for all 55 checks across 9 lenses.