3_1: Market Analytics: Analysing and Predicting aggregated Demand and Competiton

Question 1

What are Decision Models?

Lif

Answer 1

form the core of the marketing analytics approach to address marketing problems:

Question 2

What are the buillding blocks of Decision Models?
What are the Inputs and Objectives?

Answer 2

–>the buillding blocks are market response models

• Inputs: marketing actions (price, advertising)
• Response models: linkage from those inputs to the measurable outputs of concern to the firm (customer awareness levels, sales levels)
• Objectives: measse that the firm uses for monitoring and evaluating those actions

Question 3

Key (independable variables) inputs affecting Brand Sales?

Answer 3

• Base factors: Sales metrics (price, distributions) and competitors activitities
• Trade merchandising: marketing activities driven by retailers (displays, features, temporary price)
• Promotions: driving by manufactures
• Media advertising:
• Sales seasonality

Question 4

What is the variable transformation in Market Response Models doing?

Answer 4

• variable transformations refer to the process of altering or modifying the original variables to improve the model’s performance or interpretability.

Question 5

Why do we do variable transformation in Market response models?

Answer 5

• transformations are applied to the predictor variables (also known as independent variables or features) used in the model

–>By transforming the variables, analysts aim to create a better fit between the predictor variables and the response variable, leading to a more accurate and reliable model.

Question 6

Process (variable) transformation
(Average price, distribution factors, Seasonality)

Answer 6

transform independable inputs into usable data

Average price= Base price + discount
–>to better evaluate the impact of promotional discounts and base price elasticity

Distribution: split the width: number of stores, depth: how many products per store

Seasonality: is a periodic repetition or cyclic variation in time series data
Important before quantifying the impact of marketing promotions
–>it is crucial to separate the impact of seasonality to ensure no misguided recommendations

Question 7

What are the three Multivariate regression models used in Market Response Models?

Answer 7

• Linear Market Response Models
• Logistic functions
• exponential models

Question 8

What curve to we use to Model the Saturation Effect on Media promotions?
(Multivariate regression Models)

and why not linear?

Answer 8

We use a S-shaped curve to model the saturation effect on Media promotions

Not linear: since the relationship between media adstock and sales is not linear

–>However, a logistic function can map the media impact on sales

Thus we use s-shaped function since, It can reproduce market saturation at some point in time

Question 9

For what is the media saturation curve helpful?

Answer 9

The media saturation curve helps the busienss in identifying the optimal execution levels for maximum returns –> optimal in certain range in the middle

Question 10

What do we use in a case of Market Rolllout?

Answer 10

Linear Model would not capture the fast-growing sales level, thus we use an exponential model since it can capture this over-proportional growth

Question 11

What are the effects observed from the Output of Market response models?

Answer 11

Marketing promotions:

• synergy effect
• Decay effect
• hysteresis effect
• “wear out effect
• Halo
• cannibalization effect

Question 12

What is the Marketing Promotions Synergy Effect?

Answer 12

synergy effect occurs when the total effect of marketing promotions is greater than the sum of their individual parts

Question 13

What is the intuition by examining the associations in MBA?

Answer 13

examing the association between products bought together

–>can draw valueable insight that help to understand customer behavior

decision: product placement, promotions and pricing strategies

Question 14

What is Market Basket Analysis (MBA)?

Answer 14

used to analyze customer transaction data, such as purchases, services, and website interactions,

–>to identify patterns and relationships between products and services

Question 15

What does a Consumer´s market Basket refer to?

Answer 15

A consumer´s market basket refers to the list of products that they have purchased

Question 16

What is Advertising Stock?

Answer 16

Advertising Adstock describes the prolonged(anhaltend) or lagged effect of advertising on consumer purchase behavior

Question 17

What is the decay effect?

Answer 17

The decay effect refers to the fading of memory of an ad and lack of continued response to it. (usually expressed in terms of half life of the advertising)

Question 18

What is the Carryover effect?

Answer 18

describes the influence of a current marketing expenditure on sales in future periods

Question 19

How can the carryover effect and the decay effect influence the adstock?

Carryover effect and Adstock

Answer 19

The carryover effect contributes to the building of Adstock.
- Each exposure to advertising leaves a residual impact on consumer behavior (creating memory or impression)
- these impression accumulate to the Adstock
–>The more frequent and consistent the advertising exposure, the stronger the carryover effect and the higher the Adstock level

Question 20

How can the carryover effect and the decay effect influence the adstock?

Decay effect and Adstock

Answer 20

The decay effect influences the erosion of Adstock over time.
- memory or impressions created by advertising gradually weakens
- If advertisisng is not reinforced through repeated exposures, the impact diminsihes and Adstock declines

Question 21

, What does the term “Association mean in Market Basket Analysis?

Answer 21

Association: Co-occurence of two or more things (how likely it is to find two products together in the basket)

Question 22

, What does the term “Transaction” mean in Market Basket Analysis?

Answer 22

Transaction: A group of one or more items (set) that occur in an observation
–>represents a single purchase or a set of items bought together by a customer in a single occurrence (one Basket = one transaction)

Question 23

, What does the term “Rule” mean in Market Basket Analysis?

Answer 23

Rule: Incidence across transaction of one set of items as condition of another set of items
–>represents an association or relationship between sets of items in customers’ transactions. These rules are typically in the form of an “if-then” statement

For example, a rule might state, “If a customer buys item A and item B, then there is a high probability they will also purchase item C.

Question 24

What does the Metric Support mean in Market Basekt Analysis?

Answer 24

Support: Proportions of all transactions that contain the set of items (relative frequency)

–>measure of the frequency or occurrence of a particular itemset or rule in the dataset. It indicates the proportion of transactions that contain the specific items or itemsets being analyzed

Question 25

What does the Metrics confidence mean in Market Basekt Analysis?

Answer 25

Confidence: Support for the co-occurnece of all items in a rule, conditional for the set alone

—>Conditional probability that an itemset will occur, gien that another item or itemset has already occured
–>measures the reliability or certainty of a rule in Market Basket Analysis.

Question 26

What does the Metrics Lift mean in Market Basekt Analysis?

Answer 26

Lift: Support of a set, conditional on the joint support of each element indenpently

–>compares the likelihood of an itemset / rule occurring together to the expected or random co-occurrence of the items.

Lift >1 –>items tend to be purchased together
–>Never put both products on sale

Question 27

How to calculate the lift?

Answer 27

Lift calculation

likelihood of an event occurring when a specific condition is present (e.g., a customer purchases product A) to the likelihood of the event occurring in general (e.g., a customer purchasing any product)

Question 28

How do Neural networks forecast sales? Compared to multiple regression

Answer 28

Neural networks seaches many models to find a relationship with independent variables that best predicts the dependent variables
- they can capture even complex relationships that multiple regression cannot find

Question 29

What affects a product´s average price?

Answer 29

1. Manaufacturer discount
2. Temporary price reductions from the retailer
3. Stock keeping units level price difference

Question 30

What is the Histeresis Effect in Media Promotions (Sales)?

Answer 30

–>Sales increase after marketing stimulus, fade out but sales/market share stays at the higher level than before (long-term improvement of sales)

–>opposite of the new trier “wear out effect

Question 31

What is the New Trier “wear out” effect?

Answer 31

Sales reach a peak and then dceline and settledown at steady state, despite the maintenance of advertising at a high level

–>The new trier “wear out” effect occurs beyond a certain exposure level and is common for frequently purchased products

Question 32

What is the Halo effect?

Answer 32

consumers bias towards products due to their favorable experiecne with other products made by the same manufacturer

–>Driven by brand equity

Question 33

What is the Cannibalization Effect?

Answer 33

• Reduction in sales because otherr (new) prodsuct is launched
• Important to identify and distinguish the effect for correctly estimate the impact of promotions

Question 34

What are Market Share Models?
what is the difference between single brand models to market share models

Answer 34

used to understand how marketing of every brand impact the results in a competitive marketplace

–>Unlinke single-brand sales modles, they also include competition

Question 35

Formula and intution behind Market Share?

Answer 35

Market share = k*M (k= proportionality constant; M=Marketing effort)
–>the greater the marketing effort, the greater the market share should be

Question 36

What is Bell, Keeney & Little Market share theorem idea?
(attraction),

Answer 36

they argue, that the only determinant of market share is the attraction which consumer feel towards each alternative brand

Question 37

What is Bell, Keeney & Little Market share assumption?
(attraction),

Answer 37

attraction of a brand is proportional to is marketing effort and interpret attractions as probabilities

–>Attraction models

Question 38

What is Kotler´s Theorem of Market Share?

Answer 38

The theorem says:

That the market share of a brand = to the brand´s share of the total marketing effor

Question 39

What assumption do Market Share attraction Model underline?

Answer 39

attraction of a brand depends on its marketing mix

Question 40

How do Market share attration models derive market shares?

Answer 40

–>Modelling attractions as probabilities
Brand attractions are determinants of market shares!

–>they interpret market shares as purchase probabilities

Question 41

Strategic PLanning:
What do Market Response Models offer?

Answer 41

Superior allocation:
- Identify marketing channels tht max returns
- Compare tradition channels with emerging, cost effective ones

Superior execution:
- propose the right execution levels for highly effective markeitng channels (avoid saturation)
- Investigate successful campaings, their types or quality scores
- compare campaign duration

Simulation & optimization:
- Forecast sales trend as per current marketing mix plan
- Predict sales impact subject to a planned increase/drop in advertising/price
- Create an optimum marketing mix plan to achieve target revenue growth

Question 42

Challenges of Brand & Product Sales Models

Answer 42

• requires leveraging a diverse set of data sources
• missing measurment standards for external data –>data credibility
• difficulty of capturing the long-term effects of advertising campaigns as component of base sales
• ## Making market mix recommendations –>big operational challenge

Question 43

What is Non- transactional data?

Answer 43

• refers to data that is not explicitly tied to individual transactions or purchases.
• It includes data that describes characteristics, attributes, or preferences of customers, products, or other entities related to the market being analyzed

Question 44

Which curve is used to forecast sales of a new Product and what can they learn from the curve?

Answer 44

S-Curves are used to forecast sales of a new product

• Upper limit of sales
• Inflection point (Wendepunkt)
• Future profitability

Question 45

What is the usage of the Pearl curve?

Answer 45

Pearl curve is used to model the path of product diffusion by fitting the logistic curve

Question 46

What needs to be down when fitting a Logistic curve with seasonality?

Answer 46

the seasonality of sales must be incorporated into the estimation process by

–>multiplying the forecast from the S-curve with the appropriate seasonal index

Question 47

What is the Gompertu Curve?

Answer 47

is another function form that is often used to fit an s-curve to data for forecasting sales

Question 48

What is the difference between Neural network and Regression analysis in forecasting sales? (Assumptions, relationships)

Answer 48

Observations and Goal is the same

Regression analysis
Assumptions: requires statistical assumptions about the data
Relationships: relationship between IVs and DV needs to be specified

Question 49

For what can we use Moving averages?

Answer 49

Used to smooth data and eliminate seasonality

Question 50

What are the terms consiered in Multiplicative Model with Trend and seasonality?

Answer 50

Sales t = Base*trend * seasonal index month t

Trend: % monthly increase in the level of airline miles
seasonal index: % by which airlines travel for the month is above or below an average month
–>Multiplicative seasonal indexes must average 1

Question 51

What kind of Model is the Multiplicative Model with Trend and seasonality?

Answer 51

It is a non-linear forecasting model because one can raise the trend to a power and multiply it, rather than add terms involving the seasonal indices

Question 52

What are the three terms considered in an additive Model with Trend and Seasonality?

Answer 52

Predicted period t sales= Base + Trend*t + Seasonal index month t

Base: best estimate of the level of monthly airline miles at the begining of the observed period (without seasonality)

Trend: Best estimate of monthly rate of increase in airline miles travelled

Seasonal index: each month has seasontial index to reflect –> the seasonal indexes must average to 0.

Question 53

Using Neural networks to forecast sales:
(Observations, Goal, Assumptions, Relationshipps)

Answer 53

Observation: each observation contains a value for each dependent and independent variable

Goal: to make accurate predictions for the output or the dependent variable

–>Observation and Goal: same as for regression analysis

Assupmtions: No statistical assumptions about the data needed
Relationships: find patterns by learning from the data

Question 54

Which curve to use in predicting sales?
(Pearl Curve or Gompertz Curve)

Answer 54

3If future sales benefit from previous product sales –>Pearl curve

If future sales do not benefit from previous sales –>Gompertz curve