Business Intelligence

BUSINESS INTELLIGENCE OVERVIEW & APPLICATIONS

Definitions ƒ

The technology that enables companies to explore, analyze, and model large amounts of complex data. Consists of statistical modeling, data mining and multidimensional data exploration technologies (OLAP).

ƒ

Business Intelligence (BI) is normally built on one or more well defined data marts. BI models could also feed the data sources (DW) with metadata.

ƒ

Application of BI on customer data leads to Customer Intelligence or Customer Analytics

ƒ

Must be thought as an integrated process of every CRM solution (thus defining Analytical CRM)

ƒ

BI is based on several technologies & scientific areas such as information technology, OLAP, data mining, statistical modeling, text mining, visualization techniques

http://www.datamine.gr

BI, Overview of Applications Decision Making processes: well defined customer segments and KPIs can be analyzed versus time and in combination to several customer & product attributes: changes in customer base (quantitative and/or qualitative), revenue and profitability monitored in detail. What if scenarios with ROI models, on marketing, CRM or other actions. CRM & Customer analytics: Infrastructure that provides overall assessment on each single customer: profitability, loyalty, credit risk, usage patterns. Traffic information modeled and analyzed versus time along with customer profiles enable churn management, monitoring & prediction or Credit Risk assessment. Customer Loyalty: Business logic that encapsulates customer insight in order to build long lasting customer relationships: the right offer to the right customer through the right channel can help maintain high levels of Customer satisfaction. More accurate measurement of customer satisfaction is also possible through BI techniques (directly through targeted e-surveys or indirectly from QoS figures), regular stratified random sampling providing feedback to data warehouse or marketing database. Campaign Management: Customer selection, eligibility criteria, Target group definitions and profile analysis, campaign execution automation processes can be optimized using BI infrastructure. Marketing & Sales: Customer usage patterns, profiles and customer base trends may reveal significant cross-selling or up-selling opportunities. Marketing functions can greatly benefit from BI in designing new products, performing what if scenarios or closed group studies. Performance of marketing actions, special offers or campaigns can be assessed in details using customer responses and changes in usage patterns: The Closed Loop Marketing

http://www.datamine.gr

Business Intelligence, Architecture DATA PROVIDERS

Billing & Provisioning Systems Customer Profiling Account data

DATA WAREHOUSE - ANALYTICS

DSS AREA - DATA CONSUMERS

Data cleansing, Transformation to ‘flat’ data structures

Reporting Tools OLAP

DATA WAREHOUSE

Services & tariffs

Customer Base KPIs monitoring

Billing & payment history

Customer Care, Operational CRM

ETL processes

Contact History, Complaints, Activation Requests TRAFFIC processes Marketing Data Products & services properties, Campaigns, Micro& Macro segmentation schemes

Flattened customer data structures Reliable customer data with time dimension Physical Customer, Account & Contact, Customer Scores Billing data Payment behavior Segmentation data Utilization profile & Aggregate Traffic patterns

Customer Segmentation Manager

Customer Viewer CRM datamart

Operational CRM Platform Descriptive statistics, traffic patterns

Traffic Data Raw server log data-clickstream, Browsing patterns

REPORTING datamart

Statistical Modeling Sales Automations Statistical models, churn prediction, credit scoring, fraud cases, segment-clustercampaign memberships http://www.datamine.gr

Business Intelligence, Applications Data Modeling

Customer Metrics

CRM data mart

Customer Analytics

Front End Apps

Data understanding and basic statistical analysis (descriptive) • Identify critical entities, data quality problems, limitations & special cases • Collect & organize reliable information from every business function of the enterprise • Design data cleansing & validation procedures, flat data structures, synchronization mechanisms • physical customer Analyze data • build statistical models • design customer metrics, extend data structures, incorporate logic into synchronization mechanisms • customer base evolution & cluster analysis • profitability, consumer credit risk, churn rates, profiles, tenure, customer value, expected value Data structures, combining customer related entities and customer metrics scores • Also known as ‘Marketing Database’ • Flat data structures, containing hundreds of customer attributes versus time: behavioral data, demographics, payment & contact information for a series of time frames enabling modeling of change in customer data • Could be outside DW and/or Multidimensional Advanced models (statistical - data mining) on customer profiles & transactional data producing probabilities for certain events, homogenous clusters, behavioral patterns • Predictive components Integration with operational CRM system • Establish systematic data flows from/to CRM, Call center, corporate site or any other customer touch point • Development of additional front-end applications for management & decision making purposes. http://www.datamine.gr

Typical Customer Metrics Billing & Payment Statistics ƒ Total amount Billed – Amount due ƒ Billing Statistics (Averages,Variability) ƒ Statistical Trends on bill \ payment ƒ Credit Score (payment behavior) ƒ Profitability or Revenue Rank Score ƒ #accounts (by status), Product & Services, segments Customer Care or Sales Contact History ƒ Frequency of Calls - Requests ƒ Distribution by Service, Reason, Type, Priority Metadata ƒ Statistically derived Scores, clusters and segmentation schemes ƒ Marketing Research data, customer satisfaction surveys, on-line customer surveys, customer interaction data (CRM campaigns, Loyalty program memberships & usage, special offers) ƒ Micro-Macro segmentation, clustering memberships, control-placebo group memberships

Traffic analysis ƒ Email usage: IN vs OUT ƒ Email usage: distinct email addresses along with frequency analysis - dependency ƒ Email usage: metrics on volume of data, files, message size, file types transferred ƒ Certain Services usage patterns (for instance categories along with frequency analysis ƒ Distribution of sites – pages visited: Distinct Number, Frequency & average time spent) ƒ Distribution of daily traffic on predefined set of sites or categories (on-line shops, news, entertainment, educational, academic, scientific etc) ƒ Time of Day distribution ƒ Day of the week distribution ƒ Quality of Service indicators ƒ Bandwidth, daily, monthly, statistics, trends

http://www.datamine.gr

Typical Customer Dimensions & Measures ƒ ƒ ƒ ƒ ƒ ƒ ƒ ƒ ƒ ƒ ƒ

Systematic, normal, occasional user (based on frequency/duration of connections) Professional, academic, fun (based on sites visited-content preferred) Service sensitive, price sensitive Power, normal, entry level user Demographics, customer type (business-consumer, traffic based) Average Revenue Per month, expected yearly revenue VAS usage, mailbox or other services usage Tenure Use of e-commerce - On-line transactions Seasonality indexes Statistically derived clusters (homogenous groups pf customers)

In a GSM environment ƒ Calls versus SMS usage ƒ Incoming versus Outgoing Calls ƒ Small number of long duration calls versus large number of small duration calls ƒ MSISDN dependency index, based on weighted scores of distinct IN/OUT MSISDNS

http://www.datamine.gr

Additional Applications ƒ

Advanced OLAP reporting, definition of cubes that allow multidimensional views of customer data

ƒ

Monitor & analyze customer base evolution, monitor time series of critical KPIs, define customer base health indexes

ƒ

CRM program development, monitoring & optimization (closed loop marketing)

ƒ

Customer satisfaction measurement & monitoring

ƒ

Design & develop customer Loyalty procedures

ƒ

Development of micro & macro segmentation schemes

ƒ

Identify cross selling & up selling opportunities

ƒ

Define, Monitor & predict Churn

ƒ

Enhance campaign management procedures, measure effectiveness, calculate ROI

ƒ

Study bad payment behavior, develop credit scoring models

ƒ

Build optimized rules and policies for identifying fraudulent cases

ƒ

Flexible target group definitions, possibility for stratified random sampling for further customer Research – survey studies.

ƒ

Marketing activities (promotions, offers, campaigns) assessment capabilities

ƒ

Fraud detection http://www.datamine.gr

Additional Applications TARGET GROUP DEF

MARKETING USER PROFILE & STATISTICS

PERFORMANCE EVALUATION & ROI ANALYSIS

TARGET GROUP SPECS

CUSTOMER BASE SEGMENTATION & CAMPAIGN MANAGEMENT SYSTEM

CUSTOMER RESPONSE DATA

CRM DATAMART ANALYTICS & RESPONSE

CUSTOMER & CAMPAIGN DATA

CRM / CALL CENTER SYSTEM / EMAIL SERVER

CUSTOMER RESPONSE DATA

PROMOTIONOFFER

ƒ Closed loop marketing: gather

customer responses to DW ƒ Performance evaluation - ROI ƒ Control – Placebo group design

CUSTOMER

http://www.datamine.gr

Campaign management Define Campaign, special properties & Objectives

Analyze data, define target group

Release Campaign, manage Smooth execution

ƒ

Define business needs, objectives of the campaign

ƒ

Basic description of the target group, the time frame, the channels to be used,

Finalize campaign, analyze data, calculate ROI

the budget to be allocated, human resources needed or special I.T. infrastructure ƒ

Evaluation criteria: Definition of the criteria to be applied in order to assess campaign efficiency (such as expected responses, expected revenue or usage increase)

ƒ

Decide on use control & placebo procedures in order to measure campaign success (split the target group in to two random samples)

ƒ

Decide on campaign customer contact scripts, follow up procedures

http://www.datamine.gr

Campaign management Define Campaign, special properties & Objectives

ƒ

Analyze data, define target group

Release Campaign, manage Smooth execution

Finalize campaign, analyze data, calculate ROI

Descriptive analysis of data based on predefined metrics (such as ARPU, tenure, usage patterns & categorizations, cluster memberships, demographics)

ƒ

Analyze resulting target group versus other characteristics – metrics (not necessarily used as selection criteria) to verify the normality of the set of customers.

ƒ

Customer Eligibility checks: Apart from suitable customer profile (according a specific target group definition) a given customer must meet specific eligibility criteria before communicating the campaign (e.g. check of open balance or amount due or fraud indicators – for post paid services)

ƒ

Additional eligibility checks for recent customer requests, offers, contacts or other promotional activities (for exclusion or extra handling)

http://www.datamine.gr

Campaign management Define Campaign, special properties & Objectives

ƒ

Analyze data, define target group

Release Campaign, manage Smooth execution

Finalize campaign, analyze data, calculate ROI

Release the campaign, make eligible customer dataset available to call center (CTI) and/or email or other server depending on communication channels assigned to this campaign

ƒ

Monitor campaign execution progress on regular basis, collect overall progress data and make adjustments if necessary (from script enhancement to human recourses allocation)

ƒ

Monitor follow up actions for customers that have accepted the offer-proposal, ensure smooth completion of the process (POS, Corporate site, Customer Care).

http://www.datamine.gr

Campaign management Define Campaign, special properties & Objectives

Analyze data, define target group

Release Campaign, manage Smooth execution

Finalize campaign, analyze data, calculate ROI

ƒ

The closed loop marketing: Collect all resulting data to the initial data source, correlate and analyze campaign.

ƒ

Customer contact history within this campaign (whatever the response – if any) becomes part of the overall customer (contact) history record and available for future analysis, modeling & reporting.

ƒ

The same dataset is available as history of the specific campaign and of the customer base. The campaign execution event marks the evolution of the customer base (comparative reporting before and prior the campaign) for trends or pattern identification

ƒ

Compile ROI models, compare expected results with actual, analyze versus initial statistical profiles of the target group, interpret the results

http://www.datamine.gr

22 Ethnikis Antistasis Avenue, 15232 Chalandri, Athens, Greece [email protected] [email protected] http://www.datamine.gr

http://www.datamine.gr