- Raw Data, the New Oil
- Top 10 Business Questions for Analytics
- Vital Lessons Learned
- Linking Analytics to Business
- Conclusion
This chapter is from the book
This chapter is from the book
This chapter is from the book
Top 10 Business Questions for Analytics
In my more than 15 years of direct involvement in applying analytics to business, I have found many important and difficult business questions that only analytics can answer. Today, the list is growing longer as more companies are investing in and applying analytics to new problems.
Although some of these questions can be answered from simply “slicing and dicing” raw data, some are so novel that many people in business might label them as “unknowable.” I will give you one such example; you can then come up with your own list of unknowable questions to see whether analytics can help answer them.
One example is to predict the actual wallets of individual customers—and by extension the business share of their wallets. It was commonly viewed by business as unknowable because you can see only the purchases your customers make at your store, not their purchases at other stores.
Other examples of unknowable questions include what online visitors do when they leave your sites, what your customers need when they can’t even describe it to themselves, and what to ship to the customers even before they start to consider buying and ordering. The former is the basis of the current ad exchange networks, in which third-party cookies are exchanged to let websites know where the visitors may have been before coming to your site. The last example is actually what Amazon is contemplating launching soon.5
Let’s look at the first example. With a partial view, the true wallet size each customer possesses is elusive and can be inferred only through surveys. However, such survey data is often unreliable and hard to extrapolate to other individual customers. However, as to be explained in more detail in Chapter 8, I have directly used the best customers’ transactions as “lenses” to predict wallets for both business-to-business (B2B) and business-to-customer (B2C) customers. These wallets were successfully tested and deployed in many business problems.
Here is my list of top 10 questions that I feel can best be answered by the use of analytics. The questions have been divided into four sections: Financial Management, Customer Management, HR Management, and Internal Operations.
How analytics can answer the questions might not be apparent at the moment. One of the aims of this book is that you will be able to either find the answers from the examples in this book or be able to customize your own BA workflows for specific industries or business situations. Once comfortable with modules and processes, you might want to go to the list of questions in Chapters 9 and 10 to practice putting together your own analytics workflows to answer the questions.
The following sections discuss my top 10 strategic questions and tips on how analytics can be applied to provide answers.
Financial Management
- Are your business and financial goals the right ones?
- For a business to set the right goals, it is best to start from the customers and the current market. Because the entire market is rarely addressable due to its product focus or competitive landscape, a business needs to define the markets in which it wants to play. Once decided, analytics can then be used to predict the size of the addressable market and combine with revenues to obtain the respective market share.
- Based on customer-level analytics, a business can predict what investment levels are needed to reach specific goals. Again, a business should start at the most granular levels and aggregate up to generate higher-level (business unit [BU] and enterprise) views. Some of the costs may involve realignment of business and product offerings.
- Where are the areas of major opportunities?
- The conventional top-down approach to market sizing is often not divisible at lower geographic or customer group levels. It has been a formidable exercise, often rife with contention when it was time for the BUs to assign annual sales targets at the various units based on the current market size predicted by the marketing intelligence (MI) team (at IBM during my time there) using the top-down approach. It felt like a prefab house being cut up and retrofitted—often messy and not a pretty sight.
- However, it is just the opposite when you start with the smallest building blocks. You can build anything provided that you have a plan, which is the case when the addressable market is produced from predicting individual customer wallets. Armed with the customers’ current spends and propensities to spend more, a business can find ways to more appropriately engage and entice different customers to shop more. Summing the wallets over the appropriate business units, the business can size the expected opportunities and identify the major revenue opportunities.
- Are your investments adequate?
- When the opportunities are identified, the natural next question is whether you are spending enough to win. Before knowing whether the investments are adequate, it is important first to predict the effective “levers” for increasing product wallet shares among the different groups of customers. Once the model results are validated with holdouts, the incremental causal effects of the levers need to be tested in-situ. A properly designed, multifactorial test-and-learn experiment can help determine the optimal combination of levers to give the maximum impacts. With the levers known, the business can then ascertain the costs of implementing the levers.
- With the costs and impacts of the various levers determined, the business can put together a pro forma simulator to model the level of investments needed to move the needle (that is, per-unit wallet share gain). Once the optimal return on investment (ROI) has been decided, the requisite costs then need to be checked against investments in the annual strategic plan to ensure that adequate funds have been allocated to achieve the goals.
Customer Management
- Do we know enough about our customers?
- The conventional way to know customers was to hear directly from the voice of the customers through focus groups and surveys. Unfortunately, focus group and survey results cannot be easily applied to each individual customer. They also take time to organize and are costly to run regularly.
- Given the richness of today’s customer data, analytics has been shown to be capable of generating detailed individual and actionable customer knowledge. This knowledge includes predicting individual customer’s propensities to buy from your brand and product, how much customers buy from you and from your competitors, and how valuable customers are to your business over their lifetime. Analytics can also perform customer multidimensional segmentation (behavior, demographic, attitudinal, and value) and develop segment persona. From the persona, the business can then devise effective strategies to better serve the segment, improve customer experience, and increase customer satisfaction and loyalty.
- Lapses can happen despite best efforts, so the business should constantly predict and monitor customers’ propensity to churn. The business should try to understand why they churn, what went wrong, and how to fix the problem. Any fixes should also be tested with control groups to make sure they are the right options. During and after implementation, those customers with high churning propensities should be surveyed to ensure that their issues were successfully resolved.
- All vital customer knowledge assets should be shared, managed, and leveraged for all marketing, sales, operations, product development, and finance initiatives.
- What actionable customer insights do you have?
- The chief marketing officer (CMO) of a well-known steakhouse chain client received a customer insight report taken from surveys from a top management consulting firm. It showed that the client has more than 15 million customers that like its food, but they rarely visit the restaurants. It was a great insight, but where and how can they reach and entice such customers to come and dine more frequently? They can deduce from surveys what these customers might look like and devise broadly targeted mass media ads to entice them. They can’t be sure whether the strategy would work across their chain when they test at a particular region or city. There are simply too many things at work to determine the effectiveness with sufficient certainty.
- To make such insights actionable with analytics, the individual diners’ receipts should be linked to the diners—either through a frequent diner program or by their name, credit card, or checking account number. The customer insights developed for the previous question can then be used to formulate campaigns and strategies to test and validate the effectiveness of the solutions derived from analytics insights. With a properly designed test-and-learn methodology, the incremental effects of levers can be ascertained before full-scale rollout. This analytics-driven approach would save costs and also reduce risk of failure.
- Are we focused on the right social and mobile issues?
- Social and mobile uses and data should be viewed as part of the data analytics value chain. Social and mobile are just part of the media. The customers’ social, online, and mobile behavior data must be linked to target behaviors in terms of purchases and conversions. Once integrated, the issues must be germane to the BA process, actionable, aligned with, and have significant impacts on the critical strategic goals.
- Is the critical knowledge from analytics properly managed (that is, captured, stored, shared, and reused) as an enterprise asset?
- As discussed in Chapter 4, analytics innovation and knowledge tend to occur at the key intersections in which different disciplines, roles, functions, and goals “collide.” To ensure that such knowledge can flourish and be captured, stored, shared, and reused as an enterprise asset, a process such as BAP (also see Chapter 4) should be adopted. Beyond the process, key “analytics deciders” who will be defined later must be present at the intersections. For now, an analytics decider is defined as someone who is highly proficient in both business acumen and analytics. A knowledge management IT system and a functioning “analytics sandbox” support the analytics deciders with well-defined rules, ensuring a safe and collaborative environment for creative and productive collisions.
HR Management
- Do you have the right strategy for recruiting, managing, and retaining analytics talent?
- In my years spent in building and managing analytics teams for big and small companies, I rank organizational and personnel issues as the number-one cause of failure in applying analytics. Many efforts failed because the analytics had either a limiting or conflicting reporting structure; the wrong compositions of talents and skills in the business, data, modeling, and strategy functions; or simply the wrong people (nonanalytics deciders) in key leadership roles within the intersections. Alternatively, companies’ analytics efforts failed as they tried to remove the collisions at the intersections by delineating clear lines of responsibility. Unknowingly, these intersections were eliminated as everyone tried hard to stay within their own areas of responsibilities and roles. (Chapter 11 addresses some of these organizational and personnel issues.)
Internal Operations
- Are your business processes driven with insights from predictive customer analytics?
- Although companies started to use more predictive customer analytics in the past decade, many did not progress beyond tactical campaign targeting. As a result, the effectiveness of analytics can deteriorate over time. Sadly, many pioneers in analytics applications have been stuck doing the same things or have relegated analytics to just one of the support functions. More progressive companies have quietly embedded predictive analytics in their various business processes, and their CEOs rely on analytics for answers to their strategic questions.
- For example, during a benchmarking project of companies’ analytics competency, I was told that the CEO of one of the major online retailers, while preparing the annual report to the Wall Street analysts, discovered that revenues went up but the number of visitors went down. He wondered whether he should pose it as a positive or negative indication. Instead of consulting with the CFO or anyone within his senior leadership, he instead asked the director of analytics to come up with an answer. The CEO was satisfied with the reply and used the analytics recommendations during the call with the analysts.
- I firmly believe that in the next few years, companies will win or lose by how much they can integrate and embed business analytics into their various business processes (as explained in Chapter 6).
- Are your sales efforts fueled by analytics insights?
- One of my earliest successes in applying analytics was not with marketing efforts, even though my team was part of the IBM MI team; it was to help IBM sales efforts. Since then, I find that sales efforts can be greatly enhanced by analytics insights such as leads generation, leads prioritization, sales force optimization, and telesales/call center performance analytics.