Exercise 1: The rise of analytics
Task 1: Read Competing on Analytics
- What does it mean for a company to “compete on analytics”?
How is this different from simply using data or reports in decision-making?
To “compete on analytics” means that a company uses data, statistical analysis, modeling, and experimentation as a core strategic capability across the organization. Analytics is not just a support tool; it becomes a basis for competitive advantage. Davenport (2006) argues that such firms do not merely know what happened in the past, but use predictive and optimization methods to determine what will happen, what customers will do, what prices maximize profit, and how processes can be improved.
This is different from simply using data or reports in decision-making. Many firms use descriptive statistics and reports, but analytics competitors go further by:
- applying advanced models rather than only dashboards or summaries,
- embedding analytics across multiple functions rather than in isolated departments,
- coordinating these efforts under an enterprise strategy,
- and repeatedly using evidence to improve decisions over time.
In short, ordinary data use supports decisions; competing on analytics makes superior, repeatable, organization-wide decision-making a source of advantage.
- What organizational capabilities are required to compete on analytics?
Consider aspects such as leadership, culture, people, and technology.
According to Davenport (2006), companies that compete on analytics need a combination of organizational and technical capabilities. There is a need for:
Strong senior leadership, ideally from the CEO. Top executives must actively champion data-based decision-making and signal that analytics is central to strategy.
Analytics-oriented culture. Employees are expected to justify decisions with evidence, performance is measured quantitatively, and the organization develops a “test and learn” mindset in which ideas are validated through experiments.
The right people. This includes not only technically skilled analysts, but also people who can translate complex analyses into business language and work closely with decision-makers. Analytical talent must be numerous enough to matter and visible enough to influence the business.
Enterprise approach. Analytics should not be fragmented across departments with separate tools and inconsistent data definitions. Instead, it should be managed coherently across the organization, with shared leadership, standards, and infrastructure.
The right technology and data foundation. Davenport (2006) highlights the importance of a clear data strategy, integrated and standardized data, data warehouses, business intelligence tools, and sufficient computing power. These capabilities take time to build and require sustained investment.
Overall, competing on analytics depends on aligning leadership, culture, talent, organization, and technology around fact-based improvement.
- Which companies or industries compete on analytics?
Give examples and explain how analytics creates their competitive advantage.
In Davenport’s paper, several examples illustrate what it means to compete on analytics.
Amazon is a central example. Its advantage comes from using analytics in areas such as customer recommendations, supply chain management, and operational optimization. This helps it personalize the customer experience and manage large-scale complexity efficiently.
Capital One competes on analytics through extensive experimentation and quantitative analysis in marketing, pricing, and risk assessment. Its competitive advantage comes from constantly testing offers and learning which customers to target and how to manage risk profitably.
Harrah’s (now Caesars Entertainment) used analytics to improve customer loyalty, promotions, pricing, and service. Its advantage came from understanding customer behavior in great detail and using this knowledge to increase retention and spending.
Progressive used analytics to refine insurance pricing. Instead of broadly rejecting risky customers, it analyzed detailed customer segments and priced them more precisely, allowing it to profit where competitors relied on cruder categories.
Marriott applied analytics to revenue management, pricing, and customer retention. By optimizing room prices and related offerings, it improved revenue performance across its hotels.
Davenport also points to sports organizations such as the Boston Red Sox, the Oakland A’s, and the New England Patriots. Their advantage came from applying statistical analysis to player selection, performance evaluation, and other decisions that were traditionally based more heavily on intuition.
More broadly, industries that seem especially suited to competing on analytics are finance, retail, hospitality, logistics, insurance, and professional sports, because they generate rich data and make many repeatable decisions that can be improved through modeling and experimentation.
Task 2: Create an analytical notebook in Jupyter
A good notebook solution should be structured according to CRISP-DM:
Business understanding
Clearly describe the business problem, explain why it is relevant, and define the target of the analysis.Data understanding
Import the dataset, inspect its structure, and create a first descriptive overview (for example: variables, missing values, distributions, summary statistics, or simple visualizations).Data preparation
Anticipate how data should be cleaned and prepared. This may include handling missing values, selecting variables, encoding categories, or splitting the data into training and test sets.Modeling
Select a first predictive model that fits the problem. Reasonable starting points could be:- regression models for predicting numeric outcomes,
- classification models for predicting categories,
- or a machine-learning models and AI for prediction.
Evaluation
Plan to assess whether the model is useful for the business problem. Explain the chosen metric and interpret the results in plain language.Interpretation / deployment
Anticipate what the findings could mean for decision-making and what could be done next.
In the notebook, students should also indicate which parts are:
- descriptive (understanding and summarizing the data),
- predictive (building and evaluating a model),
- prescriptive (discussing what actions could follow from the results).
A complete solution does not need to finish every step perfectly. It is acceptable to leave key coding parts as TODO, propose likely next steps, or discuss modeling options that could be tried. The goal is to demonstrate a structured analytics workflow and familiarize yourself with Jupyter Notebooks.