What are companies A/B testing on their landing-page CTAs right now?

CTA is the fourth most-tested landing area with 232 detected experiments, behind hero (488), offer (301), and pricing (254).[1] The active themes are free-trial wording (77, 33%), price-in-button (53, 23%), and demo paths (25, 11%).[2] All figures are detected before/after diffs with inferred rationale, not measured lift.

232 detected CTA experiments — the #4 most-tested landing area, July 2026.

Lazyweb Research · n=232 · Published 2026-07-07

landing-pageexperimentsux-patternsmobileweb

Finding: CTA sits mid-pack in test volume, mobile-heavy

CTA draws 232 experiments and skews mobile (141 mobile vs 91 web).[1][3]

AreaExperimentsAvg impact
HERO4883.28
OFFER3013.69
PRICING2543.52
CTA2323.16

The mobile tilt matters: more CTA experimentation happens on phones than on desktop.

Breakdown: the CTA test menu

The 232 CTA experiments cluster into a handful of themes (non-exclusive):[2]

ThemeCountShare
Free-trial wording7733%
Price in button5323%
Demo path2511%
Waitlist1<1%

Beyond wording, the qualitative rows show structural swaps — removing a CLI install widget or an AI prompt box in favor of a plain account-first button.

Breakdown: structural CTA swaps

Two detected examples go beyond copy (observed change + inferred rationale):[3]

  • Factory (web): a hero CLI install widget ('curl ... | sh' + OS tabs) replaced with a single 'START BUILDING' button. Inferred rationale: enterprise-buyer traffic isn't terminal-ready, so an account-first funnel guides installation later.
  • Replo (web): an interactive 'Build With AI' prompt box replaced with a standard 'Sign Up For Free' / 'Book a Demo' pair. Inferred rationale: serious buyers want a workflow platform, not a toy demo.

The numbers

StatComputed from
CTA 232 experiments, #4 area (hero 488, offer 301, pricing 254)experiment_area_leaderboard / cta_experiments_total
Free-trial 77 (33%), price-in-CTA 53 (23%), demo 25 (11%), waitlist 1 of 232cta_change_themes
CTA platform split 141 mobile / 91 web; Factory and Replo examplescta_platform_split / qualitative[factory, replo]
Methodology. Universe: 232 detected CTA experiments within 2,160 annotated, July 2026. Method: area annotation counts, keyword incidence, and named-company qualitative rows. Caveat: observed changes with inferred rationale, not measured A/B lift.

Sources & citations

  1. [1] Lazyweb Research analysis of 2,160 annotated experiments (detected-experiment corpus), July 2026. Area annotation counts; impact is a model-assigned 1-5 relative score.
  2. [2] Lazyweb Research analysis of 232 CTA experiments (detected-experiment corpus), July 2026. Non-exclusive keyword incidence over summary+learning text.
  3. [3] Lazyweb Research analysis of 225 named-company CTA experiments (detected-experiment corpus), July 2026. Platform split and observed before/after CTA changes with model-inferred rationale.

Source: Lazyweb Research — proprietary analysis of real, in-market app screens. Cite as Lazyweb Research, 2026-07-07.

Related questions

Explore the underlying screens, flows, and A/B tests inside Lazyweb. More research