Integration of Statistics & Business with the technology - The Biggest Breakthrough in the retail industry
Data Hygiene, Integration of the data across multiple channels is few of the challenges for the retailers
Statistics is useful in helping retailers to reduce their cost & increase their profits by using the customer data
IAC 2024 - IA Fast Track to Search Focused AI Solutions
How your favorite retailers make money out of analytics
1. How Your Favorite Retailers Make
Money From Analytics?
The Fifth Elephant, 28th July, 2012
Sridhar Bollam
Vice-President - Analytics
Capillary Technologies
2. New Era of Retail Analytics
The Biggest Breakthrough - Integration of Statistics and Business with the Technology
Real-time In-store Engagement using
Predictive Analytics
Helped retailer in increasing sales by 4%
Multi Channel Integration across
Websites, Social Platforms & Stores
Access to Quality Demographic data & Lifestage
Events
Customer is an Individual
Bayesian Prediction Models helped retailers in increasing the
sales by 4%
3. Retail Industry: Scope of Data
Retail accounts to 15% of Indian GDP, $470 billion
11 Shop outlets per 1000 people
$27 billion is organized sales
60% comes from verticals other than Food & Grocery
Demographic data, Transaction data, Inventory data & Marketing data
References: Wikipedia
4. Challenges in Analytics: Retail Industry
Data Hygiene – or lack of it !
Connecting the customer online, offline & social media
Playing with scale of data
Running real-time algorithms on cloud data
Deliver ROI in less than a quarter
Limited tools to analyze data
5. Business Problems in Retail
Typical Business Problem One of the Solution
Low campaign conversion rates (2 – Engaging customers: In-Store activities
3%) to boost conversion rate up to 20%
High communication cost & less Merging 15 logistic propensity models
relevance to act together
Need to engage each customer
80 – 85% Customer Drop Rate (Very
uniquely by understanding the
high compared to other industries)
customer (Prediction Algorithms)
6. Flexible Analytics
Deliver Quick Profits Through In-store Activities
Quick Tuning
Before During After
Marketing In-store Marketing
Activity Activity Activity
Close Loop
Prediction through Real-time Engagement to Hypothesis Validation
Logistic, Discriminant & understand missing baskets Text Mining Techniques
Bayesian models Instant results
7. Real-time Recommendation Engine
An In-Store Activity
Analyze the
Recommendation
basket based
Engine
on SKUs
Shopping
Basket
What is What really
Customer Past Business
missing in goes with
Purchase Data Objective
Basket? What?
Business Real – Time Instant
Impact Execution Suggestions
9. Working of Real-Time Engine: Bayes Probability
Layer 1 Layer 2 Layer 3
γ
B C
β
C D
A Basket Probability:
D E
P(C|A,B) = P(A,B,C) / P(A)P(B|A)
E
B
N Layer Basket: P(x1,x2,x3,…,xn)
C P(xn|x1,…,xn-1) =
P(x1)P(x2|x1)P(x3|x1,x2)…
Can execute several complex rules in parallel with any operational effort
Typically retailers increase sales up to 4% using this recommendation engine
Past purchase behavior is applied over the output as a (selection or Rejection rule)of Real-Time Engine
10. How Prediction Algorithms can increase profits?
-Departmental Stores
Which Customers to Target With?
Big &
Complex
Data 15 Predictive Merging
Propensity Models Propensity
DOE
+ using Logistic & Models to find
Marketing Discriminant the best offer
Offers
Cost: $20,000
Cost: $20,000
Targeted: 20,000
Targeted: 20,000
20%
4% Predictive conversion
Conversion Random
Approach Approach
Profit: $100,000
Responded: 4000
Net Profit: 4,000
Responded: 800
11. Data Mining Techniques in Pizza Business
Pizza Loyalty Program is More Driven by Groups Than an Individual
Customer Family
Loyalty Loyalty
Program Program
Customer
Regex Phonetic
Address
Merging
Different
Edit Distance
Customers into
a Household
20% of the customers seem to use different mobile number but same address
12. Who is Your Customer?
Taste Behavioral
Clustering Clustering Differentiated on Customer Spend Parameters
Golden Pool
High
Spenders
Discount
Seekers
Potential
A1 B1 C1 D1
Fine Diners
Big Bite
Burpers +
Solo Drill down further using Significant Variables
Connoisseurs from Chi-square test
Gourmet
Travelers
Arrived @ 17K segments with 7 Dimensions
13. Linear Regression technique - Pizza Business
......
????
T1 T2 T3 T(Last) T (Next
order)
T(Next Order) is dependent on # Customer Visits, last
purchase, Micro-segment, Customer Frequency using Linear
Regression models for groups
Continuous engagement for 17K segments
Instant prediction for the next purchase
Taking care of all external events like Cricket by combining it with Propensity
Model
14. How We Grew Retailers’ Business?
4K What? Whom? When?
Conversion Rate = 3.4%
YOY Growth = 6.4 %
What to target? to $1 Investment has given ROI
whom? of $20
Customer Lifetime Value
Treating all Conversion rate = 1.8%
customers YOY Growth = 2.78%
Similarly
$1 Investment has given
Conversion rate = ROI of $9 Do it the Analytics Way!
1.3%
YOY Growth =
0.56%
$1 Investment has
given ROI of $1.8
Non – Analytics
Driven Marketing
Initiatives
2K No Data
Acquire Engage Retain Grow
*Customer lifetime value is proportional to Retail business growth