Everyone loves stories to which they can relate, which probably makes it the ideal way to conclude this book. While stories can be fun and funny, the most valuable stories are those that motivate us to think differently and take action, where the story is so compelling that the reader can't wait to put the ideas into action!
The goals of this chapter are to share some big data stories and to help you, the reader, develop inspiring stories that are relevant to your organization and motivate the organization into action.
Instead of providing a long list of the different analytics that are occurring within different industries, I'm offering a “think differently” approach for how you find and construct big data stories that are the most relevant to your organization. Instead of looking at the big data stories from the traditional industries perspective, let's look at stories from the perspective of the organization's strategic nouns, or key business entities. I find that most big data and data science stories fall into three categories of business entity analytics (regardless of industry):
Customer and employee analytics
Product and device analytics
Network and operational analytics
The advantage of looking for stories across these three categories is that it prevents organizations from artificially limiting themselves in searching for relevant big data stories. Many organizations are only interested in hearing about big data stories that are happening within their industry. That's the “safe” way to go. But sometimes the most powerful opportunities are realized from stories from other industries. Having a broader view of these big data stories can open the eyes of the business executives as to the potential of big data within their organizations.
For example, digital media organizations use “attribution analytics” to quantify the impact of different digital media treatments (messaging, websites, impressions, display ad type, display ad page location, keyword searches, social media posts, day parting, etc.) on a conversion or sales event. Think about how many different websites, display ads, and keyword searches you interact with as you decide to do something (e.g., buy a product, request some collateral, download an article, play a game, research an event, etc.). Attribution analysis looks at “baskets” of digital media treatments and activities that lead to particular conversion events across a large number of visitors and creates complex data enrichment calculations (frequency, recency, and sequencing of marketing treatments) in order to attribute sales credit to these different digital media treatments. Think “hockey assist” as in trying to measure the impact that a wide variety of digital media treatments had over a period of time to drive a conversion or sales event.1 Following is an example of how organizations use attribution analysis to maximize campaign return on marketing investment (ROMI):
Digital media attribution analysis
1. Track Activities Leading to Conversion Events. Create market baskets of keyword searches, site visits, display impressions, display clicks, and other media treatments associated with each conversion event
2. Enrich Data to Create New Metrics to Understand Drivers of Visitor Behaviors. Create metrics around frequencies, ordering, sequencing, and latencies
3. Analyze Metrics to Quantify Cause and Effect. Identify commonalities in baskets, calculate correlations and strength of correlations, and build “conversion path” models
4. Operationalize Actionable Insights. Operationalize insights into media planning and buying systems, and guide in-flight campaign execution
That same attribution analytics would work perfectly in the area of health care where physicians, nurses, and other caregivers are trying to determine or attribute the impact on a patient's wellness across a wide variety of health care “treatments” including medications, surgery, supplements, therapy, diet, exercise, sleep, stress, religion, consoling, and many other health-impacting variables. Using the digital media attribution analytics, health care organizations could determine which combinations, frequency, recency, and sequencing of health care treatments are most effective for which types of patients in what types of wellness situations. But if health care organizations only look within their own industry, they are likely to miss opportunities to learn from other industries' analytic stories and miss the opportunity to apply those stories to optimize their own key business processes, uncover new monetization opportunities, and gain a competitive edge within their industry.
These three business entity analytics buckets will help you see that the use case type is more relevant than the industry from which it came; that it provides a “think differently” moment to borrow analytic best practices from other industries. Let's discuss each of these three categories in more detail to see what stories you might uncover that could be meaningful to your organization:
Customer and employee analytics
Product and device analytics
Network and operational analytics
Customer and Employee Analytics
For organizations in business-to-consumer (B2C) industries, understanding and taking care of customers is job #1. Understanding in detail the propensities, tendencies, patterns, interests, passions, affiliations, and associations of each of your individual customers is key to increasing revenue, reducing costs, mitigating risks, and improving margins and profits.
Customers can take many forms including visitors, passengers, travelers, guests, lodgers, patients, students, clients, residents, citizens, constituents, prisoners, players, and more. Many B2C industries can benefit directly from data and analytics that yield superior insights into the behaviors of their customers including:
Retail
Restaurants
Travel and hospitality
Airlines
Automotive
Gaming
Entertainment
Banking
Credit cards
Financial services
Health care
Insurance
Media
Telecommunications
Consumer electronics (e.g., computers, tablets, digital cameras, digital media players, GPS devices)
Primary and higher education
Utilities
Oil and gas
Public service agencies
Government agencies
The foundation of customer analytics is identifying, quantifying, and predicting the individual customer's behavioral characteristics (propensities, tendencies, patterns, trends, interests, passions, associations, and affiliations) to identify opportunities to engage the customer to influence his or her behaviors. Some call this “catching the customer in the act.” The more timely the identification of these customer interactions, the better the chances of uncovering new revenue or monetization opportunities. Customer analytics include the following:2
Customer acquisition measures the effectiveness of different sales and marketing techniques to get customers to sample or trial your product or service.
Customer activation measures the effectiveness of different sales and marketing techniques to get customers to regularly use and/or pay for your product or service.
Customer cross-sell and up-sell measures the effectiveness of different sales, marketing, and merchandising techniques to get customers to upgrade the products and services that they already use or buy and/or get customers to use or buy complementary products and services.
Customer retention measures the effectiveness of sales, marketing, and customer service treatments to identify customers likely to attrite and the subsequent efforts to retain those customers.
Customer sentiment monitors the sentiment of customers across multiple social media sites, blogs, consumer comments, and e-mail conversations to flag product, service, or operational problem areas and recommend corrective action.
Customer advocacy measures how effective particular customers are at influencing other customers' actions or behaviors.
Customer lifetime value determines the current (and future or maximum) value of a particular customer.
Customer fraud monitors and flags potential fraudulent activities in real-time in order to recommend timely corrective or preventive action.
Cohort analysis determines the impact that one particular customer has on other customers in driving particular customer and/or group behaviors.
There is also a set of customer analytics around marketing. These marketing analytics include:
Targeting effectiveness measures the effectiveness of marketing's targeting efforts to reach the “right” or highest qualified prospects.
Re-targeting effectiveness measures the effectiveness of re-targeting efforts to re-target prospects that have shown an interest in a particular product or service.
Segmentation effectiveness measures the effectiveness of segmentation efforts to identify high-value prospect clusters.
Campaign marketing effectiveness measures the effectiveness of general marketing campaigns at driving customer or prospect actions.
Direct marketing effectiveness measures the effectiveness of direct-to-consumer marketing campaigns to get customers to respond to marketing requests or buy particular products or services.
Promotional effectiveness measures the effectiveness of channel or partner promotional activities, events, packages, and offers.
A/B testing tests the effectiveness of two different marketing treatments (messaging, ad types, websites, keywords, day part, and page location) to determine which marketing treatment is most effective in driving the desired customer action or behavior.
Market basket analysis determines the propensity of products or services to sell in combination with other products and services (within same basket or shopping cart). Market basket analysis also can identify time lags between purchase events (buy a boat and then two weeks later, buy water skis).
Attribution analysis quantifies the contribution of different digital marketing or media treatments in driving a customer event or activity (e.g., buy a product, download an app, play a game, request collateral, research an event).
Omni-channel marketing analysis quantifies the inter-play of marketing effectiveness across multiple retail or business channels (e.g., physical store, catalog, call center, website, social media) in driving sales results.
Trade promotion effectiveness measures the effectiveness of channel or partner promotions to drive end consumer sales.
Pricing and yield optimization determines both the timing and the “optimal” prices in order to maximize revenue and profitability for perishable products or services (vegetables, meat, airline seats, hotel rooms, sporting events, concerts).
Markdown management optimization determines the timing and amount of price reduction and promotions to reduce obsolete and excess inventory while balancing revenue, margin, and cost variables.
By the way, many of these customer analytics have a corollary for employee analytics (teachers, police officers, parole officers, case workers, physician, nurses, technicians, mechanics, pilots, drivers, entertainers, etc.). These analytics include:
Employee acquisition (hiring) measures the effectiveness of different hiring practices and recruiting personnel to identify and hire the most productive and successful employees.
Employee activation (productivity or performance) measures the effectiveness of training programs and managers to engage employees and drive more productive and effective performance.
Employee development (promotions, firing) measures the effectiveness of reviews, promotions, training, coaching, interventions, and management to identify and promote high potential employees and release low productivity employees at the lowest cost and lowest risk.
Employee retention measures the effectiveness of promotions, raises, awards, stock options, etc. to retain the organization's most valuable and productive employees.
Employee advocacy (hiring referrals) measures the effectiveness of advocacy and referral programs to acquire high potential job candidates.
Employee lifetime value determines or scores the current (and future or maximum) value of employees to the organization.
Employee sentiment (employee satisfaction, “best places to work” surveys, etc.) identifies, measures, and recommends corrective action on the drivers of employee and departmental dissatisfaction.
Employee fraud (shrinkage) monitors and flags shrinkage problems and triages those situations to identify root causes of fraud and shrinkage.
It can be useful to look at what other organizations in other industries are doing to better understand their customers and employees. For example, your organization could identify which organizations are best at leveraging customer loyalty programs to drive customer acquisition, maturation, retention, and advocacy. Then identify what data they are capturing about their customers and what analytics they are leveraging to improve the customer experience. There are many examples of organizations that understand how to optimize their loyalty programs. Just go grab a venti non-fat, no water chai latte at a certain coffee chain to experience that for yourself.
Product and Device Analytics
The second area of business entity analytics focuses on physical items—products and machines. Many of the same behavioral analytic basics that are used in customer analytics are applicable for products and machines. Like humans, products and machines exhibit different behavioral tendencies, especially over time. Two wind turbines manufactured by the same manufacturer, installed at the same time, and located in the same cornfield could develop very different behaviors and tendencies over time due to usage, maintenance, upgrades, and general product wear and tear.
Analytics about products and machines (airplanes, jet engines, cars, delivery trucks, locomotives, ATMs, washing machines, routers, traffic lights, wind turbines, power plants, etc.) could include any of the following:
Predictive maintenance predicts when certain products or devices are in need of maintenance, what sort of maintenance, the likely maintenance and replacement materials, and technician skill sets.
Maintenance scheduling optimization optimizes the scheduling of resources (technicians with the right skill sets, replacement parts, maintenance equipment, etc.) in order to optimize the replacement and/or upgrading of failing or under-performing parts or products.
Maintenance, repair, and operations (MRO) inventory optimization balances MRO inventory with predicted maintenance needs in order to reduce inventory costs and minimize obsolete and excessive inventory.
Product performance optimization optimizes product performance and mean time between maintenance (MTBM) by understanding the product's or device's optimal operation performance ranges, tolerances, and variances.
Manufacturing effectiveness reduces manufacturing costs while maintaining product quality levels and production schedules through the optimal mix of supplies, suppliers, and in-house and contract manufacturing capabilities.
Supplier performance analytics quantify supplier product quality and delivery reliability in order to minimize manufacturing line downtown.
Supplier decommits/recommits analytics understand optimal production capacities of suppliers and contract manufacturers in order to properly rebalance manufacturing needs caused by supply chain disruptions (strikes, storms, wars, raw material shortages).
Supplier network analytics triage product and supplier problems more quickly by understanding the dynamics of the underlying supplier and contract manufacturer relationships and inter-dependencies.
Product testing and QA effectiveness accelerates product quality assurance testing by optimizing the tests and/or combinations of tests that cause products, components, suppliers, and contract manufacturers to fail more quickly.
Supply chain optimization optimizes supply chain delivery and inventory levels while minimizing supply chain costs and risks associated with obsolete and excess inventory.
Optimize MRO parts inventory to determine the appropriate level of MRO parts inventory based on predicted maintenance needs.
New product introductions optimize product and marketing mix to increase the probability of success when launching new products, product extensions, and/or new product versions.
Product rationalization/retirement determines which products to divest or retire, and when, based on that product's impact on customer value and inter-related profitability of other products (market basket analysis).
Brand and category management analysis determines optimal pricing, packaging, placement, and promotional variables of individual brands and products within brands to drive overall brand and category revenues, profitability, and market share.
Product-centric industries most impacted by product and device analytics include:
Consumer packaged goods
High-tech manufacturing
Appliance and electronics manufacturing
Sporting goods manufacturing
Food and beverage
Automotive
Agriculture
Farm machinery manufacturing
Heavy equipment manufacturing
Pharmaceuticals
Financial services
Banking
Credit cards
Insurance
Network and Operational Analytics
The third area of business entity analytics focuses on network and operational analytics. The “internet of things” (IoT) and wearable computing (Fitbit, Jawbone, Garmin) has increased the level of interest (and the volume and variety of data) about what is happening across vast and complex human and machine/device networks. More than ever, we are an interconnected world where the actions of one person or device in a social or physical network can have a “butterfly effect” on all of the people and devices across that network.3
Networks can take many different shapes and forms including ATM networks, retail branches, supplier networks, device sensors, in-store beacons, mobile devices, cellular towers, traffic lights, slot machines, and communication networks.
Analytics about networks and operations could include any of the following:
Demand forecasting forecasts network demand (average demand, surge demand, minimal viable demand) based on predicted network usage behaviors, patterns, and trends.
Capacity planning predicts network capacity requirements in all potential (what if) working situations.
Reduce unplanned downtime to identify, monitor, and pre-emptively predict the failure of the drivers of unplanned network downtime.
Network performance optimization predicts and optimizes network performance across multiple usage scenarios (network traffic, weather, seasonality, holidays, special events) in real-time.
Network layout optimization optimizes network layout in order to minimize traffic bottlenecks and optimize network bandwidth and throughput.
Reduce network traffic to triage network traffic bottlenecks and provide real-time incentives and/or governors to reduce or re-route traffic during overload situations.
Load balancing identifies and rebalances network traffic based on current and forecasted traffic needs and current network capacity.
Theft and revenue protection identifies, understands, and recommends the most appropriate revenue protection actions based on theft situations across the network.
Predictive maintenance predicts when network nodes are in need of maintenance, what sort of maintenance, the likely maintenance and replacement materials, and technician skill sets.
Network security identifies, understands, and recommends the most appropriate actions based on unauthorized network or device/node entry or usage situations across the network.
Industries most impacted by network and operational analytics tend to be industries that run or manage complex projects or systems. These industries have to coordinate multiple vendors and suppliers across multiple sub-assemblies or sub-projects in order to deliver the end product or project on time and within budget. Some of these industries include:
Large-scale construction (skyscrapers, malls, stadiums, airports, dams, bridges, tunnels, etc.)
Airplane manufacturing
Shipbuilding
Defense contractors
Systems integrators
Telecommunication networks
Railroad networks
Transportation networks
There are many, many more examples of customer, product, and network analytics. The list above is a good starter point. And while investigating analytic use cases within your own industry is “safe,” better and potentially more impactful analytic use cases can likely be found by looking for customer, product, and network analytic success stories in other industries. Bucketing the analytic use cases into those three categories helps the reader to contemplate a wider variety of analytic opportunities and best practices across different industries.
Think differently when you are in search of the analytics that may be most impactful to your organization. Don't assume that your industry has all the answers.
Characteristics of a Good Business Story
The final step in the book is to pull together the “thinking like a data scientist” results and the sample analytics to create a story that is interesting and relevant to your organization. While it can be useful to hear about what other organizations are doing with big data and data science, the most compelling stories will be those stories about your organization that motivate your senior leadership to take action.
You know from reading books and watching movies that the best stories have interesting characters that have been put into a difficult situation. Heck, that sounds like data science already. To create compelling stories, you are going to need the following components to create an interesting and relevant story that is unique to your organization (think about the process in relationship to your favorite science fiction adventure movie):
Key business initiative (survival of the human race)
Strategic nouns or key business entities (pilots, scientists, aliens)
Current challenging situation (aliens are going to conquer Earth and exterminate the human race)
Creative solution (infect the alien ships with a computer virus that shuts down their defensive shields)
Desired glorious end state (aliens get their butts kicked, and the whole world becomes one united brotherhood)
Let's see this process in action:
Let's say that your organization has as a key business initiative to “reduce customer churn by 10 percent over the next 12 months.”
Your strategic noun is “customer.”
The current challenging situation is “too many of our most valuable customers are leaving the company and going to competitors.”
The creative solution is “developing analytics that flag customers who have a high propensity to leave the company, create a customer lifetime value score for each customer (so that your organization is not wasting valuable sales and marketing resources saving the ‘wrong’ customers), and deliver messages to frontline employees (call center reps, sales teams, partners) with recommended offers to deliver to the customer if a valuable customer has a score with an ‘at risk’ propensity to leave.”
The glorious end state is “dramatic increase in the retention of the organization's most valuable customers that leads to an increase in corporate profits, an increase in customer satisfaction, and generous raises for all!”
This is an easy process if you understand your organization's key business initiatives or what's important to the organization's business leadership.
Summary
Broaden your horizons with respect to looking for analytic use cases. Instead of just looking within your own industry, look across different industries for analytic use cases around:
Customer and employee analytics
Product and device analytics
Network and operational analytics
Since this is the last chapter of the book, put a cherry on the top of your Big Data MBA by developing a compelling and relevant story that you can share within your organization to motivate senior leadership to action. Make the story compelling by tying one of the above analytic use cases to your organization's key business initiatives, and make the story relevant by leveraging your “thinking like a data scientist” training. That way you ensure that all the work you have put into reading this book and doing the homework can lead to something of compelling and differentiated value to the organization. And heck, maybe you will get a promotion out of it!
Congratulations! For a special surprise, go to this URL: www.wiley.com/go/bigdatamba. And don't share this URL with anyone else. Make other folks read the entire book to find this “Easter egg” surprise.
Now you have earned your Big Data MBA! Go get 'em!
Homework Assignment
Use the following exercises to apply what you learned in this chapter.
Exercise #1: Identify one of your organization's key business initiatives.
Exercise #2: Apply the “thinking like a data scientist” approach to identify the relevant business stakeholders, key business entities or strategic nouns, key decisions, potential recommendations, and supporting scores.
Exercise #3: Now create a story that weaves together all of these items with a relevant analytics example that can help senior leadership to understand the business potential and motivate them into action. Use your strategic nouns to help you find some relevant analytic use cases outlined in this chapter.