Datasutram Markets
Overview
My Role: Product Designer
Impact: Increased user retention by 75% and reduced support volume tickets by 40%
Duration: 6 months
Collaborations:
Product Manager
Engineering and Data Science teams
Sales and Business teams
Background
Traditionally, these decisions relied on intuition and followed a spray and paint model with high churn rates.
Challenge: Poor Engagement and Adoption
Despite its potential, the platform faced challenges with adoption and engagement, and clients dropped off.
My Role: Founding Product Designer
Led the UX research and end-to-end design of the platform, along with usability testing, iterative design through client interactions.
Problem
Poor user retention in the previous platform
User Research Method
Interviews: 3 internal business/sales team members and 4 external customers.
Usability Testing: 2 sales representatives and 4 customers.
Pain points
Ambiguous visualizations: Too much left to intuition, leading to disengagement.
Lack of trust: Predictions lacked transparency, making users hesitant to trust the tool.
Frustrating filters: Filters often returned nil or overly broad results, frustrating users.
User scenario
Tara wants to to open stores in high potential hotspots
Tools
Data Analysis and GIS Software
Frequent site visits
Cross-functional collaboration
Motivations
Balance traffic and visibility with operational efficiency.
Complement existing networks while avoiding cannibalization.
Maximize market penetration in high-potential areas.
Pain Points
Balancing the cost of prime locations with potential returns.
Adapting to changing consumer behavior.
Ensuring accessibility for walk-ins and delivery services.
Challenge 1
Lack of Actionable Insights
The platform was underutilized because clients struggled to interpret its broad and ambiguous visualizations. Key insights were left to user intuition, undermining confidence in the tool.
Solution: Introduced ML-powered actionable insights:
Predictions of high-potential properties based on demographic, market, and internal data.
Clear, ranked suggestions for potential store locations
Challenge 2
Lack of Trust in Predictions
Clients were skeptical about trusting the platform’s predictions.
Solution
Enhanced map visualizations and data analytic charts to support predictions and improve transparency.
Introduced a human-in-the-loop model where final decisions (e.g., phone calls, negotiations) were made by site selection teams to mitigate risks.
Recommendations in gallery view with supporting visualizations
Recommendations in list view with supporting visualizations
Challenge 3
Filters returning nil results
Filters often returned nil results or required multiple iterations to refine outputs, and had a lot of data points, which seemed irrelevant to many users. This frustrated users.
Solution
Customized data points per customer
Displayed frequency distribution charts to show potential results before applying filters.
Provided real-time previews of filtered results to reduce trial and error.
Filters with distribution and real time filtered results
Outcomes
Improved User Experience
Improved Engagement
40% increase in platform retention as users found insights actionable and intuitive.
Enhanced Workflow
Clients reported a 2x faster site selection process with improved targeting.
Scalable Design
The modular platform catered to diverse clients while being consistent for legacy users.
Process
Low fidelity prototypes
These helped in communicating ideas with the engineering and business teams for quick feedback.
Iterative design
Out of the many designs, iterative design was most impactful in redesigning the user flow. After the first round of user testing, I simplified the flow from a 7-step to a 3-step process.
Conducted 3 rounds of usability tests with metrics like ease of use, actionability, and time-to-completion.
Collaborated with cross-functional teams to ensure feasibility and alignment with business needs.
Initial flow
Final flow
Collaboration
Engineering
Partnered with Engineering to integrate ML insights seamlessly and functionalities
Product
Worked with the Product team to prioritize datapoints, use cases and feature prioritization
Sales and Marketing
Collaborated with Sales and Marketing for use cases, client interactions and onboarding
Reflections and Learnings
Improved User Experience
Cross-functional collaborations
Involve product and engineering teams early to ensure alignment and feasibility. Sales team is great for direct customer insights.
Focus on transparency and explanations
Providing actionable insights, explanatory data and human-in-the-loop validation improved user confidence in the predictions.
Augmentation rather than automation
The site selection teams were comfortable using the platform once they understood that their work is not going to be automated.
View Other Projects
Financial Planning is crucial but has a steep learning curve. I designed an LLM-driven Financial Advisor for GenZ users to explore personalized strategies to manage their money. Trust-building was a focus in the interactions.
Smart cities are enable by IoT resources for efficient functioning but data collection is opaque. I designed the IoT Privacy Assistant for 17,000+ residents to reduce friction in managing their personal data, reducing the multi-channel process to a single click.
Collective Things
While second-hand furniture is economical, it lacks the emotional connection that helps create a sense of belonging in a new place.
This project explores how shared furniture can help foster a sense of belonging, turning impersonal rooms into their homes.