Do market leaders run more experiments than challengers?
Yes, on average — though the median ties. Across 121 leader apps we detected 2,189 distinct UI variations (avg 18.1 per company) versus 15.3 per company for 85 challengers, and leaders own the high tail (max 221 vs 66).[1] Both groups have a median of 10, so the average gap is driven by a handful of prolific leaders.[1] These are detected before/after UI diffs with inferred rationale, not confirmed A/B outcomes.[1]
Market leaders average 18.1 detected UI experiments per app vs 15.3 for challengers (2,189 across 121 apps) — Lazyweb Research, July 2026.
The finding
Leaders show more detected experimentation on average and dominate the extreme tail:
| Group | Apps | Distinct experiments | Avg / app | Median / app | Max |
|---|---|---|---|---|---|
| Leaders | 121 | 2,189 | 18.1 | 10 | 221 |
| Challengers | 85 | 1,299 | 15.3 | 10 | 66 |
The median is identical (10), so the typical leader and typical challenger test at a similar clip. The average gap (18.1 vs 15.3) comes from prolific leaders like DoorDash (221), Wattpad (85), Brainly (77), Zoom (74), and Grammarly (67).[1][2]
How to apply it
Two lessons. First, the median tie means you don't need a giant experiment program to look like a leader — a steady ~10 detected variations is table stakes on both sides. Second, the tail is where leaders separate: the biggest names run dramatically more (DoorDash's 221 dwarfs the challenger max of 66).[1][2] If you're already at the median, the leader move is sustained volume over time, not a one-off burst.
Caveats
These are detected before/after UI diffs with LLM-inferred rationale — not measured lift or confirmed A/B tests, so 'more experiments' means more observed UI variation, not more wins.[1] The average is skewed by the tail (DoorDash 221), which is why the tied median (10) matters more for a typical team. Only 121 of 286 leader-flagged apps have detected experiments, so this describes the captured sample.[1]
The numbers
| Stat | Computed from |
|---|---|
| Leaders: 121 apps, 2,189 distinct experiments, avg 18.1/app, median 10, max 221 | experiment_volume_leaders_vs_rest |
| Challengers: 85 apps, 1,299 experiments, avg 15.3/app, median 10, max 66 | experiment_volume_leaders_vs_rest |
| Top leader experiment volumes: DoorDash 221, Wattpad 85, Brainly 77, Zoom 74, Grammarly 67, Audible 64, Chime 56, ChatGPT 53, Flo 52 | named_leader_high_experiment_volume |
Sources & citations
- [1] Lazyweb Research analysis of 3,488 detected experiments across 206 apps (leaders vs challengers), July 2026. COUNT DISTINCT experiment_id per company from detected before/after UI diffs; 121 leader apps (2,189) vs 85 challenger apps (1,299). Detected variations with inferred rationale, not measured A/B lift. Joined to companies.market_leader on lower(company_name). ↩
- [2] Lazyweb Research analysis of named market-leader experiment volumes, July 2026. Market_leader='Yes' apps with the most detected distinct experiments. ↩
Source: Lazyweb Research — proprietary analysis of real, in-market app screens. Cite as Lazyweb Research, 2026-07-07.