It Final - Part 2 (Data Mgt.)

Front 1

billy beane method

Back 1

rethink normal baseball
instead find inefficiencies and exploit them -- scientific data driven approach to key decisions

analyze data differently -- better models of performance = better aquisition
- player acquisition
- using existing players

Front 2

method of improvement for white sox case

Back 2

goal = win more games w/ less $$

ex: more runs = better hitting vs. batting average
hold opponents to few runs scored

player aquisition -- work w/ undervalued assets

Front 3

emergence of mgt.

Back 3

prior to 1900 - no pro. mgt.

today: performance of important institutions depends on pro. mgt.

Front 4

mgt. 1.0

Back 4

mgr. as gunslinger

shoot from the hip/ gut feeling

performance depends on scarce mgt.

strict hierarchy

Front 5

mgt. 2.0

Back 5

face based decisions

DDS: decision support system (database, decision models, decision-marker insight, better tech.)
-- evolution: better data, better models, better tech., pushed lower in company

less reliance on superstar mgt.
DDS allows ordinary people to perform at extraordinary levels

what we think vs. what we know

competing on analytics

Front 6

competing on analytics

Back 6

sophisticated IS and analysis across all company functions

PRIMARY FOCUS = development and maintenance of analytics

FACT BASED DECISIONS IMPORTANT PART OF COMPANY CULTURE

hire people w/ best analytical skills

prop. metrics for eval. business processes

share data and analysis with customers and suppliers

TEST AND LEARN CULTURE: conduct experiments

Front 7

objective and deliverables of Prime Supermarket case

Back 7

obj: improve performance of FJD category at prime

deliverables:
- growth and mkt. share analysis for each juice and brand
- variety/product mix analysis for each type of juice and brand
- data analysis program to help the company make future decisions

Front 8

changes in retail industry

Back 8

POWER SHIFT TO LARGE RETAILERS: from mfg. "push" to consumer "pull"; bulk buying, large stores, economies of scale, control of shelves at extended store chains; store bran recognition and acceptance

PRODUCT PROLIFERATION: excess demand for shelf space, slotting fees, excessive product promotions

INTEGRATION OF SUPPLY CHAIN PARTNERS

new system more complex -- scan data

Front 9

retail performance review process

Back 9

retailer strategy and positioning

growth and mkt. share analysis

product mix/variety analysis

recommendations

Front 10

prime's competitive positioning

Back 10

differentation: better variety, customer service, everyday low prices on key items

target mkt: food and pharmacy customers (women 25-54)

primary category role: profit contributor

Front 11

DDS concept

Back 11

database
decision models
decision makers insights
interactive analytical modeling process

Front 12

DDS design principles

Back 12

reliability: output correct and consistent

audibility: user able to trace steps to generate output

modifiability: capable of easy modification

elegant simplicity: not complex, approp. degree of automation, approp. calculation/processing method

Front 13

DDS effectiveness test

Back 13

purpose -- goals? what questions DDS answers?

data - type and qty./source?

calculations -- which parts are calculated? what formulas are used?

changes -- easy to modify model or change data

Front 14

structured approach vs. ad hoc approach

Back 14

STRUCTURED: SDLC appropriate, simple model, many deliverables, large data set, BUILDER IS NOT THE USER

AD HOC: smaller data set, few deliverables, simple model, BUILDER IS THE USER

Front 15

DDS design and development process

Back 15

1. design
2. build and test
3. document

Front 16

DDS design 4 step process

Back 16

1. OUTPUT FORMS
-- METRICS: what to measure and how
-- TABLES AND GRAPHS: layouts and results groupings

2. PROCESSING LOGIC
-- MODELS to generate output metrics
-- EXCEL TOOLS/ METHODS to populate output form

3. DATA MODEL
-- SOURCES
-- STRUCTURE: table row and col. design
-- LINKS: among tables

4. EXCEL MULTISHEET LAYOUT
-- WHAT GOES WHERE in file

Front 17

best DDS design processes

Back 17

goal: elegant simplicity

approach: art, not science

process: iterate among 4 steps until the goal is met

Front 18

DDS design tips

Back 18

work backwards from end product

catch and fix problems early so they don't spiral out of control

ACID TEST: total stranger must be able to replicate analysis with new data set with no help from system developer

Front 19

prime supermarkets avail. data

Back 19

04-05 data for product SKU for Prime and for all firms combined competing in SE region:
-- annual sales, annual vol in sales (gal), ave. price/gal for all sales (w/ and w/out marketing promotion),

data classified into 4 subcategories (generic type of juice) and 8 mfg. brands

Front 20

growth rate metric

Back 20

% growth rate for mkt: (mkt. sales 05 - mkt. sales 04)/mkt sales 04

% growth rate for prime: (prime sales 05 - prime sales 04)/prime sales 04

Front 21

mkt. share metric

Back 21

prime's mkt. share = (primes sales in 05)/(total mkt. sales in 05)

Front 22

variety metrics

Back 22

variety index ratio = (# prime SKUs)/(# mkt. SKUs)
-- calculate for total categories, subcategories, and brands

TOP 10/BOTTOM 10 comparison: 10 SKUs carried by prime with LOWEST sales for yr & 10 SKUs carried by prime with HIGHEST yearly sales that PRIME DOES NOT CARRY

Front 23

uses for spreadsheets

Back 23

small simple databases

grid for drawing (flow charts)

financial models

stat. analysis

project mgt. (task/timeline)

Front 24

alt. methods for deliverable processing

Back 24

brute force: sort, output tables = data source for graph/chard

conditional calculations: ex: "sumif"

pivot table: report and chart

Front 25

data model

Back 25

describes how data is organized, managed, and accessed

efficient and effective organization of data: database principles tempered by analysis reqs. of excel based DSS

Front 26

database

Back 26

collection of tables that organizes and stores related info.

Front 27

table

Back 27

lists of rows and cols. containing data about a subject

Front 28

columns

Back 28

field/attributes of a subject

Front 29

rows

Back 29

records of individual instance

set of attributes

Front 30

database design principles

Back 30

MINIMIZE REDUNDANCY: duplicated data wastes space; increases likelihood for error

EFFICIENT USE OF STORAGE SPACE AND PROCESSING CAP.

ENSURE CORRECT INFO.: reports based on wrong data lead to bad choices

Front 31

good database design

Back 31

info. divided into subject based table to reduce redundant area

tables are linked together to retrieve data

ensure accurate info.

DATA MODEL IND. OF ANALYSIS AND PROCESSING REQS.

Front 32

DBMS design process

Back 32

PURPOSE: range of expected uses

SOURCES: identify and collect required data

DIVIDE INFO. INTO TABLES: each subject is separate table

DEFINE COLS./FIELDS FOR EACH TABLE: info. attributes assoc. with each entity/subject

SET UP TABLE RELATIONSHIPS: common fields to link across tables

Front 33

DBMS vs. excel DSS

Back 33

DBMS: linked tables consolidated through queries
-- FIELD STRUCTURE: based on subject attribute relationships inherent in data
-- independent of analysis tools and reqs.

EXCEL DSS: fewer tables, consolidate all info. to be analyzed in a single table;
-- FIELD STRUCTURE: based on analysis reqs. for grouping/filtering
-- EXCEL ANALYSIS TOOL/method influences design

Front 34

why analyze growth?

Back 34

doesn't stand still -- usually grows or declines

steady state is declining relative to to competitors that are growing

growth masks mistakes and incompetence

Front 35

why analyze mkt. share?

Back 35

economies of scale advantage

brand recognition -- go for the best

leverage with suppliers

correlated with LT profitability

new venture myth of capturing small amt. of mkt. share in a big mkt.

GE rule: if not 1,2,3 -- get out -- disadvantages too great to overcome

Front 36

quadrant charts

Back 36

plot growth/mkt. share info. as quadrant bubble charts

purpose: facilitate interpretation through visualization
-- 4 quadrants = benchmarks to eval. results

Front 37

Quad. chart midpoint selection criteria

Back 37

choose midpts. for x and y axis that yield most meaningful characteristics of bubble positions in chart based on overall goal of analysis

-- look for benchmark values for x and y that provide the best context for comparison for data plotted
-- choose measure external to or broader than scope of data plotted if you can
** overall mkt. share for comp.
** all stores in chain as benchmark for comparing performance

Front 38

why analyze product mix?

Back 38

important part of consumer shopping choice

very important in space constrained environment

source of differentation and comp. advantage

one retail mkt. mix elements: price, promotion, place/dist., product/shelf space utilization