Case Study 04
Cricket Betting — Conversion Redesign
Reducing Cognitive Load at the Decision Moment
At a glance
- Problem
- Working the betting funnel stage by stage in Tableau, the largest loss was at match selection — before anyone committed to a bet. The screen buried the one thing the decision turned on.
- What I decided
- Reworked the hierarchy with match clarity first, framed odds and progressive disclosure — trading completeness and density for less ambiguity at the moment of decision.
- Result
- Conversion at match selection went from 8% to 32% in a two-week A/B test on 5% of traffic. It ran during IPL season, so I read it as evidence of where the constraint was, not as a lasting 4× on the business.
- My role
- Owned the funnel analysis, the hierarchy redesign, the test hypothesis and the iteration decisions. The PM led experiment design; the data team validated the result.
01 — Problem Statement
Problem Statement
Conversion was low not because users didn't want to bet — but because the decision moment was cognitively expensive. In a real-money context, ambiguity amplifies hesitation.
The biggest drop-off in the cricket betting funnel occurred at the match selection stage — before users committed to any betting option. That made it the highest-leverage stage in the funnel: every downstream screen, no matter how well designed, was only ever seen by the users who survived this first decision.
Analytics revealed the cause: users struggled to understand match clarity and odds meaning. The information hierarchy buried the primary decision variable. Too many elements competed for attention at the critical moment — the screen was answering questions the user hadn't asked yet, before it answered the one they had.
02 — My Role vs. Team Contribution
My Role vs. Team Contribution
I owned
- Funnel drop-off analysis and insight synthesis
- Information hierarchy redesign — match clarity first
- A/B test hypothesis framing
- Iteration decisions between test variants
Team contributed
- PM: experiment design lead, coordinating with data
- Data: Amplitude analysis and statistical validation
- Engineering: variant implementation
03 — Where the Funnel Actually Broke
Where the Funnel Actually Broke
I found this in Tableau, not in a design review. I was working through the betting funnel stage by stage rather than trusting the aggregate drop-off number, and match selection — the step before anyone commits to anything — was the largest single loss in the funnel.
That made it the highest-leverage stage: every screen downstream of it, however well designed, was only ever seeing the users who had already survived this one. So I took the stage apart with the data team before touching any UI.
What the Funnel Data Showed
- Match clarity was the first thing users needed to resolve, but it wasn't the first thing the screen answered — it was buried under secondary information.
- Odds were shown in their raw form, which reads as a betting-literate power-user format, not a first-decision format.
- Too many elements — visual and informational — competed for attention at the exact moment a user needed to make one specific call.
- The pattern held across match types: this wasn't a one-off layout problem, it was a hierarchy problem.
If match clarity was resolved first, and odds were reframed instead of shown raw, users would convert at a meaningfully higher rate — because in a real-money context, the cost of ambiguity is felt immediately, before a user ever reaches the odds themselves.
04 — The Hardest Decision
The Hardest Decision
Decision
Reworked information hierarchy with match clarity first, simplified odds framing
In real-money contexts, the cost of a wrong decision is felt immediately. Reducing ambiguity at the decision moment is more valuable than visual richness or information completeness.
| Factor | Option A | Option B — Chosen |
|---|---|---|
| Information priority | Feature-rich layout — all info visible | Match clarity first — progressive disclosure |
| Odds presentation | Raw odds — complete but high interpretation load | Framed odds — reduced cognitive cost |
| Visual density | High density — power user optimized | Reduced density — guides scanning toward decision |
05 — Rollout Strategy & Learning
Rollout Strategy & Learning
Controlled A/B test with 5% traffic split over 2 weeks. The experiment ran during IPL season — acknowledged as a confounder, but traffic volume remained stable across control and variant groups during the test window, which kept the comparison meaningful even with a live-cricket-driven surge underneath it.
What went wrong
Some micro-interactions initially underperformed and required mid-test iteration on feedback states and transition cues — the hierarchy was right, but the moment-to-moment feedback wasn't yet reinforcing it.
How it was corrected
Iterated on feedback states after reviewing interim data jointly with PM and data team. Every variant call was a joint decision, not a solo design call made in isolation from what the numbers were actually saying mid-test.
The IPL season context is acknowledged. The number is directional validation that cognitive load was the constraint — not an unqualified 4× claim about the business.
06 — What the Test Showed
What the Test Showed
The variant ran against a 5% traffic split over two weeks, during IPL season. A four-fold lift is a large claim for a hierarchy change, so the caveat below is part of the result, not a footnote to it.
- Conversion: Baseline 8% → variant 32% at match selection — a 4× relative lift, +24 percentage points.
- What that does and does not mean: It is a large effect on a single funnel stage, measured in a two-week window. I would not present it as a durable 4× on the business; I present it as evidence that ambiguity at the decision moment was the binding constraint.
- Confound: The window overlapped IPL season. Traffic volume held steady across control and variant, so the comparison stands — but the absolute level is seasonal, and a replication outside IPL was the obvious next test.
07 — Org-Level Impact
Org-Level Impact
- Established the principle: in high-stakes transactional contexts, information hierarchy matters more than information completeness.
- The funnel audit methodology — mapping drop-off to cognitive friction points — became reusable for other betting products.
Reduce the cognitive cost of the decision moment before optimizing anything else in the funnel.
08 — What I'd Do Differently
What I'd Do Differently
- Run the test outside IPL season for a cleaner baseline.
- Instrument match selection time as a leading indicator of cognitive load, not just conversion rate.
- Test a larger traffic split earlier to reach statistical significance faster.
09 — Design Principles Demonstrated