Lykdat SaaS
LykDat's unique selling proposition (USP) is its visual search feature, positioning it as a fashion tool similar to what Shazam offers for music.
Year
2021
Role
Business Analyst
Context
As a result of this feature, review analytics is a necessary component of our B2B platform: review analytics entails the sentiment analysis on sold products over categories like price, style, fit, material, build quality, etc. Fashion retail chains (our customers) leverage these analyses to make better manufacturing and design decisions.
As a Business Analyst who bridges the gap between business needs and technical implementation, I was part of the team tasked with automating the review analytics process.
Business Opportunity
Due to the ever-increasing need for data by end users, a necessary feature we offer is review analytics. Review analytics provides customers with the unique insights needed to improve products and tailor marketing strategies; however, a considerable percentage of new customers did not have data analysts as part of their teams.
This gap in effectively accessing data led to a lack of a direct way to “talk” to the underlying data, leading to untapped insights and a reduction in customer retention rate year on year (9.8%)
Problem Statement
Insights required by different stakeholders required a high degree of customisation, and external stakeholders were unwilling to invest in data analytics personnel to carry out the required co-development of customised data analytics.
As a result, the standard reports created by our internal data analysts, though simplified, were a bit complex to interpret and not targeted to the stakeholders' needs. Also, the insights delivery time by our team took a long turnaround time which began to negatively affect our brand perspective as customers viewed us as unresponsive.
Proposed Solution
Update the current review analysis process to feature automated insight discovery from our extensive product review source data. The new process aims to minimize overall effort/time spent searching for valuable, actionable insights, thereby empowering customers to speedily make data-driven decisions.
My Process
Identifying Stakeholders and Understanding the Project Scope
Before kick starting the project, we needed to be certain of the problem and its proposed solution. As the BA on the project, I had to identify and collaborate with stakeholders to understand the current review analytics process and gather insights about the constraints of the current process and their expectations for this project.
Stakeholders
External: Manufacturing Lead, Sales Analysts
Internal: Customer Success Managers, Data Analyst
Eliciting and Analyzing Requirements
After identifying the stakeholders, we elicited requirements using interviews and survey forms. The purpose of elicitation was to determine the project scope, gather functional and non-functional requirements, and analyze requirements. These were useful in creating documents that ensured transparency and minimized conflicting requirements.
"Customers often ask us for quick answers about trends in their reviews, but it takes too long to get those insights. We need a faster way to deliver targeted insights without overloading them with unnecessary details."— Lauren, Customer Success
"The current process is labor-intensive. We spend too much time customizing reports, only to find that they still don’t fully meet the stakeholders' needs. Automation could help free up time for more strategic analysis."— James, Data Scientist
"We need insights that are specific to our product lines and highlight products/materials customers are complaining about the most—without us having to spend hours digging through reports."— Chi, Manufacturing
Documentation
I created a Business Case detailing the risks, alternatives, costs and benefits of the proposed project, and a flowchart that showed a high level overview of the current review analytics process and a FRD document. These were shared with stakeholders.
Placeholder
Project Planning
In preparation for the solution design, I planned the following for the project implementation.
Collaborated with the technical team to determine the scope of automation (e.g., data collection, sentiment analysis, dashboard creation).
Defined roles and responsibilities for the projects.
Established timeline, tools, and techniques to be used. E.g., data modelling, agile framework, and user stories.
Solution Design
I collaborated with the developers and data team to translate the requirements into a solution design. A key part of the solution design was to integrate our platform with a third party automation software. We weighed the pros and cons of creating an automation bot in-house and integrating with the current system VS purchasing the api to a third party automation software and integrating it with our platform. The latter option prove to be more beneficial and we embarked on the design of the system.
The current (as-is) state was composed of the following:
Customer portal - for standard review analytics reports.
Visual search API - for clients to connect to for visual search capability.
Visual search engine - the brains behind visual search, leveraging proprietary computer vision technology.
PostgreSQL database - for storing review analytics and visual search data.
The future (to-be) state should compose of the following:
Customer portal - for standard review analytics reports.
Visual search API - for clients to connect to for visual search capability.
Visual search engine - the brains behind visual search, leveraging proprietary computer vision technology.
Standardised Azure PostgreSQL database in Star Schema with extensive documentation.
Cimba.ai natural language chatbot built on the star schema to “talk” to the underlying data for quick natural language, table, chart and graph responses without the need for training in data analytics.
Solution Implementation & Testing
Some of my responsibilities during the implementation & testing phases include:
Facilitating communication between the technical team and non-technical stakeholders.
Managing new requirements and scope creep.
Setting up test cases and supporting user acceptance testing (UAT). You can find the test case documentation here.
Review of Solution
Performance Metrics from Improved State:
Quantitative Improvements:
Reduction in Insight Discovery Turnaround Time: Average turnaround time decreased from X hours to Y hours (e.g., a 50% reduction in time-to-insight).
Increased Customer Retention: Year-over-year retention improved by Z%, driven by enhanced usability and satisfaction with the analytics platform.
Cost Savings Realized: Potential savings of $X per new use case through actionable insights provided without additional data analytics personnel.
Dashboard Utilization: Increased usage of dashboards by external stakeholders by W%, reflecting better alignment with their needs.
Qualitative Improvements:
Enhanced User Experience: Stakeholders now interact with an intuitive, conversational interface, reducing the perceived complexity of analytics.
Improved Internal Brand Perception: New clients now view the product as less resource-intensive and more user-friendly, enhancing reputation and market positioning.
Scalability: The introduction of a natural language chatbot allows seamless scaling of insights delivery without requiring additional human resources.
Baseline Comparison:
Previous adoption rate for standard dashboards: A%
Improved adoption rate for natural language-driven insights: B%
Customer satisfaction scores (before vs. after implementation): C vs. D
These metrics reflect a significant shift toward a data-driven, user-focused approach to analytics, aligning better with the diverse needs of stakeholders in the fashion intelligence and manufacturing sectors.
Tools used in this project
Gap Analysis
Solution prototypes/demo
Decision framework