optimization as a golden layer - chris diener, svp analytics, absolutdata
DESCRIPTION
Chris Diener, SVP - Analytics, AbsolutData delivered a session in MRA insights and strategies conference, 2013, on the topic ‘Optimization as a golden layer’, where he discussed optimization and constrained optimization and then showed how it can be applied effectively across a number of common and emerging MR technologies. AbsolutData is a global leader in applying analytics to drive sales and increase profits for its customers. AbsolutData has built strong expertise and traction with Fortune 1000 companies across 40 countries. We specialize in big data, high end business analytics, predictive modeling, research, reporting, social media analytics and data management services. AbsolutData delivers world class analytics solutions by combining their expertise in industry domains, analytical techniques and sophisticated tools. Visit us here : www.absolutdata.comTRANSCRIPT
Optimizationas a golden “layer”
Presenter: Chris Diener, SVP – Analytics, AbsolutData
MRA Insights and Strategies ConferenceJune 10-12, 2013
Venue:
Vision
Mission
To become the
Most Impactful, Most Respected,
Most Powerful
Analytics Firm that has ever existed
To help
forward looking organizations excel
Through optimal use of data
Recommendation Systems
Constrained Optimization
Linear Programming
Machine Learning
Genetic Algorithms
These are the tools of the 21st Century
Ever Heard of This?
Are You
Optimizing?
MR Industry Conditions – Tough!
Client perspective
Order takers instead of consultants
Do the work instead of liaising and thought partnering
Questions of relative contribution compared to “Analytics” folks using non-survey data
Vendor perspective
Vendor perspectiveCommoditized: lack of true differentiation of services
Self-help explosion
Budget squeeze between economy and “Analytics/Social” spend
How do you differentiate?
How do you win?
Problem: An Insight is not an Insight, is not an Insight
RESEARCH
MUST
MUST NOT
Impact Decision Making
Remove specific pain or solve a specific problem
Merely Describe
Just address issues or answer questions
What do they need
to do?
Virtually all research falls
short
Recommend! (make a stand, be an expert, be specific)
Dig! (identify specific decisions
and actions)
How do you do this?
Support! (with a story)
Ensures Recommendation
Empowers Recommendation
Optimization
Make you a more valuable resource
Promotes Specificity
Bringing a System to its Peak Performance
Need constraints Need a goal
Need a system – or a Model
Searching and recommending
Need levers to pull to get to that goal
DATA
Identification of critical outcomes and management levers
Building a model to find the predictive
relationships between levers and
outcomes
RECOMMENDATIONS
SIMULATOR ENGINE
Potential constraints
Conjoint Product and Product
Line Management
Segmentation For communications
or product development
Targeting Database or
Demographic Classification
PositioningBrand Image Management
Examples of Optimization
How Optimization Ensures Differentiating Quality of
Work
Segment prioritization
Brand positioning
Conjoint
Descriptive
Segmentation
Fashion Focused – 16%
Young Passionate about
fashion Keep up with the latest
trends Own relatively higher
number of jeans Willing to pay more
money for her jeans
Bargain Hunters – 28%
Interested and involved in this category
Love to shop but on a budget
Actively looks around for deals and sales
She own an average number of jeans in spite of spending less than average on jeans
Classic – 26%
Middle aged working women
Though she is not very style focused, she doesn’t want to look out -dated
Wants classic & latest trends to suit her age
More likely to buy specialty store brands such as Gap & Old Navy
Utilitarian Buyers – 20%
Lower income, white More inclined
towards functionality than style
Buy jeans only when necessary
Budget conscious Shops at discount
stores Buys Lee Jeans
Urban Dresser – 6%
African American and Hispanic
Want to look hip and trendy
Willing to pay more Large share of
purchases in urban brands
However more likely to buy jeans on sale and in off-price stores
16
Prescriptive
Segmentation
S6
S1
S3
S5
S2
S4
S12
S8
S11
S7
S10
S9Ability to Win
Attractiveness
Higher
Lower
Lower Higher
Segment Prioritization Matrix: Growing Brand Share Go after Segment S12 first and then consider segment S4 as a secondary target
Conjoint
PrescriptiveDescriptive
2%
4%
8%
13%
18%
24%
Margin Impact
1%
2%
3%
4%
6%
9%
Sensitivity= -1.61
Price product at $19K and offer a 5-year Warranty
PrescriptiveDescriptive
Quick dry
Fresh all day
Dry all day
Skin Comfortable
Refreshing
Suits Sensitive Skin
Confidence Giving
No Stickiness
Effective against Odour
Low Cost
Low Cost
No Stickiness
Confidence Giving
Suits Sensitive Skin
Refreshing
Comfortable on skin
Dry all day
Fresh all day
Quick dry
Brand positioning
Spend $50M to:• Increase Comfortable by 2 pts• Increase Confidence Giving by 5 pts
Finding the best product to make
Finding the first three product changes to make
Knowing who to target first and with what message
Knowing how much to invest in raising awareness of which specific brand qualities
Key is identifying a Goal that is Actionable
Frame your research in terms of specific decisions
Identify critical decision criteria that can be the basis of a maximization goal
Define the relationships between levers and the goal
Using the model, search for lever combinations that get the best goal outcome
How to do it
Recommend the best lever combination(s)
Optimizing Using Social Media
22
Using Social Media crawlers such as Buzzmetris, Radian 6, Alterian to collate unstructured user generated data from Social Networks, Blogs, Forums etc
Use Text Mining software's or custom verbatim coding methods to create quantitative data for reporting and modeling
Mining Social Media Data
Text Mining
Brand Tracker Report
Level 1 ModelSentiment Drivers
Level 2 ModelBrand Metric Drivers
Report Social Media Brand Metrics such as Brand Sentiment Score, Awareness / Share of Voice (SoV) along with visuals/frequency for Themes/Topics, Imagery, Brand Attributes, Product Attributes, Source etc
Multivariate Regression model to identify key drivers of the Brand Sentiment in the Social Media comment at Author (Respondent) Level as DV and Theme/Topic, Imagery, Persona, Brand/Product Attributes, Source etc as Independent Variables
Multivariate Regression model to identify key drivers of the Brand Performance Metrics (Brand Equity Score from Traditional Research, Sales/Volumes) aggregated at Week/Month Level using week level SM Metrics - Brand Sentiment/Satisfaction, Brand/Competitor SoV, Imagery, Personas etc
1
2
Details on the Driver Models
23
Reporting and Sentiment Driver Analysis Brand Key Driver Analysis
with key drivers of Consumer Sentiment drivers of Key Brand Metrics
Weekly dashboard withSM Frequency at category level• Brand Share of Voice/ Awareness• Competitor Share of Voice• Sources of conversation• Brand Sentiment Score (BSS)• Competitor Sentiment Score• Conversation theme cloud• Imagery (Existing and Emergent)• Product/Brand Attributes• Key Influencers
Drivers Analysis using Multivariate Regression
BSS = f( Imagery, Product/Brand Attributes, Benefits , Personas, other themes)
Aggregated Metrics from various sources (Weekly/Monthly buckets)• Weekly/Monthly Sales or volumes• Brand Equity Score (BES), Purchase Funnel Metrics from Traditional Research• Macro Economic Factors (US Gov, BLS)• Traditional Media Spend• % Share of Voice (SoV) - Brand, Competitor• Brand and Competitor BSS• Number or % mentions for imagery, Attributes, key themes• SoV by Personas etc
KDA using Multivariate Regression
Sales/Volumes or BES= f(Macro Factors, SoV variables, Mentions or Share of imagery/Attributes/ themes)
1 2
Optimizing product, people and message in the same system
Advanced Optimization Applications
Inclusion of social media or other big data with survey data for better application and insight
Segmentation
Targeting optimization
Brand purchase driver modeling
Marketing effectiveness and attribution modeling
Path to purchase modeling
CRM lifetime value calculations
Promotion optimization in direct marketing
So/Lo/Mo models for better targeting
If you need help with Analytics or Research, please write to us:
[email protected]@[email protected]
For Media related queries [email protected]
For all other queries [email protected]
25
HEAD OFFICE
314 Marble Arch Tower,55 Bryanston Street,London W1H 7AA
Phone : + 44 207 868 2240
UK OFFICE
DLF Cyber City SEZ,Building#14, 4th Floor, Tower B,DLF Phase-III, Sector 24 & 25A, Gurgaon-122002,
Phone: +91.124.4953.400
INDIA OFFICE
AbsolutData AnalyticsMiddle East JLTOffice 1604, Tower BB1Mazaya Business AvenueJumeirah Lake Towers
Phone: +97150-1577257
DUBAI OFFICE
1851 Harbor Bay Parkway,Suite 125, Alameda,California, USA – 94502
Phone: +1 510 748 9922Fax: +1 510 217 2387
What does your research look like?
Thank You
Where is your GOLDEN LAYER?