digitl data

March 30, 2018 | Author: varshasri | Category: Data Analysis, Brand, Information Technology, Business, Technology


Comments



Description

LESSON 6: “Insights from Digital Data” Key questions in your plan direct your analysis 5 Primary analysis techniques used to analyze data 3 Each technique Important rules of If there is no “60offers a unique view context ensure second story” there on a key question impact in analysis is no story at all . . Documenting your plan at this stage is critical to success Business Objective Grow Loyalty Key Question Data —> Source(s)* How has consumer interest in our brand trended over time? Search Volume —> Google Trends Customer Inquiries —> CSR Database What consumer group is our strongest advocate? Consumer Groups —> Segmentation Study Twitter Volume —> Twitter API Which marketing programs have grown advocacy? Marketing Events —> Company Intranet Site Hashtag Volume —> Topsy Note: * Here “source” is used in a way synonymous with “tool” . There are five primary categories of marketing 
 data analysis Predictive Descriptive Causal Inferential Exploratory Source: Adapted from Jeffrey Leek. “Types of Data Science Questions” Note: A sixth analysis approach (“Mechanistic”) has been omitted . There are five primary categories of marketing 
 data analysis Uses data on some object to predict the future value of others First kind of analysis performed Commonly applied to raw data Depends heavily on having the right data and data quality Must remember that if X predicts Y. “Types of Data Science Questions” Note: A sixth analysis approach (“Mechanistic”) has been omitted Commonly the goal of surveys Results depend heavily on population and sample Seeks to discover connections and patterns in the data Good for defining future studies Should not be used alone for generalizing or predicting . X does not cause Y Predictive Descriptive Generalizations typically not possible without modeling Uses a small set of data to say something about larger set Seeks to determine what happens to one variable when a change is made to another Usually seen as average effects Usually the “gold standard” for data analysis Causal Inferential Exploratory Source: Adapted from Jeffrey Leek. “Types of Data Science Questions” Note: A sixth analysis approach (“Mechanistic”) has been omitted .These categories offer graduated levels 
 of analysis depth Predictive Descriptive Causal Inferential Exploratory Source: Adapted from Jeffrey Leek. “Types of Data Science Questions” Note: A sixth analysis approach (“Mechanistic”) has been omitted Identify when interest started for the brand. and other interesting patterns in the data .Q: “How has consumer interest in [X] trended 
 over time?” Determine whether consumers are or are not interested in your brand Offer recommendations for actions that will successfully drive brand interest in the future Predictive Descriptive Causal Explain what has been done in the past that successfully drove brand interest and other influential brand attributes Inferential Provide a description of what a typical consumer who is interested in our brand looks like and how they behave Exploratory Source: Adapted from Jeffrey Leek. when it ended. g.. “Types of Data Science Questions” Note: A sixth analysis approach (“Mechanistic”) has been omitted Outcomes analysis Other regression and performance correlation analyses ..Different analyses require the collection of 
 different data Media optimization modeling Attribution modeling Other response modeling (e. customer & channel response) Predictive Descriptive Web traffic data reports Web server performance data reports Web transaction data reports Competitive intelligence Clickstream analysis Usability studies Voice of customer studies Causal Inferential Experimentation and testing Other surveys when a sample of a population is included Multivariate & A/B testing Site optimization Other campaign optimization (e.g. “Heavy up” media tests) Exploratory Source: Adapted from Jeffrey Leek. goals. or even prior performance to give some kind of context • Include insights — in words — to summarize performance and recommend actions in every analysis Source: Kaushik.Three rules of context ensure impact in your analysis • Never produce an analysis that reports a metric all 
 by itself … period • Use benchmarks (internal or external). “Five Rules for High Impact Web Analytics Dashboards” (2007) . President FCB Chicago (2006) . You can even tell the story of Dostoyevsky’s ‘The Idiot’ in 60 seconds.“ If there is not a 60 second story. then there is no story at all. ” Michael Fassnacht. LESSON 6: “Insights from Digital Data” Key questions in your plan direct your analysis 5 Primary analysis techniques used to analyze data 3 Each technique Important rules of If there is no “60offers a unique view context ensure second story” there on a key question impact in analysis is no story at all . com/2006/07/makingaccountability-sexy.com/2006/06/veil-ofstatistics.blogspot.html • The Veil of Statistics: 
 http://marketinggeek.html • Making Accountability Sexy: 
 http://marketinggeek.Supplemental reading for this lesson • Dialog Intelligence: 
 http://marketinggeek.com/2006/12/dialogintelligence.blogspot.html .blogspot. Retrieved from http://jtleek. Jeffrey Leek. Avinash Kaushik. “Five Rules for High Impact Web Analytics Dashboards”. 2007.References 1. “Types Of Data Science Questions”.com/modules/01_DataScientistToolbox/ 03_01_typesOfQuestions/#1 2.kaushik.net/avinash/five-rules-for-high-impact-webanalytics-dashboards/ . Retrieved from http:// www.
Copyright © 2024 DOKUMEN.SITE Inc.