How Many Detected Experiments Touch Social Proof, and What Changed?

Lazyweb Research detected 120 before/after experiments across 62 companies whose diff mentions social proof — about 2.5% of the 4,814-experiment corpus [1][2]. Separately, 48 experiments are annotated directly in a 'social proof' change area, and 18 hero-area experiments also touch social proof [3][4]. Most social-proof changes are additions (37 added vs 3 removed among classifiable diffs) [5].

120 of 4,814 detected experiments (62 companies) mention social proof; among classifiable changes, additions beat removals 37 to 3 — Lazyweb Research, July 2026.

Lazyweb Research · n=120 · Published 2026-07-07

experimentsux-patternslanding-pagedesign

How much social proof gets tested

Of 4,814 detected experiments spanning 362 companies (2022-10-10 to 2026-07-04), 120 mention social proof — testimonial, star rating, review count, user count, or 'trusted by' — across 62 distinct companies [1][2]. That makes it a mid-frequency tactic: more touched than free-plan removal (10) but far behind discount badges (1,133) [6][7].

Where it shows up and which direction it moves

By annotated change area, 'social proof' is a named area on 48 experiments; hero remains the most-tested area overall at 405 experiments, 18 of which also mention social proof [3][4][8]. Direction of the social-proof changes:

DirectionCount
Added37
Removed3
Mixed6
Unclassifiable74

Among the changes we could classify, additions outnumber removals roughly 12:1 [5].

How to apply it

If you're deciding whether social proof is worth a test slot, the corpus shows it's tested less than pricing or CTA copy but almost always in the additive direction — teams add it and keep it (0 of 29 add-screens later removed it) [9]. A concrete pattern from the corpus: voice-notes scaled up its testimonial wall and framed an aggregate '500,000+' user claim rather than a bare assertion [10].

Caveats

Keyword tagging counts experiments that *mention* social proof, not a curated set of pure social-proof tests; treat 120 as an upper-ish bound on the theme [2]. Learnings are LLM-inferred rationale for observed diffs, not confirmed test outcomes.

The numbers

StatComputed from
120 experiments mention social proof, 62 companiestactic_mentions__social_proof
4,814 detected experiments across 362 companiestotal_detected_experiments; distinct_companies_in_experiment_corpus
48 experiments annotated in the 'social proof' areaannotation_area_breakdown: social proof 48
18 hero-area experiments also mention social proof (of 405)hero_social_proof_experiments
37 added / 3 removed / 6 mixed / 74 otherdirection_split__social_proof
10 experiments mention free-plan removaltactic_mentions__free_plan_removal
1,133 experiments mention discount badgestactic_mentions__discount_badges
hero is the most-annotated area at 405 experimentsannotation_area_breakdown: hero 405
0 of 29 add-screens later removed social proofretention__social_proof
voice-notes scaled testimonial cards; 'Loved by 500,000+'qualitative: voice-notes
Methodology. Universe: 4,814 detected before/after UI experiments (362 companies, 1,358 canonical screens), of which 120 mention social proof; keyword tag on what_changed+learning plus annotation areas, July 2026 pull. Learnings are LLM-inferred, not measured A/B results.

Sources & citations

  1. [1] Lazyweb Research analysis of 4,814 detected experiments (362 companies, 1,358 canonical screens), July 2026. Keyword-tagged on what_changed+learning; distinct companies via control screenshot join.
  2. [2] Lazyweb Research analysis of 120 detected experiments (social-proof-mentioning subset), July 2026. Tagging matches experiments that mention the tactic, not a curated pure-tactic set.

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