Seven Questions To Ask Before You Invest In Data

​

 

The importance of data as a competitive strategy tool is nothing new.  Two hundred years ago Napoleon said:  “War is information.” Napoleon’s military genius drove him to learn as much as possible about troop position, enemy strength, topography, weather and any other factor that could influence the outcome of a battle.  So the value of information has been known for centuries, but no one talked about “Big Data” during the Napoleonic era. What has changed?  

 

Several things:  First, the volume of data being generated in our increasingly digital world is enormous and growing at truly exponential rates. IBM reports that 2.5 quintillion bytes of data are generated every day – to the extent that 90% of the data on the planet today did not exist 2 years ago.  IDC projects the total volume of data available will increase 50-fold in the decade ending 2020.  

 

What is driving this tremendous growth in data? Forbes recently described trends fueling the explosion of data: (http://www.forbes.com/sites/bernardmarr/2015/09/30/big-data-20-mind-boggling-facts-everyone-must-read/#36092f166c1d) 

 

Here are several examples of major data generators:

 

➢    Google responds to 40,000 search queries every second.  

 

➢    Facebook users send more than 30 million messages every minute.

 

➢    Last year more than one trillion photos were taken, and 80% of them were taken on smart phones.

 

➢    The number of sensors in Internet connected devices is expected to top one trillion by 2020.

 

Second, the cost of storing data has declined dramatically, as you would expect given the growth in volume. Mkomo.com charted the drop in the cost of data storage over the past 35 years.  In 1981, a 5.0 MB drive cost $3,500 or roughly $700,000 per GB.  By 2014 you could purchase a 3.0 TB drive for $99.99, or $0.03/GB. How’s that for a 99.999996% cost reduction!  And, the cost of data storage should continue to decline sharply as storage increasingly moves to the cloud, taking advantage of the massive scale of players such as Amazon, Oracle, Google and Apple. 

 

Third, the tools required to process and analyze this growing mountain of data have become much more effective and affordable than earlier iterations.  This point is worth emphasizing, because many veteran marketing managers in financial services had less than favorable experiences with early CRM systems implemented in the ‘90s and ‘00s.  In fact, the bad experience was not confined to financial services. Bain & Company recently estimated that 90% of first generation CRM systems ended up as a complete write-off.

 

Why were early attempts to integrate sophisticated data analytics into marketing and customer relationship management so ineffective? Many factors contributed to the first-gen CRM debacle, but very high on the list is the fact that many organizations jumped into the deep end of the CRM pool without having any clearly defined objectives for what they expected the systems to produce. 

 

How do we avoid the same trap twenty years on?  One approach would be to avoid data analytics altogether, and simply assume the way we’ve processed information for the past 50 years will serve us well into the future. That approach is short sighted at best and foolhardy at worst.  Remember Napoleon’s words:  “War is information.”  Whether you choose to utilize advanced data analytics or not, your competitors almost certainly will. Failure to advance your data strategy could mean that within a few years your competition will know more about your customers than you – not an enviable position, and certainly not strategic.  

 

Also, the opportunity cost of not utilizing data in financial services is huge because the banking sector (including credit unions) possesses more data than any other industry except communications and media.  The applications of data within financial services are virtually limitless. IBM has developed a short list of applications including: customer insight, customer experience management, channel execution, business strategy, risk management and marketing – to name just a few.

 

An equally bad approach would be to jump at the first data tool or vendor that promises to bring the miracle of “Big Data” to your organization, and there are many preaching just that in the market today.  Unfortunately, there is no “one size fits all” solution when it comes to applying data analytics to financial services, or any other sector for that matter.  Success in implementing data solutions requires careful thought and rigorous analysis of the objectives and capabilities of your organization. 

                  

A good starting point is to answer two fundamental questions:  First, how does the use of data fit with the mission of your organization? Second, what outcomes are desired at the end of the process?  For example, one primary objective could be profit maximization.  A very different objective could be the maximization of member value.  Data can help an organization optimize its actions to achieve virtually any objective, but first you must define what those objectives are.  The Navy has a saying:  “No wind favors him who has no set port.”  Don’t expect data to tell you what your objectives should be. That is up to you.  

 

Let’s assume you have answered the a priori questions about what you want a data program to accomplish. At the next level we have identified seven key questions that are useful in choosing the best data solution for your organization.  Here is the list:

 

1)    What questions are we trying to answer with more data?   

 

Questions at this level can be strategic or tactical.  It is helpful to develop a list of twenty marketing or operational questions you would like to answer. You may be able to answer many of these questions today without the use of a new system or process.  Others may require new tools. But the process of posing the questions will help determine what type of system or process you need.  Here are some examples:

 

➢    What factors could help us predict in advance that a member will leave our credit union at a later date?

 

➢    How does member profitability vary relative to their home distance from our nearest branch?

 

➢    What factor(s) determine whether a member views us as their primary financial institution (PFI)?

 

➢    Which of our members will have the highest propensity to seek a mortgage loan in the next twelve months?

 

➢    How do channel preferences vary between age groups and other demographic segments?

 

2) What type of data strategy best fits our needs?

  

Data analytics programs can range from very simple to highly complex.  A “simple” use of data would be an events-based marketing campaign in which a special offer is presented to a customer at an anniversary or birthday date.  

 

Somewhat more complex would be a rules-based campaign based upon a set of logical rules.  For example, a credit union marketing manager might determine that a member’s linkage of their checking account to a Target Red Card and a PayPal account means that member views the credit union as their primary financial institution.  Those “PFI” members could then receive special offers per the “rules” of the campaign management design.

 

Even more complex would be a true machine based learning approach (also often referred to as AI) in which a computer using proprietary algorithms chews through massive amounts of data seeking to identify patterns and correlations that might be beyond the capacity of a human mind to process. This latter approach is closest to what has commonly become known as “Big Data Marketing Analytics”.  

 

However, depending upon the needs and capabilities of your organization, big data is not necessarily better or more practicable than a simple data application.  In any data strategy, you need to accomplish four tasks:

 

➢    Gather the relevant data in a usable fashion (see #3 below)

 

➢    Analyze the data to identify valuable patterns, trends and correlations

 

➢    Connect the data analytics output to an appropriate marketing campaign management tool

 

➢    Analyze the results of the marketing programs themselves to determine the effectiveness of the various marketing treatments and, in the case of a big data application, allow the machine learning to take place

 

Each of the above three steps needs to be consistent with the other two.  A world-class analytics capability might be wasted if the organization lacks the campaign management capabilities to put the analytics to good use.  It is critical to carefully evaluate your organization’s capabilities from end to end before investing heavily in any one aspect of a data roadmap.

 

3) Do we first need to build a data warehouse?

 

Let’s first describe what a data warehouse is.  Wikipedia says data warehouses are “central repositories of integrated data from one or more disparate sources.  They store current and historical data and are used for creating analytical reports for knowledge workers throughout the enterprise.”  As such, a data warehouse is more than just a database. StackOverflow says that a data warehouse must contain “the means to retrieve and analyze data, to extract, transform and load data, and to maintain the data dictionary.”

 

Before we get too technical, let’s get back to our question.  Do we need to build a data warehouse to gather data from various internal (e.g. our core data processor) and external (e.g. credit bureau) sources to aggregate, cleanse and normalize data before we can begin to analyze it? A few years ago many financial institutions would have answered yes to this question. Now, however, we are seeing new data analytics tools that perform the data warehousing functions themselves.  In fact, some state of the art analytics tools will intentionally bypass a data warehouse preferring to access the data as close to its source as possible. This latter approach saves significant dollars and time to market.

 

Again, the key is to consider a full end-to-end approach to the data roadmap before constructing the individual components.    

 

4) Who should own data within our organization?     

 

Ultimately, every senior leader in the organization should be a stakeholder in the data implementation strategy, just as every part of the business should benefit from the use of data.  But that does not mean that everyone must own the data process.  While a data task force can be extremely useful, I am convinced that one senior leader in the organization should be tasked with driving the data strategy. 

 

I’ve seen some organizations in which Marketing owns data.  In others data is the primary domain of the Chief Information Officer.  Occasionally the CFO might be the primary driver.  And rarely the CEO takes the lead directly.  The answer to the question of who owns data will vary according to the skills of individual leaders and the overall needs of the organization, but appointing one individual as the owner establishes accountability and increases the likelihood of sustained progress in the exploration and implementation of the data plan.

 

5) Should we build, buy or partner to develop data analytics tools?   

 

The state of the art is advancing very rapidly in the data analytics field, and the playing field is already crowded.  A recent report by Raab Associates identified 140 companies competing in the machine based learning sector across 23 different product categories.

 

In my view, very few credit unions possess the scale and technical capabilities required to build their own data systems.  Even if it were feasible, given the availability of so many robust systems in the market today, why not take advantage of expert teams whose sole focus is the development and continued enhancement of data analytics technology?

  

Increasingly, many of the best data tools are being made available on a SaaS basis, requiring limited front-end investment and relatively minimal implementation time.  This approach means you are always in position to take advantage of the latest and greatest advances.  

 

6) Do we need to have a data analyst on staff?  

 

The answer to this question will depend upon the processes and tools you choose to implement.  Be very careful to obtain a clear answer to this question before you embark on purchasing a data system, because experienced data analysts are in short supply and can be very expensive to hire and retain.  One of my clients offered $300K plus incentives to a skilled data analyst, only to have that offer doubled by one of the major casino operators.  The gaming industry, by the way, is a heavy user of data analytics in the quest to find new ways to keep players at the table as long as possible!

 

7) What are the ethical and regulatory limits to the use of data?

 

The legal and regulatory limits to the use of customer data in financial services continue to evolve rapidly and need to be considered carefully prior to implementing a data strategy.  Equally important, although perhaps less well defined, is the consideration of whether an organization should self-impose certain ethical limits to the use of data.  

 

For example, should a credit union set limits regarding when member communications driven by data analytics might be considered too intrusive or an invasion of privacy?  Should a member have the ability to opt out of any marketing communications?  Should certain events-based marketing concepts simply be off-limits?  

These are not easy questions to answer, but a data task force within the organization should give them careful consideration early in the data process. Several high profile retailers including Target and Wal-Mart have experienced blowback from publicized incidents in which the use of data arguably went too far.

 

CONCLUSION

 

The potential benefits of a well-constructed data strategy are immense, but the task of deciding the best path forward can seem daunting.  Taking time to answer the seven questions above can be an excellent first step in moving your organization towards a deeper level of member centricity.  You may be very pleasantly surprised at the powerful insights even a rudimentary data program can generate.  

 

Winston Churchill may have captured it best:  “Out of intense complexities, intense simplicities emerge.”  And now we can let the machine do the sorting.