top of page
ConverSight.ai
Conversight.ai is a data analytics platform that helps users find their way to pro actively analyze the performance of business data and make better decisions for a better tommorow.
TEAM
CEO, CTO, 2 UI Front end developers, 3 Customer success, 1 Product Manager
MY ROLE
Product designer strategizing user experience with the CEO and product owners till coordinating the execution with the development to customer implementation team
TIMELINE
May 2024 - July 2024
Project Context
A tool to empower Supply Chain Demand Planners with easy forecasting of business data using no-code methods
A Quick Overview
Whats the problem?
Business leaders and managers in the supply chain industry often spend much time forecasting sales and product requirements for different locations.
Its a painstaking process of collecting all the historical data and build multiple formulas to see where we are going to go next.
Why does it take 5 days work, 5 different platforms, and 5000 errors???
When will I make business decisions?
How does Conversight provide AI-powered value?
With high-end technology, LLMs,
Conversight build models to run forecasts
by creating different configurations, helping users to forecast.
My Goals for building demand forecasting tools
Curating the user experience to build forecast models for a user group who are not tech savvy or have been using Excel sheets to make the calculations
Breakdown complex forecasting automation models into an easy workflow with simple, non overwhelming User Interface
Guiding users throughout while educating about the quality of data they are forecast.
Minimal Inputs and Maximum Output
Achievements
10%
Revenue growth
50 new customers
Increasing engagment and adoption
70%
Reduction in customer implementation time
Outcome
See how I broke down the product into an experience
1
Building idea to Research
We adopted multiple different methods and sources to uncover hidden truths for the opputunities.
Domain Research and Technical understanding from data scientists on how to build forecast models to understand the terms and Navigational attributes.
Conversations, and and brainstorming with the product managers, Product owners, Leadership team and developers to curate user and feature requirements.
Created and curated User interviews to understand the user's pain points, needs, and motivations, to derive.
Why do Forecasting?
Demand forecasting predicts future customer needs using historical data and trends.
It helps businesses plan inventory, optimize operations, and meet demand efficiently.
What did we discover from Market Research ?
Anaplan
Limited Custom Scenarios:
​
Tools may lack granular scenario-building for niche or complex markets.
Inventoro
c3ai
Tellius
Gaps in Real-Time Data:
Insufficient integration with dynamic external factors like weather or economics.
Accessibility Issues:
AI insights may not be intuitive enough for non-technical users to act swiftly.
Understanding different stakeholders
Wearing multiple faces to collaborate with different stakeholders and understand from their shoes
Product owners
Data scientist
PHASE 1
Primary User
Secondary User
PHASE 2
Customer success
Developers
PHASE 3
How did we deep-dived into user mindsets by interviewing them?
Conversing with 15 planners who forecast regularly with the customer success teams, building out the resources of user needs
Learning about Gen AI and LLM capabilities that CS has developed.
Understanding the ability of forecast data models and the technicalities built inside conversight.
Technical understanding
Product understanding
Users
Stakeholders
Developers
Simplification of forecasting, launching, modelling, guide
Guided actions, insights, Cta, navigation flow
Seperate Components for scalability
Number of models to build, pricing, cloud solutions
Forecast modelling components, levers, metric
Metrics and impact with reasoning along side every product
Showing historical data along side forecast data
Controls for API calls, socket push
Flexibility and guiding factors for the modelling.
Analysis at SKU level, identification on outlier numbers, mathematical model
Restriction on running cost and speed at which the models
Collecting data with the same component from exisiting product
Key Findings
Learned about the company's growth and business unit restructuring, highlighting the need for forecasting in the supply chain.
Discovered product requirements and the team’s expectations for the platform's features and functionalities
Identified day-to-day challenges and pain points faced by case workers and managers, emphasizing the necessity for a more efficient and user-friendly interface
2
Research to Synthesis
Putting together all findings into jouneys mapping patterns, themes , needs and wants from all perspectives.
Mapping insights into user journey touchpoints:
Organized findings into user flows capturing critical moments where users make, validate, or adjust forecasting decisions.
Identifying patterns, behaviors, and friction points:
Synthesized interviews, observations, and survey data to surface recurring themes like decision anxiety, model confidence, and preference needs.
Framing actionable user needs and business goals:
Converted patterns into structured need statements, bridging user expectations and business outcomes to set a strong foundation for solution brainstorming.
Behavior Mapping
Evaluating existing systems, Users, and behaviors by collaborating cross-functionally to find the missing gaps and opportunities.
Methods of Synthesis
Mapped user journeys across systems and behaviors to trace decision flows, and used affinity mapping to cluster insights into patterns, needs, and opportunity areas.
What is the Design Opportunity we derived
Enhance usability with a clear, step-by-step user flow that provides intuitive guidance and relevant information for non-tech users to create accurate forecasts.
From
Time extensive setup for customer success teams
To
Easy learning curve and usability
Aiming to build a
scalable, simple integration framework,
for flexible automated forecasting on data a
get better visibility on business data
with enhanced visualisations
3
Synthesis to Solution
We derived the user flow, phasing out the product development growth, choosing different methods of evaluation and iterating based on internal feedback on wireframes, mid-fidelity.
Based on research findings, I mapped three alternative user journeys to solve the forecasting configuration challenge, ensuring a balance between user control and system guidance.
worked closely with developers and product stakeholders to review technical feasibility, business goals, and user mental models for each approach, facilitating structured decision-making discussions.
After multiple iterations and validations, we finalized the workflow that optimized decision clarity, user confidence, and forecast accuracy — leading to a scalable high-fidelity prototype for implementation.
User Journey Mapping
For a shorter time to value while handling huge amounts of data
Solution A
Config → Method → Forecast
Start by creating a config, choose forecasting method, then generate forecast.
"Users felt disconnected from data and lacked visibility into what they were forecasting."
Solution B
Data → Method → Forecast
Start by creating a config, choose forecasting method, then generate forecast.
"Caused decision fatigue. Users struggled choosing methods without setting parameters/preferences."
Solution C
This was choosen
Data → Method → Forecast
Start by creating a config, choose forecasting method, then generate forecast.
" Gave the user a structured, thoughtful flow aligning forecasting strategy with data, making the experience predictable, scalable, and user-intent driven."
Visualizations of different iterations
Layouts, navigation, Infomation overload
Round 1
Technical constraints affecting the navigation
Round 2
Round 3
Cost of model running, security and access, and feature prioritisation
Solutions driving Business Outcomes
User Adoption
Higher adoption rate as users experience success quickly without feeling overwhelmed.
Customer Service Tickets
Drop in customer support load; users can independently and confidently create forecasts.
Revenue Impact
Increased realized revenue from better inventory alignment to demand patterns.
Forecast Accuracy
Significant lift in forecast accuracy, resulting in smarter inventory, production, and distribution decisions.
Look at the how the product turned out
1
Give users the actions they can perform easily on top of the configurations
Quick Actionable items to perform on top of data and tables
Clear, clean data loaded tables for ease in usability and readbility.
Guiding users and helping them learn about forecasting with simple models along the way
Constant status updates of what the system is doing based on the action given.
2
Mananging to show a lot of numbers to consume them easily and in order.
Navigation made easy and bold ensuring flexibility
Actions and workspace divided in proper ratios making the user feel comfortable
Adding tooltips at every point to convey the action they can perform at any point making the user autonomous.
3
Navigation through step by step process with flows enriching users in controlling the flow
Intuitive actions for the users
Clean layouts, tags, and findings for large sets of enterprise data
Prioritising information flow and keeping them on under clicks.
Problems to solve
An Interesting Design Problem I solved
What was the problem
Balancing High data density and fragmented actions overwhelmed users, and Lack of trust in AI-suggested actions due to unclear recommendations and scattered manual override options.
Actionable Clarity
Prioritized AI recommendations directly next to forecast accuracy, making user decisions obvious and confident.
Seamless Decision Making
Created a compact action panel (Include, Exclude, Apply Growth) minimizing context switching and accelerating action-taking.
Guided Trust in AI
Introduced smart labels and intuitive badges (Included, Excluded, Manual) that explained AI suggestions conversationally and visually.
Focus Through Smart Filters
Enabled users to slice data by outliers, segment patterns, and variance levels — spotlighting what matters most without overwhelming.
Forecast Component Designs
Built feasible components for forecasting on top of libraries to reduce development resources.
Key aspects I achieved?
A path of discover , ideation and create while constantly learning and recreating.
Flexible Approach
There is no one-size-fits-all process; it's about adapting and combining multiple methods to solve the problem effectively
User Validation
Frequent testing with real users ensures the product meets their needs and provides valuable insights for improvement.
Stakeholder Sync
Collaborating closely with stakeholders ensures feasibility and alignment with the product roadmap throughout the project.
What did I do for end-to-end development of the project
01
Consistent Framework
Standardization ensures design consistency, making the user experience seamless across various touchpoints.
02
Structured Records
Comprehensive documentation captures key decisions, enabling smooth collaboration and future reference.
03
Cross-team Synergy
Effective collaboration across teams fosters alignment, combining diverse perspectives for stronger outcomes.
04
Scenario Narratives
Storytelling through use cases brings design solutions to life, helping stakeholders visualize the user journey and impact.
HI! Do reach out to me to know more about how I solved this problem for Conversight and created a simple experience for a data complex medium
bottom of page