We deliver measurable results
Get to know the client
The client is one of the world’s largest suppliers of athletic shoes and apparels. It’s among the most valuable sports businesses and ranked among the Forbes Global 2000 companies in 2024.
Challenges
Targeted solutions for
Low Conversion Rates
Despite attracting a significant number of visitors, the client’s website faces low conversion rates, with major drop-offs occurring at critical stages of the purchase journey.
Key pain points to be fixed:
– High Drop-off at Product Detail Page (PDP): Over 54.34% of users exit the site after viewing a PDP. The lack of engaging or optimized content – such as “Shop the Look,” elevated product storytelling, or tailored recommendations – limits the Add-to-Cart (ATC) rate.
– Friction at Add-to-Cart and Checkout Initiation: A substantial portion of users abandon the journey after adding items to their cart. Time-consuming address inputs and a lack of pre-filled or auto-complete fields lead to drop-offs, particularly at the billing step. However, users interacting with membership prompts (“Join” or “Login”) show a 20% increase in checkout completion rates.
– Checkout Page Abandonment: A lengthy and complex checkout process with up to 20 input fields spread across 5 pages, combined with limited payment options by region, further discourages users from completing their purchases.
Inefficient inventory operation
The client struggled to maintain optimal inventory levels due to the limitations of traditional demand forecasting methods. This led to two either stockouts, which resulted in lost sales opportunities and customer dissatisfaction or excess inventory, which tied up capital, increased storage costs, and risked product obsolescence.
To resolve this, SmartOSC has to tackle two core issues:
– Lack of Data Readiness for Machine Learning: While the client had a well-organized data warehouse for storing historical transactions, the data lacked the necessary granularity and attribute diversity required for machine learning models to generate accurate, actionable insights.
– Inaccurate Demand Forecasting: The forecasting approach was heavily reliant on past sales trends, failing to incorporate external variables such as seasonality, promotional activities, or market shifts. As a result, the client had an incomplete and often misleading picture of future demand.
Better customer segmentation
With a large and diverse customer base, the client aims to personalize the shopping experience and improve marketing effectiveness by segmenting customers based on their preferences, behaviors, and purchasing patterns.
Several challenges hindered this effort:
– Lack of Data Integrity: Customer-provided information such as age, gender, and location, was often inaccurate due to minimal input validation. On top of that, data privacy restrictions limited access to certain individual-level insights, making it difficult to build reliable customer profiles.
– Bot Prevalence: A growing number of users employed bots to quickly purchase new products for resale. This artificially inflated transaction data and skewed behavioral insights, undermining the integrity of customer segmentation models.
– High Segmentation Costs: Marketing communications, including emails and push notifications, incurred real costs. Without accurate segmentation, the client risked inefficient targeting, leading to unnecessary marketing spend and reduced profit margins.
SOLUTIONS
Our tools for success
Optimizing Conversion Funnel
SmartOSC worked with the client to analyze and enhance every stage of the customer journey to improve engagement and reduce drop-off:
– Product Detail Page Enhancements: Enrich product descriptions with clear feature highlights and persuasive CTAs like “Get Yours Today.” Introduce “Shop the Full Look” modules to increase cross-sell potential and boost average order value.
– Streamlined Checkout & Membership Engagement: Simplify form fields (combine “First Name” and “Last Name”), restructure form flow, and incorporate targeted incentives (limited-time offers, exclusive member discounts) to drive urgency and lift conversion rates.
– Expanded Payment Options: Collaborate with development teams to integrate regionally relevant payment gateways (PayPal, UnionPay, Visa), reducing friction for international customers.
Enabling Inventory Forecast with Machine Learning
We implement advanced ML solutions to enable data-driven inventory management:
– Feature Engineering for Demand Signals: Integrate seasonality, promotional events, and campaign performance into model inputs to better reflect real-world demand fluctuations.
– Robust ML Pipeline: Build a scalable and modular ML pipeline encompassing preprocessing, training, packaging, and continuous monitoring to ensure consistent accuracy and model lifecycle efficiency.
– Model Performance Optimization: Apply and compare various forecasting models (Regression, ARIMA, STL, LSTM), using results to directly inform inventory planning across BAU and promotional periods.
Implementing Customer Segmentation & Data Governance
We ensure segmentation strategies are based on accurate, secure, and actionable data:
– High-Fidelity Data Sourcing: Detect and correct anomalies across customer profiles, cross-referencing with transactional data to ensure validity
– Data Privacy & Compliance: Apply data masking techniques in collaboration with engineering teams to comply with privacy policies and governance standards
– BOT Detection & Cleansing: Implement automated detection of bot traffic to preserve the integrity of behavioral insights
– Dynamic Segmentation Models: Continuously test and refine clustering methods (RFM, k-means, hierarchical, OPTICS) to ensure segmentation evolves with user behavior and business needs
