The Landing Page Validation Guide: Measuring Demand Before the Product Exists
A landing page test (smoke test) collects behavioral demand evidence while the product doesn't yet exist: you put your value proposition on a page, send traffic from the target segment, and measure how many take a concrete step (email, pre-registration, pre-payment). It sits one rung above the interview's "words" evidence: here, people do something.
The Experiment's Design: What Are You Measuring?
A landing page test actually tests two things at once, and they must be separated:
- Message test: Does the value proposition resonate in this segment? (Click-through rate + time on page)
- Demand test: Does resonance convert to action? (Conversion: signup/pre-order)
The one-page, one-segment, one-promise rule applies: testing two promises on the same page hides which one works. If you run variants (A/B), change a single variable headline OR price OR image.
Page Structure: The Converting Skeleton
- Headline: Your positioning formula "for [audience], we solve [problem] with [solution]"
- Subtitle: One sentence on how it works + the key difference
- Visual/mockup: A product concept visual an honest concept, not a fake "screenshot" of a product that doesn't exist
- 3 benefit blocks: Outcome-driven, with numbers
- Trust signals in place of social proof: With no customers yet: team/methodology/"first 100 users get X" framing
- One CTA: The action the test measures no other exits on the page
The CTA's commitment level sets the evidence strength: leaving an email (weak-medium) < survey+email (medium) < "pre-order for $X" (strong). Start with email for early ideas; add a pre-payment layer when the signal gets serious.
Traffic: Whom Are You Measuring?
The result's meaning depends on the traffic's segment purity: low conversion on irrelevant traffic proves nothing. Source options:
| Source | Strength | Weakness |
|---|---|---|
| Targeted ads (Meta/Google) | Segment control, speed, scale | Costs budget; ad quality affects the result |
| Community posts | Pure segment, zero cost | One-shot; community rules |
| Interview participants | The warmest audience | Small sample, possibly biased |
If testing with ads, aim for ~200-500 visitors per page for a minimally meaningful sample; declaring "2% conversion" on 30-50 visitors is reading noise.
Thresholds and Interpretation
Write the threshold before the experiment (hypothesis discipline). Rough references: on cold ad traffic, 5%+ email signup is good and 10%+ is a strong signal; the 1-2% band points to a message or segment problem. On a pre-payment CTA even 1% can be meaningful the higher the price of the action, the stronger the evidence. Read the qualitative layer alongside conversion: a post-signup mini-survey ("what drew you most?"), the domains of incoming emails (company addresses in B2B), and ad comments (the objections are written there).
FAQ
Is a page for a product that doesn't exist ethical?
Yes, on the honesty line: "we're building this join the early access list" is transparent and legitimate; a fake "buy now" button and feature lists presented as live burn trust (and carry legal risk in some countries). If taking pre-payments, the delivery commitment and unconditional refund must be explicit. When the test ends, tell registrants the truth: "we measured demand, we launch on X date" that communication is also the foundation of your first customer community.
Conversion came in low is the idea bad, or the page?
Separate three layers: a low ad click-through rate means a message-segment mismatch (the promise isn't resonating); good clicks but low page conversion means a page problem (clarity, trust, CTA weight); both fine but a drop-off at the payment step means a price/commitment problem. Fix in order and retest with one variable. If no combination passes the threshold in three rounds, the problem isn't the page but the hypothesis consider a segment or problem pivot.
Should I run the landing page test before or after interviews?
The ideal order: 5-10 interviews first, then the landing page. A page without interviews means writing the promise in your own words a headline written without knowing customer language becomes the test's weakest link. Interview findings are the page's raw material: the customer's words in the headline, the three most-heard pains in the benefit blocks. The reverse order (page first) is defensible only for very clear problems with known search volume.
Will competitors see the page and steal my idea?
The practical answer: idea theft happens far less often than founders fear execution capacity is worth more than ideas, and the odds of a competitor monitoring your test page are low. Comforting measures anyway: test under a generic name before finalizing the brand, use narrow geographic ad targeting, and describe the "what" rather than the "how" (promise the outcome without revealing the mechanism). The cost of delaying the test out of secrecy almost always exceeds the cost of the theft risk.