Client
Domain: Display Products for Shops
Channel: E-Commerce Website
Objective
The primary goal was to develop a user-friendly navigation system that effectively fulfills search intent.
Challenges
Upon initial analysis, I discovered that navigating the website to find intended products was difficult due to poor organization. Users struggled because:
- The product categorization did not align with user expectations.
- Customers were accustomed to shopping on competitors’ websites, which had better navigation structures.
Process
To identify user pain points, I analyzed:
- Sales Data to recognize which products performed well.
- Google Analytics to track user behavior, bounce rates, and conversion rates.
- SEO Keywords to understand user search patterns.
- Competitor Research to benchmark best practices.
Findings
- There was a significant gap between user search queries and the way products were displayed.
- Sales data showed that easily discoverable products had higher sales.
- Google Analytics revealed low visit-to-cart conversions and small cart sizes, indicating missed opportunities for upselling related products.
Solution
To address these issues, I:
- Exported product data and used a programmatic approach to generate tags for each product.
- Mapped SEO keyword data to associate relevant keywords with each product.
- Used Python and AI to automate tag generation, generating a total of 182 tags, followed by manual review and proofreading.
- Categorized the tags into four major user search intent categories (Pillar Topics):
- Shop Type (e.g., Jewelry Store, Retail Store, Pharmacy)
- Placement Type (e.g., Countertop, Wall-mounted, Freestanding)
- Product Type (e.g., Display Showcase, Brochure Holder, Signage Stand)
- Material Type (e.g., Acrylic, Glass, Metal)
- Leveraged AI (Python Script + OpenAI API) to systematically segregate tags into these four major categories.
- Conducted an SEO expert review to finalize and refine the tags.
- Structured the website navigation based on these categories, implementing a mega dropdown list with 48 tags per Pillar Category. With 3000 products analyzed, I generated four major categories—Shop Type, Placement Type, Product Type, and Material Type. Each category contained around 48 tags, with products belonging to one or more tags from each Pillar Category, ensuring a more intuitive navigation experience.
Outcome
- The website navigation was revamped, improving user experience and search efficiency.
- Customers could now search intuitively, e.g., “I want to buy an Acrylic Display Showcase for my Jewelry Store to set up on a Countertop.”
- Acrylic (Material Type)
- Display Showcase (Product Type)
- Jewelry Store (Store Type)
- Countertop (Placement Type)
Final Thoughts
While AI (OpenAI API) could have been used exclusively for the entire categorization process, cost estimations were prohibitively high. Instead, I adopted a hybrid approach, using AI for logical decision-making while keeping execution costs at just 2% of an AI-only approach.
Important Note for Fellow Readers
- Avoid jumping straight into AI solutions without cost analysis.
- AI queries charge per input token, meaning even an incomplete process incurs costs.
- A programmatic approach combined with AI optimization can significantly reduce costs while improving accuracy and efficiency.
This case study highlights how Python and AI can be strategically leveraged to enhance UI/UX, improve navigation, and drive conversions on an e-commerce platform.
contactme@pradeepdabane.com for any suggestions or to know more how I do it.