All Posts by ganpati

About the Author

Jan 20

Important Metrics for Analytics Managers

By ganpati | Analytics Tools

Key Performance Indicators (KPI’s) are the vital navigation instruments used by managers and leaders to understand whether they are on course to success or not. The right set of KPI’s will shine light on performance and highlight areas that need attention. Without the right KPI’s managers have no handle on the business, a bit like a pilot without the controls.

The problem with easily available technologies to store data is that most companies collect and report a vast amount of everything that is easy to capture and measure. As a consequence their managers end up drowning in data while thirsting for insights.

The organizations should have clear objectives and strategic directions in identifying the right KPIs for their businesses. Remember, navigation instruments are only useful if we know where we want to go. Therefore, first define the strategy and then closely link KPIs to the objectives.

The KPI’s that are most critical for any business are those that: measure financial performance, help understanding the customers, gauge market size and marketing efforts, measure operational performance, helps to understand employees and their performance and measure environmental and social sustainability performance.

Based on your area of interest you may consider focusing on a subset of these KPI’s. Nevertheless, when you grow your responsibilities as a manager and the importance of analytics grows within the organization more and more KPI’s become important to track and their interplay becomes important.

Jan 18

Business Analytics Interview Questions (Part 2)

By ganpati | Getting Started

In continuation to our earlier discussion on Business Analytics Interview Questions, in this post we will discuss some other topics on which questions are frequently asked in business analytics interviews.

Customer Lifetime Value

Historical lifetime value simply sums up revenue or profit per customer. Predictive lifetime value projects what new customers will spend over their entire lifetime. Which of the above will be more important for you as an analytics manager? Under what circumstances you will prefer one over the other?

Under new approaches of modeling CLV, all customers are considered different – and it is this recognition of what academics call “customer heterogeneity” that makes all the difference.How will you approach the calculation of CLV considering “customer heterogeneity”?

Suggest a comparative analysis of below two approaches to calculate CLV:

i. A historical, average revenue per customer per month metric (ARPU) is calculated for 2 years and multiplied by 24
ii. A linear regression model that latches onto how cohorts of users are changing over time and extrapolates to project CLV. (here is a guide to regression analysis)

Suggest an approach to compare CLV across digital (web based) and traditional (agent based) marketing channels for an insurance company. Once you calculate the CLV what approach would you suggest as an analytics manager to increase the ROI of marketing expenditure.

Predictive Sales Analytics

Most principles used in B2B selling today have been in place since the late nineteenth century and coincide with the rise of large mass manufacturing firms. National Cash Register, Westinghouse Electric and others created large, organized sales forces and with them, standardized sales and sales management techniques. How do you think the rise of predictive sales analytics shall drive genuine innovation in these areas?

Predictive analytics is an estimated $5 billion market that has seen $1.2 billion in VC funding specific to sales analytics. This shows there’s a strong economic case for deploying predictive sales analytics.

Even with a CRM in place, companies lose significant time creating and rolling-up company-wide sales forecasts. Compounding the problem is that forecasts are based on human judgment leading to inaccurate forecasts. How do you think Predictive Forecasting can be used to better manage this situation?

One benefit of sales analytics technology is that it can facilitate the alignment of sales and marketing and fulfill the promise of Sales Enablement. The end result is that marketing has the insight needed to produce content that drives engagement and revenue. How do you think this idea can be applied at a pharma company to boost sales.

By predictive lead mining, companies aim to identify high-quality, hidden prospects—those that share characteristics of your best customers. This is done by of analyzing a combination of CRM data, campaign results, opportunities won and lost, and Internet data mining.How do you think you can leverage that analytics to identify prospects at your firm. (for example, using lead information, teams can prioritize sales calls based on which leads are most likely to close and improve conversion rates throughout the entire customer lifecycle)

How would you use predictive sales analytics to lose fewer deals due to uncompetitive pricing and win more deals at profitable pricing?

Some Important Domains in Analytics

Behavioral analytics

Behavioral analytics is a recent advancement in business analytics that reveals new insights into the behavior of consumers on eCommerce platforms, online games, web and mobile applications, and IoT.

How will you formulate a strategy to capture Real-time vast volumes of raw event data across all relevant digital devices and applications used during sessions for an e-commerce company?

How will you plan for an automatic aggregation of raw event data into relevant data sets for rapid access, filtering and analysis?

Ability to query data in a number of ways, enabling users to ask any business question is important for behavioral analytics. What is the tool that you would prefer for a start up in order to build such capability?

Give few examples of built-in analysis functions such as cohort, path and funnel analysis.

Cohort analysis

Give examples of some actionable metrics. (An actionable metric is one that ties specific and repeatable actions to observed results [like user registration, or checkout].)

Give examples of few vanity metrics. (The opposite of actionable metrics are vanity metrics like web hits or number of downloads) which only serve to document the current state of the product but offer no insight into how we got here or what to do next.

Jan 15

Business Analytics Interview Questions

By ganpati | Business Analytics

When you are applying for a business analytics position, the questions that you will be asked are aimed towards your understanding of the business situation, your ability to assess the environment and the industry in which the firm operates and your ability to give a structure to the problem and developing a framework for arriving at a solution. Here are some typical scenarios discussed in business analytics interviews:

Basic Questions on Applying analytics to Business Problems

How will you start while approaching an analytics problem?/ What will be your first step in solving a business analytics problem? (You may refer to the article on first step to an analytical problem for framing an answer)

A large part of business relevant information originates in unstructured data that can’t be stored in SQL, spreadsheets or other structured databases. It’s generally text heavy and may contain text, numbers or other data types like images or audio files. How will you plan to unlock the potential of such data for a traditional company where employees and managers place heavy emphasis on structured databases and spreadsheets.

Questions Based on Business Experiments

1. You have to choose from multiple actions to decide which one will lead to increased sales? How will you make a decision?

2. There are multiple products on your portfolio and you need to prioritize which products should be released first in order to maximize revenue. How will you make a call?

3. You have to take a call on discontinuing one product from a set of N products. How will will you do that while minimizing the revenue impact while maximizing the cost savings from discontinuing a product.

4. You have to select from three different channels for your next marketing campaign. How will you maximize the response rate to the campaign.

5. As a manager you can only select two channels to proceed with the next rund of recruitment. How will you select which channel is better?

Customer Analytics

How would you use the following types of customer data to analyze the customers and get a better understanding of problems and opportunities for your customers for an insurance company (you can replace insurance with any other industry)-
a). Demographic data like gender, age, geography and income
b). Behavioral data like purchases, registration data, browsing, and device usage data
c). Interaction data like clicks, navigation paths and browsing activities
d). Attitudinal data like opinions, desirability, branding and sentiments

Technology is changing how customers interact with products even for traditional products like insurance, banking and Mutual Funds. As a marketing manager for a finance company that offers multiple products to its customers, what data will you collect to understand the changing behavior of your customers. How will you apply this understanding to gain new customers.

You have joined a company that has ignored the data revolution for years and is now losing its customer base to new generation of competitors. They don’t have enough infrastructure to collect or store data related to existing customers. How will you approach the problem as an analytics manager and what methods will to adopt to immediately arrest the customer attrition and gain lost customers in short to medium term.

How do you plan to measure customer satisfaction for your company? What are the questions that you will ask as part of a survey data collection?(Think about descriptive data collection)

If you ask the question to your customers “How Likely are you to recommend the product to a friend or a colleague”, what are the things you should keep in mind? (How will you define the promoters and detractors. Will you subtract the detractors from promoters?)

How would you justify a certain metric of customer satisfaction (think about finding a correlation with profitability and other managerial outcomes, comparing the correlation with alternative metrics)?

What are the pros and cons of using surveys vs other indicators like store purchase data (what customers are buying and when they are buying it) for measuring customer satisfaction?

How do you measure word of mouth dynamics from customers? As an analytics manager how would you approach collecting the word of mouth data from your customers? (for example how to capture data on who are your customers are talking to, how are the brands being mentioned and so on)

What are the advantages of using passive or unobtrusive ways of data collections from a customer analytics perspective?

Churn Related Problems

1. As a consultant you are trying to identify the customers of a telecom company who are most likely to churn. What data will you collect and what approach will you adopt.

2. On a similar line, you are tasked with identifying employees of an IT company who are most likely to move out to a different company. How will you approach the problem. (approach the problem with techniques such as survival analysis)

3. In retail banking, you are expected to develop a model for identifying the best customers who will be eligible for pre-approved loan based on customer profile attributes (apply techniques such as logistic regression)

Geographical/Location Based Analytics

A bit advanced and challenging but worth the effort – customising promotional messages and targeted promotional offers based on proximity of a customer to the store. For example, if you can track through a mobile GPS data that the target customer is parking near a shopping plaza defined within 100 mts of radius of your store, you can immediately send an enticing offer through SMS offering discount on a product if purchased within the next one hour

Customer and Product Affinity

1. You are tasked with automating product recommendations for any organization with a substantial product catalog and transaction volume to increasing competitiveness via product affinity. How will you approach this problem. (Think market basket analysis)

2. Determine frequently bought items together in a super market for purpose of cross-selling (Refer to the article on Association Rule Mining for a thorough understanding of such problems)

Forecasting

How will you apply time-series for forecasting of demand for products using ARIMA modeling

Application of Analytics in Marketing and Advertisement

1. Image analysis – advanced and challenging – translating images and other high dimensional data into numerical or symbols data to detecting events, surveillance, etc.

2. Sequence analysis – modeling a customer purchases as a sequence – customer first buys a computer, then speakers then webcam

3. Customer segmentation for targeted marketing (used clustering)

Applications in Finance- Fraud Analytics

1. How will you apply analytics in identifying credit card fraud (Anamoly / outlier detection in purchases data)

Pricing Analytics

1. In a airline flight booking operations how will you set decision rules about closing the slots for a particular price range

2. Hotel room bookings – when to say no rooms are available even when there are vacant rooms

Customer Analytics- Measuring Customer Experience and Retention of Customers

1. How will you derive sentiment analysis from product reviews, forum comments and tweets to identify possible point of discontent among customers

Unlocking Customer Data

You as a manager want to begin connecting to your customer data with the right tools and start analyzing your customers’ transactions. What will be your first step? (Ask if there is a data warehouse in place with a well-defined data dictionary)

Value of Data

Data is becoming a new source of value in large part because of what we termed its option value. Give few instances where you think data possesses option value for the business.

Earlier the emphasis was on companies that collected data. Now, the emphasis has shifted to companies that analyze data. In this context there are two types of companies at the different ends of the spectrum. Companies with data (like Twitter, Facebook, Reddit etc.) and companies with ideas to apply it in day to day business (consulting companies, technology vendors and analytics providers). How do you think these two can bridged to exploit hidden value? (think about companies holding involving consultancy firm to extract value from that data)

What changes do you think are likely to happen to the Big Data value chain in short-term (2-3 years) and medium term (5 years)?

How do you think companies that deal with data for one purpose can leverage its value through ancillary purposes? For example air ticket booking websites like Kayak or hotel booking platforms like Trivago has lots of data for customer preferences. How do you think they can leverage that outside their domain of expertise (which is selling air tickets and booking hotels).

Can you visualize a business model for a payment company (like Visa) that can forgo its fee payment and process transactions for free in return for access to more data and then use that data to derive and sell insights.

Site a few examples where a company can derive value from publicly available data. (for example flyontime started to gather open data to predict flight delays, Prismatic aggregates and ranks media contents across the web based on text analytics)

How value can be generated by cross-applying Big Data skills by collecting the data from multiple sources and using them to create an insight that can’t be generated by any of these standalone entities. (like for example, Climate Corporation collects environmental and other data to provide insights to farmers)

Data as Strategy

When Google collects any sort of data, it has secondary uses in mind. For example, GPS data collected from Google Street View and Google Maps ended up in its Training self-driving cars. Amazon on the other hand focuses on primary use of data and derives marginal secondary benefits from the huge data it collects. Give few example on how Amazon can use its data for secondary purposes to derive value. (its recommendation for example uses clickstream data but rarely uses it to predict the economy or other such things. Though it tracks underlined parts of Kindle books, it rarely shares that data with Authors, Publishers etc. to improve their products).

How data can be used to transform the business model of a car manufacturer and what sources would you look for to collect that data? (For example, Insurance companies collect enormous data based on accidents and collisions. By using telematics devices they also track the usage of different parts of a car. This data can be harnessed by the auto manufacturers themselves to improve the quality of their cars and even to reshape relationship with their parts suppliers )

Other Applications

1. How will you recommend the redesigning of store layouts based on data of customer purchase/ movement within the store