Please enable JavaScript.

Coggle requires JavaScript to display documents.

Data Analytics (Customer Analytics (Case studies (Evolution: (1950-s --…

- - - - Desktop surveys
      - Offline surveys
      - How should you measure?
        
        Itemized categories
        
        Comparison to what???
        
        Comparative
        
        Big loss of info
        
        Ranks
        
        Could be too many comparisons
        
        Likert scale
        
        Paired comparison
        
        Only for small number of comparisons
        
        Continuous
      - Main attributes
        
        Validity
        (predictive validity)
        
        Reliability
        (test-retest)
      - Survey design
        
        Result are generalizable to population
        
        Develop a detailed list of 'bits' of info
        
        2.Determine appropriate data collection
        
        3.Write draft questions
        
        4.Design
        
        Test/pivot
        
        Get approval from all parties
        
        Design again
        
        Production
        
        Non-response is important too (follow them and know why?)
        
        Right format for questions
        
        Open-ended
        
        Wide range for respondents to answer
        
        Could create a base and insights for future qualitative research
        
        Close-ended
        
        Useful for qualitative researc
        
        Hard questions after the easy
        
        Avoid order bias (example about 'want-to-buy' question after different previous questions)
      - Mobile surveys
    - - 0-10 scale:
        9-10 -- promoters
        7-8 -- passives
        0-6 -- detractors
        NPS = promoters - detractors
      - Similar to ASCI index
      - Correlated with profitability
        BUT link is not so good
        
        Satisfaction Axis -- Satisfaction is significant in polar parts of the scale. In other words, it influences profit poorly in the medium parts
      - Is not a metric itself
    - - Nielsen
      - IRI
      - Spins
      - Compete-com
      - OnScore
      - Ipsos
    - - Own media
      - Paid media (advertising)
      - Earned media (likes, reposts, shares)
    - - Correlation (X related with Y)
      - Temporal antecedence (X happened before Y)
      - No third factor
  - - - Probability predictions
      - Important predictors
        
        Recency (how long since the last purchase)
        
        Frequency
        
        Monetary value
  - - - 1950-s -- Store level analytics:
        Are the coupons effective?
        How prices related to sales?
        Regional ads efectiveness
      - 1960s-1970s -- Direct mail (almost household level):
        What is the impact of mail ads on cust. behavior?
        How do individual prices relates sales?
        How does product assortment influences purchase behavior?
        What types of ads are the most effective?
      - 1980s -- Store scanners (closer to customer):
        Track customers over time
        person level coupon- and discount usage
      - 1990s-2000s -- modern age, internet
    - - All the ads that you see (display), increase the effectiveness of the last click (SEM)
      - GRP (Growth Rating Points) = Reach * Frequency
      - Ad channels do not cannibalize each other in long run, but could in the short run
    - - Social network bring additional value to promo (word-of-mouth)
    - - Customer's way
        
        Order deviation (you buy products in the wrong, ineffective way)
        
        Order deviation is associated with more purchases. Travel one is not.
        
        Planned and unplanned purchases
        
        Travel deviation (you buy products in the right way, but with a lot jiggling)
    - - Kohl's uses wi-fi on your smartphone to look your internet history and give you ad in the real time when you in the store. For example, it can see if you look for shoes, then notice that right now you are in the shoe store and send you the ad
      - Netflix use the machine learning system in order to recommend a show
      - Amazon ship goods before you buy them
- - - - How to determine right amount of units to order?
      - Сценарный анализ (через вероятность)
      - Скользящее среднее
      - От сезонности можно избавляться
        
        Простое среднее
        
        Среднее по сезонам
        
        Фактор сезонности (2 / 1)
        
        Убираем сезонность (каждое значение на соответствующий фактор 3)
      - Прогнозы нужно корректировать на ошибку (фактический спрос / прогноз)