Advanced digital analytics techniques that blend offline and online data enabe e-commerce players to best target consumers. Weblog data and clickstream are powerful web analytics tools, but they must be merged with offline data analysis - such as loyalty cards, demographics, third-party information and sales data - to deliver maximum power for e-marketers.
The Work Ahead in Insurance: Vying for Digital Supremacy
Making Online Shopping Smarter with Advanced Analytics
1. • Cognizant 20-20 Insights
Making Online Shopping Smarter
with Advanced Analytics
Merging offline and online data can help retailers to customize targeting,
resulting in incremental increases in e-commerce revenues.
Executive Summary of a prospect’s willingness to buy, transactional
data points such as offline sales are key indicators
Online shopping has emerged as one of the most
of actual sales. This paper provides insights on
popular Internet activities, providing a variety of
how merging offline and online data can support
products for consumers and a multiplicity of sales
customized targeting, resulting in incremental
challenges for e-commerce players. Research
increases in e-commerce revenues.
suggests that online sales has grown 16.1% year-
over-year from March 2010 through March 2011. 1 Constraints and Opportunities
Both brick and mortar and pure play online In most companies, unfortunately, online and
retailers are competing for the attention of offline sales are processed in technical silos.
the online consumer. A real opportunity for Offline shopping data is very important because
companies lies in applying advanced digital many people still want to touch and see products
analytic techniques that integrate offline and first-hand before they buy. Conventional wisdom
online data sets to optimize on-site store inven- suggests that about 70% of all consumers will
tories. One example from retail is how Best Buy make an offline purchase as a result of online
is leveraging data to become more customer- marketing. Thus, offline data is extremely
centric. When Best Buy determined that 7% of beneficial, and mining it along with clickstream
its customers were responsible for 43% of its data provides a valuable resource for improving
sales, the company segmented its customers customer satisfaction, product development,
into several archetypes and redesigned stores sales forecasting, merchandising and visibility
to address the buying habits of these particular into the profit margin for goods and services.
customer segments, thereby enhancing the
The ability to identify and reach consumers at the
in-store experience and increasing same-store
most crucial point in their buying decision-making
sales by 8.4%.2
process, as reinforced by Google’s “Zero Moment
While optimization and other analytical of Truth”3 study, demonstrated that in-store or
techniques like A/B/multivariate testing, visitor online transactions are heavily influenced by
engagement, behavioral targeting and audience insights gained in the moment before a purchase
segmentation can point toward a high likelihood decision is made. Thus, steering consumers
cognizant 20-20 insights | june 2012
2. Applying Advanced Analytics
Data
Association
ETL Profiling Rules
Logistic Variables Statistical
Regression Creation Control Process
Weighted Segmentation Statistical
Average Control Process
Classification Linear
Time Series Regression
Advanced Analytics
Output
Figure 1
toward a particular product or service and then Joining geographic, creative, time of day (e.g.,
capturing them at the point of sale is a very morning, afternoon, evening, etc.) data and
powerful solution. mapping these attributes against time and
location-based sales data can highlight hidden
This is why Google’s search
An e-commerce player marketing is so appealing
interactions between online and offline sales
activity. This means e-commerce players can
that can capture the to marketers; consumers’ develop more effective multichannel strategies.
consumer’s attention desires are instantly tele- The end result is that real additional sales can
graphed by their actions.
and activity further be increased. Of course, e-commerce companies
have an option to use advanced Web analytics to
up the funnel and Creating a holistic customer
create a specific segment for visitors who arrived
picture is not a new idea but
combine it with very few companies have
online via offline campaigns. Web analytic4 tools
advanced analytics achieved it. Constraints
provide similar solutions, but they are not by any
means a complete set of offerings nor do they
techniques can begin have included database
create a single view of your customer. Organiza-
and hardware technology
to assume the mantle that could not effectively
tions need to pull all the data attributes, offline
of an analytics leader. support the solution; data
and online, into a single database, which would be
further refined by advanced analytics techniques,
management issues; Web
and use the unified data for precision targeting as
analytic systems that are not focused on the avail-
depicted in Figure 1.
ability of the underlying data; not getting the right
people to focus on the data; the division of online Figure 2 illustrates that by combining Website
marketing and traditional marketing skill sets; etc. data (i.e., clickstream information, etc.) with
loyalty card insights, sales data and third-party
Leading the Charge information, e-commerce players can gain critical
For e-commerce companies, the challenge is understanding into customer behavior. To reach
that customers are often relatively far into the the forefront of the data-driven digital revolution,
purchase funnel when they arrive at a particular e-commerce players need to acquire key tools,
site. An e-commerce player that can capture the analytic prowess and necessary skills. Not only
consumer’s attention and activity further up the do e-commerce players need to invest in the right
funnel and combine it with advanced analytics advanced analytics tools, but they must also gain
techniques can begin to assume the mantle of an access to mathematicians and statisticians to
analytics leader. create models and interpret findings.
cognizant 20-20 insights 2
3. Blending Multiple Sources of Sales Data for Optimal Positioning
1 2 3
Online + Offline We Reach — Right Customer at Right
Completes the Jigsaw Puzzle
Customer Behavior Time with Right Products
Enable Complete View Integrate Promotion, Web Visit, Sales Leverage Digital Analytics to
of Purchase Cycle Transaction, Customer Data Gain Customer Insights
ONLINE
Predict the
Understand Customer “Right Time to Sell”
Privacy
k Buying Choices
Quality Feedbac
g
Demo Addre
ss Multi
Channel Predictive
Segmentation
Score Modeling
Model g Optimization
Catalo
y
egistr t Accoun
t
Gift R Paymen
Social tion Market Basket Factor Probability
Media Promo Item Analytics Analysis to Buy
Sales on
Locati
Store Locator Probability
Analytics ROI to Drop Off
isit
bV
We
Assessing Impact of
OFFLINE Online Over Offline
Figure 2
Organizations need to apply more complex data Advanced Analytics Framework
calculations to generate a more accurate picture in Action: E-Commerce Gold
of customer behavior and transactional activity.
While e-commerce sales will continue to grow
Measuring offline marketing campaigns statistics
over the next four years, eMarketer5 estimates
is not as easy as many people think. There are
that the rate of growth will decline steadily. This
numerous different tools that can help organiza-
makes it even more vital for e-commerce players
tions improve data accuracy; however, many are
to be strategic about how they use clickstream or
third-party solutions that are not integrated into
Web log data to reach consumers.
a Web analytics solution.
The Art and Science of Connecting Offline and Online Data
Data Preparation
Segmentation Profiling
Campaign Data
1.0
Web Visit Data 0.8
Raw Analysis 0.6
Demographic Data Data 0.4
0.2
0.0
Sales Transaction Data
Predictive
Finalize models by taking Lift Chart*
Rank Ordering Modeling
inputs from business teams
Data
High Score Model
Medium Score
Business
Low Score Input Output
Optimize
Targeting
Figure 3
cognizant 20-20 insights 3
4. Benefits of Merging Online and Offline Data: Scenario One
Situation: Client has huge online and offline data which needed integration for actionable analysis.
Solution : Profile customers based on their online purchasing/browsing behavior with offline transactions.
Visit % of Visits per
Seasonality
freq. Visitors Visitor 1. Profiling helped client
High Regular 15.98% 44
Seasonal 1.04% 26
in recognizing the
Buyer
Low Regular 0.10% 5 potential buyers and
Online Data
Visitor Profiling High
Seasonal
Regular
4.59%
1.71%
4
38
up sell seasonal buyers.
Web Visit Non Seasonal 0.11% 26
Buyer Low Regular 1.45% 10
Web Seasonal 1.03% 8
Registration Channel Preference 2. Channel preference
% Customer
eCom Retail Both helped client in
Online
Large Frequent Buyer 5.0% 2.0% 2.7%
Transactions reaching out to
Channel Preference Large Buyer 11.7% 0.8% 0.6%
customers with their
Purchase
Behavior
Frequent Buyer 5.4% 3.5% 6.5%
Email preferred channel.
Medium Buyer 30.7% 2.0% 3.0%
Campaign
Small Frequent Buyer 0.5% 0.4% 0.8%
Banner Ads Small Buyer 21.0% 1.5% 1.9%
Design Preference % of 3. Taste of a buyer
% of Sales
Offine Data Purchasing Segment Customers
helped client in
Preference Modern 40.82% 43.75%
reaching them with
Sales Utilitarian 17.55% 11.10%
Transaction Organic 10.76% 5.19% targeted offers.
Classic 8.37% 11.04%
Demography
% Identified Sessions % of Identified visitors
4. Visitor and session
Store Location Visitor and Concept Dec-10 Jan-11 Dec-10 Jan-11
recognization was
Session Brand A 16% 15% 8% 6%
improved by 30%
Address
Recognization Brand B 22% 19% 11% 9%
to 50%.
Brand C 17% 13% 8% 6%
All Brands 16% 14% 7.8% 6.3%
Figure 4
Figure 3 illustrates a typical process/methodology In the near future, organizations will likely deploy
followed by a digital analytics center of excellence more sophisticated and integrated Web/digital
(CoE) to merge offline and online data in order to analytics tools to more easily combine online
build predictive models that optimize customer campaign data with offline insights. This will
targeting. include:
We have applied this methodology at several of • Getting the Web log or clickstream data from
our clients resulting in the scenarios presented in Web analytics tools.
Figures 4 and 5. • Merging it with point-of-sale data.
Benefits of Studying Online Customer Behavior: Scenario Two
Situation: : Unstructured huge clickstream data with not much knowledge of what to do with it.
Solution : Web path analysis revealed browsers intent and needs which helped to customize offerings.
Purchased with web Visit
30% people are using Express Checkout
100%
% of Unique Customers
60
18 70%
Thousands
71%
Online Data Express Checkout 50 80% 62% 61%
65%
Web Analysis 40
14
60%
48%
Browser
Web Visit 30
42 40%
Log
Web
20
33
20%
10
Registration 0%
0
Web Path Retail Catalog eCommerce
Online Online Purchasing
Category 1 Category 2
Transaction Channel
CHECKOUT SIGN IN
Transactions After E-mail EXPRESS CHECKOUT SIGN IN 30 Days prior 90 Days prior
Web Page
Email Click-through
View
Campaign 1. Checkout analysis helped client 2. Customer behavior of browsing
in estimating the return on new before buying helped client in
Banner Ads development on products Web site. focusing more on information
Web on their website about products.
Browser Store
Registration
Offine Data Distance Store Area Distance from browsers % Activities
Activities # of Visits after onsite
Sales Analysis search
Location % Customers with % Order Value
Transaction Name in 50 Miles with in 50 Miles Total Product Search 700,000
Item View Exit after Search 150,000 22.5%
Demography NULL Search 7,000 1.0%
Grove 96% 88%
Product Search 260,000 39.3%
Store Location
Onsite Onsite Search Westchester 96% 82%
Using Search Again 160,000 24.1%
search
Address
e Effectiveness Whitman 95% 79% Top or Left
50,000 7.1%
Navigations
Homepage 20,000 2.4%
3. Browser behavior around a
store helped client in providing 4. Onsite search effectiveness helped
customized offer on specific client in customizing product
stores. information for better search results.
Figure 5
cognizant 20-20 insights 4
5. Generating Collective Intelligence have created a unique solution framework (see
Figure 6).
Marketplace (Online + Offline)
Conclusion
Sales, Web, Demography, Store Location, E-mail,
Banner Mobile, Social Media, Data, etc.
The solution framework shown in Figure 6 can be
customized to meet the unique requirements of
Analytical Tools
each company. Because it is technology agnostic,
Omniture, Webtrends, SAS, SQL, it can seamlessly integrate with any ERP system,
Coremetrics, etc.
is
the Web, social media, DW/BI, CRM and POS
Filtering
Synthes
Group of Experts
Leaders, Scientists systems. Also, it takes data in whatever form is
and Experience Facilitators given and then it works toward synthesizing each
Validation and
of the data sets in a format that can be further
Application sliced and diced using our proprietary algorithms
and models. It thereby brings offline and online
data together to enable business users to make
Highly Specialized
Information Silos
insightful decisions to move their businesses
forward.
Collective Intelligence
New Technologies, New Knowledge, For example, an organization selling laptops or
New Philosophy
Insight Builder Tool mobile devices that captures the correct 1% of
its prospects can generate an additional 3% in
increased revenues. It can do so by applying a
Figure 6
tried and true process/methodology (as illus-
trated in Figure 3), followed by a digital analytics
• Studying online and offline customer behavior. center of excellence to merge offline and online
• Refining that data with the use of advanced data and build predictive models to optimize
analytic techniques, set methodology and customer targeting.
algorithms, applied by the right domain
Thus, identifying the most valuable customers
experts.
using online and offline data allows companies to
• Then, validating/refining the data for pinpoint more accurately identify those who are transac-
targeting. tion minded and can help boost their revenues.
A data-driven optimized strategy is the key, and
The end result: Enabling companies to reach the
that is possible only with offline and online data
right customer with the right product offerings
integration.
at the right time and place. With this in mind, we
Footnotes
1
http://www.ipaydna.biz/Online-sales-witness-16.1-percent-growth-year-over-year-n-43.htm
2
http://www.fico.com/en/FIResourcesLibrary/Best_Buy_Success_2271CS_EN.pdf
3
http://www.thinkwithgoogle.com/insights/library/studies/the-zero-moment-of-truth-macro-study/
4
http://en.wikipedia.org/wiki/Web_analytics
5
http://eMarketer
About the Author
Ashish Saxena is the Leader of the Digital Analytics CoE within Cognizant’s Enterprise Analytics
Practice. He has over 15 years of experience in e-commerce, multichannel retail, digital and advanced
analytics space. He also holds several patents in the e-commerce/mobile space. Ashish can be reached at
Ashish.Saxena3@cognizant.com.
cognizant 20-20 insights 5