Which growth UI patterns are actually rising versus plateauing in 2025?

Across 730 companies captured in 2025, the clearest risers are biometric login (4.4% to 13.3%, roughly 3x) and AI-assistant screens (2.4% to 8.8%, roughly 4x); social proof crossed into the majority at 53.3% [1]. The clear plateaus-past-peak are countdown timers (12.5% peak in 2024 to 10.1%) and streaks (14.0% peak to 8.1%) [2]. Use this to prioritize: back the risers, treat the peaked patterns as mature levers to test rather than trends to chase.

In 2025, biometric login (13.3%) and AI assistants (8.8%) were the fastest risers; countdown timers and streaks had already peaked (July 2026).

Lazyweb Research · n=730 · Published 2026-07-07

ux-patternsmonetizationretentionmobileexperimentsdesign

Risers, plateaus, and majorities at a glance

All figures are share of distinct companies captured that year (reliable window 2023-2025):

Pattern202320242025Verdict
Biometric login4.4%7.8%13.3%Rising (steepest) [1]
AI assistant2.4%6.0%8.8%Rising [1]
Widget upsell3.4%5.3%7.5%Rising (steady) [1]
Passkey login1.0%1.0%4.1%Rising (new, 2025) [1]
Social proof25.6%48.9%53.3%Now majority [1]
Referral10.6%22.1%19.0%Plateaued [2]
Countdown timer5.1%12.5%10.1%Past peak [2]
Streak6.8%14.0%8.1%Past peak [2]
AI media generator2.4%3.5%3.4%Flat [2]

How to prioritize from this

For a roadmap bet, weight the monotonic risers - biometric login, AI assistants, and the steady widget upsell - since you would be joining growing behaviors rather than plateaued ones. Social proof at 53.3% is effectively table stakes: its absence, not its presence, is now the outlier [1]. The peaked patterns (countdown timers, streaks, referral) are mature: still common and worth A/B testing against your baseline, but not evidence of a rising tide - adopt them for fit, not momentum [2].

One trend to distrust

The apparent paywall decline (labeled share 43.7% -> 28.1%) is a labeling-coverage artifact, not a real drop: the is_paywall flag is NULL on 21,824 of 23,407 canonical screens [3]. Do not include 'paywalls declining' in any prioritization; use tag-based patterns for direction instead.

Caveats

Every figure is deduped by distinct company and tag-matched, so all values are lower bounds bounded by tag recall [1]. Trends normalize to companies-captured-per-year because capture cadence is highly uneven; never plot raw counts [4]. 2022 (n=1) and partial-2026 are excluded as unreliable, and per-category time cuts are unsupported [4].

The numbers

StatComputed from
2025 risers/majorities: biometric 13.3%, AI assistant 8.8%, widget 7.5%, passkey 4.1%, social proof 53.3%biometric_login_prevalence_over_time / ai_assistant_prevalence_over_time / widget_upsell_prevalence_over_time / passkey_prevalence_over_time / social_proof_prevalence_over_time
Peaked/flat: countdown 12.5%(2024)->10.1%, streak 14.0%(2024)->8.1%, referral 22.1%(2024)->19.0%, AI media gen ~3.4% flatcountdown_timer_prevalence_over_time / streak_prevalence_over_time / referral_prevalence_over_time / ai_media_generator_prevalence_over_time
Paywall labeled share 43.7%->28.1% is an artifact; is_paywall NULL on 21,824/23,407 canonical rowspaywall_screen_prevalence_over_time_CAUTION
2025 cohort 730 companies; trends normalized to companies/year; 2022 (n=1) and partial-2026 excluded; category time cuts unsupportedcapture_cohort_by_year / capture_cadence_note / smallSampleWarnings
Methodology. Universe: 47,578 capture-dated screenshots across 809 tracked mobile apps (730 companies captured in 2025). Method: for each pattern, share of distinct companies captured each year whose screens carry the pattern's tag family. Reliable window 2023-2025; 2022 (n=1) and partial-2026 excluded; per-category time cuts unsupported. Caveat: all values are tag-recall-bounded lower bounds.

Sources & citations

  1. [1] Lazyweb Research analysis of 809 tracked apps (mobile-app corpus, 47,578 capture-dated screenshots), July 2026. Per-pattern prevalence = share of companies captured each year whose screens carry the pattern's tag family, deduped by company_name.
  2. [2] Lazyweb Research analysis of 809 tracked apps (mobile-app corpus, 23,407 canonical screens), July 2026. is_paywall labeled share and its NULL-coverage gap (21,824/23,407); the apparent decline is a coverage artifact.

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