Optionslab

OptionsLab AI

Designing for Financial Trust: Improving New User Retention by 25% for an AI Trading Simulator.

Context

Role: Product Designer
Timeline: 6 months
Team: 2 engineers, 1 PM, 1 data analyst
Platform: Web application for retail options traders

Impact at a Glance
Impact at a Glance
Impact at a Glance

31% increase in user confidence | 42% reduction in decision time | 23% increase in active trading sessions

The Business Challenge
The Business Challenge
The Business Challenge

Options trading is one of the most high-stakes activities in finance, yet the necessary analytical tools often lead to user confusion and anxiety. The core problem we identified was Analysis Paralysis, users viewing the complex Greeks data (Delta, Gamma, Theta) on the Trade Analytics screen did not know if they should sell now, hold, or adjust the trade. Our goal was to embed AI into these high-friction points to increase user confidence by 30% and reduce emotionally driven trading.

The Problem
The Problem
The Problem

Options traders were paralyzed by their own data.

Early user interviews revealed a critical insight:

"I see all these numbers—Delta, Gamma, IV—but I still don't know if I should sell. I end up just... waiting." — User Research, Week 1

Research Insights (n=24 user interviews, 847 session recordings)

  1. Analysis Paralysis: Users spent avg. 8.3 minutes analyzing a single position without taking action

  2. Greek Literacy Gap: 73% of users couldn't explain what "Theta decay" meant for their position

  3. Risk Blindness: Users didn't realize when they were overexposed until losses occurred

  4. Trust Gap: 89% wanted guidance but didn't trust generic "AI chatbot" interfaces

Key Insight: Users didn't need more data—they needed interpreted data at decision moments.

Options traders were paralyzed by their own data.

Early user interviews revealed a critical insight:

"I see all these numbers—Delta, Gamma, IV—but I still don't know if I should sell. I end up just... waiting." — User Research, Week 1

Research Insights (n=24 user interviews, 847 session recordings)

  1. Analysis Paralysis: Users spent avg. 8.3 minutes analyzing a single position without taking action

  2. Greek Literacy Gap: 73% of users couldn't explain what "Theta decay" meant for their position

  3. Risk Blindness: Users didn't realize when they were overexposed until losses occurred

  4. Trust Gap: 89% wanted guidance but didn't trust generic "AI chatbot" interfaces

Key Insight: Users didn't need more data—they needed interpreted data at decision moments.

Solution 1: AI Exit Strategy - Converting Anxiety into Action
Solution 1: AI Exit Strategy - Converting Anxiety into Action
Solution 1: AI Exit Strategy - Converting Anxiety into Action

The Design

Instead of forcing users to interpret Greeks, I designed a contextual AI button that appears directly into the trade analytics view. When clicked, it generates a structured 3-points analysis.

The Design Decision

Our design prioritized transparency and focus. We strategically used a full-screen modal to separate the AI's plain-language conclusions from the complex analytics screen. This reduced the user's cognitive load by 41%. The structured, multi-part analysis format was chosen because it tested 2.8x more trusted than conversational AI, allowing users to quickly audit the machine's logic and move directly to action.

Impact

  • 42% reduction in time-to-decision (from 8.3 to 4.8 minutes)

  • 31% increase in self-reported confidence scores

  • 19% increase in executed trades after viewing analysis

Solution 2: AI Risk Guardrails — Preventing Overexposure
Solution 2: AI Risk Guardrails — Preventing Overexposure

The Design

I introduced a persistent Portfolio AI Risk Score (1-10 scale) visible on the main trade list screen. The score uses ML to compare current exposure against the user's 90-day risk baseline.

Color-coded system:

  • Green (1-4): Within comfort zone

  • Yellow (5-6): Elevated, monitor

  • Red (7-10): High exposure, review recommended

Design Decisions

Why persistent vs. triggered alerts?
Persistent visibility creates ambient awareness—users self-regulate before risk becomes critical.

Why a 1-10 score vs. percentage?
User testing showed that percentages felt "too exact" and judgmental. A 10-point scale felt more advisory and actionable.

Why non-modal?
Keeping it on-screen (not a pop-up) positioned it as helpful context rather than an interrupting warning.

Impact

  • 47% reduction in over-leveraged positions

  • 23% decrease in support tickets related to risk questions

  • 89% of users reported the score as "useful" or "very useful"

Solution 3: AI Strategy Confidence Scoring — Validating Complex Decisions
Solution 3: AI Strategy Confidence Scoring — Validating Complex Decisions
Solution 3: AI Strategy Confidence Scoring — Validating Complex Decisions

The Design

I designed a contextual AI Strategy Suggestion that surfaces directly within the user's watchlist. When the system detects favorable conditions for a specific stock, a subtle badge appears on the position card showing the recommended strategy.

Tapping the badge reveals a compact validation modal with:

  1. AI Strategy Suggestion Header: Clear labeling with sparkle icon signaling intelligent recommendation

  2. Confidence Score: "75% confidence Score" - translating complex probability into digestible metric

  3. Reasoning Transparency: "Based on current market volatility and your risk profile" - shows what data informed the suggestion

  4. Quick Context: User can see their existing position details (Exp dates, Strikes, Buy/Sell) in the same view

Design Decisions

Why surface suggestions in the watchlist vs. a separate "recommendations" tab?

The Watchlist is the natural decision point. Embedding suggestions here reduced the time-to-action by 35%.

Why a badge on the card vs. inline text?

The badge creates visual hierarchy without cluttering. It was 3.4x more engaging than inline text.

Why a 75% confidence format vs. Low/Medium/High?

Numerical percentages are more credible and specific. This format made users feel 2.1x more confident than subjective terms.

Impact

  • 35% faster time-to-execution for AI-suggested complex trades

  • 61% reduction in strategy-related support tickets ("Can I trust this?" / "How did you calculate this?")

  • +12 NPS points with users specifically praising "intelligent suggestions" and "market responsiveness"

The Results

Qualitative Feedback

"For the first time, I feel like I understand what's happening with my options."
— Beta user feedback

"The AI score stopped me from making a stupid trade. It probably saved me $3K."
— User interview, post-launch

What I Learned

1. AI Trust Requires Transparency

Users don't trust "magic." Showing the reasoning process was more important than the conclusion itself.

2. Constraint Breeds Creativity

The requirement to avoid financial advice (regulatory) pushed me to design educational tools that were more effective than direct recommendations.

3. Speed = Trust in Fintech

Fixing the chart lag had an outsized impact on perceived credibility; sometimes interaction design is as important as information design.

Create a free website with Framer, the website builder loved by startups, designers and agencies.