Download - Ponència d'Érik Brieva a la Festibity 2015
© Strands Inc. 2015
STRANDS RETAIL
© Strands Inc. 2015
Tr i l l i ons o f Cus tomer & P roduc t “Even ts ” Happen Every Day in On l ine Re ta i l
Most of these Customer & Product Events Contain Hidden Patterns and Relationships
© Strands Inc. 2015
Processing Millions of Customer & Product Events Every Day
Home page view
Category page view
Product page view
Multiple Product Views Product Clicks
Add-to-Cart Events
Cart Abandonment
Events
Multiple Item Cart Additions Purchases Product
Bounces
Products viewed next Product price Brands Viewed
Most Often Email Opens Total Cart Size per Customer
STRANDS Analyzes and Understands every “Event” in Retail…
© Strands Inc. 2015
And from this Big Data, STRANDS Detects Relationships & Patterns Between Customers and Products
STRANDS Understands The Big Data of People & Products
© Strands Inc. 2015
• When a customer looks at Product A, they also tend to show interest in
Products B, C, and D
• Customers who spent €25- €50 on their last visit will tend to spend €80 - €100 on their next visit
• For a product @ €500, conversion is 2%, but for a similar product @ €549, conversion is 3% (= 65% increase in revenue/customer)
Examples of Big Data patterns STRANDS produces and takes action on every day
© Strands Inc. 2015
STRANDS SECRET SAUCE:
OUR B IG DATA ALGORITHMS ENABLE US TO…
SHOW THE RIGHT PRODUCT
TO THE RIGHT CUSTOMER
WITH THE RIGHT PRICE
AT THE RIGHT TIME
STRANDS Increases Conversion 4X and Revenue 20% Average Performance from Actual Strands Customers Report
© Strands Inc. 2015
Data Taken from Actual Strands Client Performance Report
Case Study: e-market of natural products
© Strands Inc. 2015
Case Study: zooming windows
© Strands Inc. 2015
Case Study: e-commerce of consumer electronics goods
© Strands Inc. 2015
Case Study: supermarket online shop
© Strands Inc. 2015
A/B test to compare the logics for a home page widget:
A: widget logic based on returning the most visited items.
B: widget logic based on STRANDS user-to-item personalization algorithm.
The results show that STRANDS user-to-item personalization algorithm provides better results in clickthrough than a bare popularity-based algorithm.
A/B Testing Example 1
© Strands Inc. 2015
A/B test to compare the logics for a product page widget:
A: widget showing greenlisted products. That is, a marketing manager at Giggle has provided a list of handpicked products to recommend for each product page.
B: widget logic based on STRANDS item-to-item algorithm. The algorithm has been configured for cross-selling purposes.
The results show that STRANDS item-to-item personalization algorithm provides better results in click-through than a list of items handpicked by a marketing manager for each case.
A/B Testing Example 2
© Strands Inc. 2015