Quantcast
Channel: Blog | Dell
Viewing all articles
Browse latest Browse all 17822

The Next Frontier: Solution Engineering – Part I

$
0
0
EMC logo

The trouble with big data is that there is no one single, shiny solution.  One can’t just install Hadoop or predictive analytics or an appliance, and say that they have a big data solution.  Our industry has struggled with this dilemma before, as data warehousing and business intelligence technologies sought relevance within organizations over the past 10 to 15 years.  To be successful with big data and advanced analytics – like its brethren data warehousing and business intelligence before it – requires a new “engineering” skill – solution engineering.

There’s engineering for many disciplines – systems engineering, electrical engineering, mechanical engineering – so why not solution engineering?  Solution Engineering would be defined as:

“A process for identifying and decomposing an organization’s key business initiatives into its business enabling capabilities and supporting technology components in order to support an organization’s decision-making and data monetization efforts.”

Lego’s Metaphor

Surprisingly, the Solution Engineering process is similar to playing with LEGO bricks.  The most successful LEGO projects are those that have an end in mind – a thoroughly defined, well-scoped solution.  Do I want to build the pirate ship, the castle, or the space ship?  With LEGO bricks, I can build all three, plus many, many more.  However, each “solution” requires a different set of bricks in different configurations and with a different set of instructions.  Much like LEGO bricks, it is critical to your big data business success to identify upfront what solution your organization is trying to build, and then to assemble the right data and technology capabilities, in the right order, to deliver a successful.

Solutions Engineering Process

Here is a Solution Engineering process for architecting and developing data and analytics-enabled business solutions.  This process requires an upfront investment to grasp how your organization “makes money,” to understand the organization’s strategic nouns and how they power your organization’s value creation process, and fathom the organizations key business initiatives[1] around which to focus your Solution Engineering efforts.

Step 1:  Understand How The Organization Makes Money.  Pretend that you are the General Manager of the organization, and contemplate how the organization can “make more money.”  For example, what can the organization do to increase revenues, decrease costs, reduce risks, or increase compliance?

There are many levers that organizations can tune in order to make more money.  Increasing revenue, for example, can include initiatives such as increase the number of premium or “gold card” customers, increase store or site traffic, reduce customer churn, increase revenue per shopping occurrence, increase private label sales as percentage of market basket, increase cross-sell/cross-sell effectiveness, and optimize promotional effectiveness (see chart below).

Figure 1: Potential Areas for Making More Money

Next, spend the time to identify and understand your organization’s strategic nouns and identify how those nouns drive the “money-making” capabilities of the organization.  For example if you’re in the airline industry, hubs are a very important “noun” of your business, and any way that you can increase the number of flights per hub (e.g., decrease airplane turn-times, improve terminal/ramp efficiencies) means more flights per day, which equals more money.

Invest some time actually using your organization’s products or services and/or observing or using competitive products.  Experience first-hand how the organization works, their value propositions to their customers and partners, and how their money-making efforts work.

For example, if your organization is in the Business-to-Consumer (B2C) market, you can leverage customer engagement data (consumer comments, emails, social media, blogs) to uncover insights that can help to optimize the customer engagement process (profiling, segmentation, targeting, acquisition, activation, maturation, retention, advocacy) in order to create more “profitable” customers (see chart below).

Figure 2:  Customer Profitability Distribution

Step 2:  Identify Your Organization’s Key Business Initiatives.  The next step is to do some primary research to understand your organization’s key business initiatives.  This includes reading the annual report, listening to analyst calls, and searching for recent executive management speeches and presentations.  If possible, interview senior business management to understand their top business opportunities (as well as their perceptions to the key challenges that might prevent the organization from successfully executing against their top business opportunities).  Check out my blog “Big Data MBA: Reading the Annual Report for Big Data Opportunities” for some clues as to what to look for in reviewing an organization’s annual report.For each identified business initiative, capture key information such as business stakeholders (roles and responsibilities), key performance indicators, and the metrics against which success of the business initiative will be measured, timeframe or roadmap for delivery, critical success factors, desired outcomes, and key tasks.  Check out my blog “Most Excellent Big Data Strategy Document” for a process for decomposing an organization’s business strategy into its key business initiatives.

Step 3:  Brainstorm Big Data Business Impact.  The next step in the Solution Engineering process is to brainstorm how big data and advanced analytics can impact the targeted business initiative.  Big data and advanced analytics can power an organization’s key business initiatives in five ways:

  1. Provide access to more detailed structured data (at the lowest level of transaction granularity), which enables more granular, more detailed decisions.  For example, detailed structured data (like customer loyalty transactions) facilitates decision-making and data monetization opportunities at the individual customer, seasonal/holiday, and local levels.
  2. Provide access to new unstructured data sources (both internal unstructured data sources like web logs, consumer comments, and emails, as well as external unstructured data sources like social media and mobile) that enable more robust, more complete decisions.  These new, diverse data sources provide new variables, metrics, and dimensions that can be integrated into analytic models to yield actionable, material business insights and recommendations.
  3. Access to machine or sensor-generated data (smart grids, connected cars, smart appliances), which enables more timely operational decisions including the ability to support predictive maintenance.  I would also put smartphone-generated location data into this category which enables location-based analysis that can drive real-time consumer offers and engagement recommendations.
  4. Provide high-velocity/low-latency data access (where we’ve reduced the time delay between the data event and the analysis of that data), which enables more frequent, more-timely decisions and data monetization.  This could include the creation of on-demand customer segments (based upon the results of some major event like the Super Bowl) as well as real-time location-based insights from real-time smartphone data.
  5. Finally, enable predictive analytics, which provides new opportunities to uncover causality (cause and effect) buried in the data and exploit experimentation when dealing with customers.  Predictive analytics can enable a different mindset with your business stakeholders, encouraging them to use new “verbs” – like optimize, predict, recommend, score, and forecast – as they explore new data monetization opportunities.

Note:  I will provide some specific examples of asking probing business questions in my next blog that can provide the starting point for the Solution Engineering development process.

Step 4:  Decompose the Business Initiative Into Use Cases.  The next step is to conduct a series of interviews and ideation/envisioning workshops in order to brainstorm, identify, define, aggregate, and prioritize the use cases necessary to support the targeted business initiative.  For each use case, we want to capture the following information:

  • Targeted personas and stakeholders (including their role, responsibilities, and expectations)
  • Business questions that the stakeholders are trying to answer (or could be trying to answer if they had access to more detailed, more diverse data sources)
  • Business decisions that the stakeholders are trying to make (and the supporting decision processes including timing, decision flow/process, and downstream stakeholders)
  • Key performance indicators and key metrics against which business success will be measured
  • Data requirements (including sources, availability, access methods, update frequency, granularity, dimensionality, and hierarchies)
  • User experience requirements (which should couple closely with the user’s decision making process
  • Identify analytic algorithms and modeling requirements (predict, forecast, optimize, recommend)

Step 5:  Prove Out the Use Case.  Now is the time to bring data and technology to bear to prove out the solution.  This is a good time to introduce a “Proof of Value” where the desired solution is flushed out using the full depth of available data and full breadth of technology capabilities.  At this point, we should have a solid understanding of the desired data and the necessary technology capabilities to build out the Proof of Value.  This process should include:

  • Gather required data, both internal as well as external data sources, and explore the use of third-party data (see data.gov) to help broaden the quality of the analytics
  • Define data transformation and enrichment processes
  • Define analytic requirements
  • Define user experience requirements

Step 6:  Design and Implement Big Data Solution.  Based upon the success of the Proof of Value, it’s now time to start defining the detailed data and technology architecture, and develop a roadmap for integrating the analytic models and insights into the operational and management systems.  The implementation plan and roadmap will need to address the following:

  • Data sources (both internal as well as external data sources) and data access requirements
  • Data management capabilities (master data management, data quality, data governance)
  • Data modeling capabilities (data schema, key-value pairs)
  • Business intelligence capabilities (performance monitoring, reporting, alerts, dashboards, KPIs)
  • Advanced analytic capabilities (statistics, predictive modeling, data mining) including real-time analysis capabilities
  • User experience requirements (operational systems, management systems, and dashboards)

Solution Engineering Tomorrow’s Business Solutions

While Solution Engineering might not be tomorrow’s sexy job, Solution Engineering will become more and more important as 1) the amount and variety of data continues to evolve, 2) technology capabilities continue to expand (fueled by both venture capitalists and the explosive growth of the open source movement) and 3) mobile devices and smaller form-factor mobile apps redefine the user experience.  As the data and technology sands are shifting under our feet, it will become even more important that we are focused on delivering business solutions that have a high return on investment and a short payback period.


[1] Business initiatives are cross-functional strategic business projects with senior management ownership, typically 9 to 12 months in length, with clearly defined financial or business metrics.  It’s not unusual to find business initiatives called out in an organization’s annual report (“President’s Letter to the Shareholders”).

Update your feed preferences

Viewing all articles
Browse latest Browse all 17822

Trending Articles