3_1: Market Analytics: Analysing and Predicting aggregated Demand and Competiton Flashcards
What are Decision Models?
Lif
form the core of the marketing analytics approach to address marketing problems:
What are the buillding blocks of Decision Models?
What are the Inputs and Objectives?
–>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
Key (independable variables) inputs affecting Brand Sales?
- 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
What is the variable transformation in Market Response Models doing?
- variable transformations refer to the process of altering or modifying the original variables to improve the model’s performance or interpretability.
Why do we do variable transformation in Market response models?
- 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.
Process (variable) transformation
(Average price, distribution factors, Seasonality)
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
What are the three Multivariate regression models used in Market Response Models?
- Linear Market Response Models
- Logistic functions
- exponential models
What curve to we use to Model the Saturation Effect on Media promotions?
(Multivariate regression Models)
and why not linear?
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
For what is the media saturation curve helpful?
The media saturation curve helps the busienss in identifying the optimal execution levels for maximum returns –> optimal in certain range in the middle
What do we use in a case of Market Rolllout?
Linear Model would not capture the fast-growing sales level, thus we use an exponential model since it can capture this over-proportional growth
What are the effects observed from the Output of Market response models?
Marketing promotions:
- synergy effect
- Decay effect
- hysteresis effect
- “wear out effect
- Halo
- cannibalization effect
What is the Marketing Promotions Synergy Effect?
synergy effect occurs when the total effect of marketing promotions is greater than the sum of their individual parts
What is the intuition by examining the associations in MBA?
examing the association between products bought together
–>can draw valueable insight that help to understand customer behavior
decision: product placement, promotions and pricing strategies
What is Market Basket Analysis (MBA)?
used to analyze customer transaction data, such as purchases, services, and website interactions,
–>to identify patterns and relationships between products and services
What does a Consumer´s market Basket refer to?
A consumer´s market basket refers to the list of products that they have purchased
What is Advertising Stock?
Advertising Adstock describes the prolonged(anhaltend) or lagged effect of advertising on consumer purchase behavior
What is the decay effect?
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)
What is the Carryover effect?
describes the influence of a current marketing expenditure on sales in future periods
How can the carryover effect and the decay effect influence the adstock?
Carryover effect and Adstock
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
How can the carryover effect and the decay effect influence the adstock?
Decay effect and Adstock
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
, What does the term “Association mean in Market Basket Analysis?
Association: Co-occurence of two or more things (how likely it is to find two products together in the basket)
, What does the term “Transaction” mean in Market Basket Analysis?
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)
, What does the term “Rule” mean in Market Basket Analysis?
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.
What does the Metric Support mean in Market Basekt Analysis?
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
What does the Metrics confidence mean in Market Basekt Analysis?
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.
What does the Metrics Lift mean in Market Basekt Analysis?
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
How to calculate the lift?
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)
How do Neural networks forecast sales? Compared to multiple regression
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
What affects a product´s average price?
- Manaufacturer discount
- Temporary price reductions from the retailer
- Stock keeping units level price difference
What is the Histeresis Effect in Media Promotions (Sales)?
–>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
What is the New Trier “wear out” effect?
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
What is the Halo effect?
consumers bias towards products due to their favorable experiecne with other products made by the same manufacturer
–>Driven by brand equity
What is the Cannibalization Effect?
- Reduction in sales because otherr (new) prodsuct is launched
- Important to identify and distinguish the effect for correctly estimate the impact of promotions
What are Market Share Models?
what is the difference between single brand models to market share models
used to understand how marketing of every brand impact the results in a competitive marketplace
–>Unlinke single-brand sales modles, they also include competition
Formula and intution behind Market Share?
Market share = k*M (k= proportionality constant; M=Marketing effort)
–>the greater the marketing effort, the greater the market share should be
What is Bell, Keeney & Little Market share theorem idea?
(attraction),
they argue, that the only determinant of market share is the attraction which consumer feel towards each alternative brand
What is Bell, Keeney & Little Market share assumption?
(attraction),
attraction of a brand is proportional to is marketing effort and interpret attractions as probabilities
–>Attraction models
What is Kotler´s Theorem of Market Share?
The theorem says:
That the market share of a brand = to the brand´s share of the total marketing effor
What assumption do Market Share attraction Model underline?
attraction of a brand depends on its marketing mix
How do Market share attration models derive market shares?
–>Modelling attractions as probabilities
Brand attractions are determinants of market shares!
–>they interpret market shares as purchase probabilities
Strategic PLanning:
What do Market Response Models offer?
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
Challenges of Brand & Product Sales Models
- 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
What is Non- transactional data?
- 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
Which curve is used to forecast sales of a new Product and what can they learn from the curve?
S-Curves are used to forecast sales of a new product
- Upper limit of sales
- Inflection point (Wendepunkt)
- Future profitability
What is the usage of the Pearl curve?
Pearl curve is used to model the path of product diffusion by fitting the logistic curve
What needs to be down when fitting a Logistic curve with seasonality?
the seasonality of sales must be incorporated into the estimation process by
–>multiplying the forecast from the S-curve with the appropriate seasonal index
What is the Gompertu Curve?
is another function form that is often used to fit an s-curve to data for forecasting sales
What is the difference between Neural network and Regression analysis in forecasting sales? (Assumptions, relationships)
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
For what can we use Moving averages?
Used to smooth data and eliminate seasonality
What are the terms consiered in Multiplicative Model with Trend and seasonality?
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
What kind of Model is the Multiplicative Model with Trend and seasonality?
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
What are the three terms considered in an additive Model with Trend and Seasonality?
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.
Using Neural networks to forecast sales:
(Observations, Goal, Assumptions, Relationshipps)
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
Which curve to use in predicting sales?
(Pearl Curve or Gompertz Curve)
3If future sales benefit from previous product sales –>Pearl curve
If future sales do not benefit from previous sales –>Gompertz curve
