AUGUST 2015

It’s not uncommon to ask the question, “Did we make the right decision?”. The process of reconsidering a business decision often stems from uncertainty associated with either the lack of usable data or the use of questionable data, which can negatively impact a well-intended data-driven decision. The lack of usable data generally arises from poor information accessibility due to either gaps in employees’ skill levels or the fact that information is often walled off in silos—or both. The struggle with questionable data often comes about when existing data mining methodologies focus on traditional KPIs (Key Performance Indicator) without evolving to meet changing business requirements. This phenomenon hinders you from addressing increasingly important data sources, and uncovering previously unknown—but often critical intelligence from sources.

Just like driving a car, the consequence of making a business decision without checking your blind spots could be serious. But it is possible to mitigate these risks and improve the speed and quality of a decision using a holistic analytics approach that takes full advantage of the 3V’s – Volume, Velocity, Variety – of Big Data.  The 3V’s essentially mean that we are dealing with a situation where a massive amount of data, in all kinds of format (e.g. text, video, audio, database) and from diverse sources (e.g. social medial, email, chats, website, sensors, transactions), is growing explosively.  Instagram alone reports an average of 70 million1 photos uploaded per day!

There are big opportunities associated this phenomenon. While there are countless use cases of how to tap into big data for actionable insights to drive sales and beat competition, let’s look at a simple case of customer intelligence. Every customer touchpoint – clickstream, web chat, call center, social media, sales, marketing, etc. – represents one of many views into the true voice of the customer.  We can no longer afford to just focus on the traditional analytics based upon transactional data such as answering typical questions like - how much did we sell? How many new customers did we acquire? What were the best-selling models?  Fortunately or unfortunately, there are a lot more pieces of the puzzle. 

It’s now common to monitor your brand and those of competitors in social media now.  So, can we easily analyze the social media reviews of competitors’ products and tease out the key concepts of these reviews.  What about identifying the most influential customers with huge Twitter followings who might have tweeted about a less-than-stellar experience with your company or your competitors?  And, please don’t forget all the customer inquiries coming into your call center as well. What are the key concepts from those conversations?  Effective analysis of audio data from call center conversations resulting from marketing campaigns might review unprecedented insights on various aspects of the campaign. In some cases, it might tell you that the call center resources and training were not optimized for supporting a sales campaign despite having a very compelling offer.
Can we effectively answer these and other critical questions in our quest for a clear and comprehensive understanding of the voice of our customers? The answer is that it depends on various factors:

  1. Technology
  2. Human Resource And Skill Level
  3. Data Volume, Data Currency And Data Access 

The right technology can help address issues related to (2) and (3). So, let’s take a look at the key factors in technology selection:

1. Build A Unified Data Analytics Environment For All Data
Niche analytics tools have been around for quite some time, and it’s likely that your organization already has diverse pockets of data analytics activity. If so, each of these activities is probably providing a siloed view—addressing only limited business requirements and exposing you to the risks of shadow IT.
Compounding the issue of silo growth is the variety of data—including traditional structured data from CRM, ERP, EDW, machine and application logs, and other unstructured data (text, video, audio) from social media and corporate repositories. These types of data should also be addressed to achieve that coveted “big picture” view of information. By establishing a single environment for data analytics across all of your data sources, you’ll not only improve staff productivity but also deliver much-needed 360-degree intelligence.
 
2. Create Access To Robust Analytics Capabilities
Analytics are only as good as the data we provide. One of the key success factors in getting users to adopt a single data analytics environment is to make sure they have access to needed analytics functionality. While most companies are good at leveraging structured information such as ERP data, they often struggle to derive structure from unstructured content (blogs, tweets, etc.) that can be used for richer analytics. The environment must be seamlessly integrated into proven data analytics engines that can truly address the big data challenges with the following functions:

  • SQL-based analytics for everything from reporting and dashboards to advanced analytics such as Live Aggregate projections and Geospatial
  • Hybrid analytics that use statistical techniques and natural language processing (NLP) to detect concepts, patterns, trends, and relationships without being limited by traditional linguistic rules associated traditional NLP
  • The ability to process and analyze all forms of data formats (for example, database, clickstream, sensors, email, documents, social media, video, and audio) across diverse data sources from both behind and outside the firewall

The value from unstructured data starts to become apparent when we can derive structure from it and incorporate that into our analytics. For example, if we use technology to extract topics of conversation from a stream of call center audio and incorporate that with customer churn information, we might derive a highly accurate model that predicts churn while a person is on the phone with the call center—giving the business the ability to save a valued customer.

3. Deliver An Intuitive User Experience
A key success factor is quality and ease of use of the user interface. Not only should the environment be highly intuitive and simple in nature, but it should provide straightforward data visualization and data exploration for business users. They should be able to engage in self-service analytics without extensive data science training and there should be an easy-to-use capability for developers to extend the environment to address future requirements.
 
4. Get The Perfect Fit—Without Breaking The Bank And Running Late
Break away from relying on shrink-wrapped applications that have limited flexibility. Don’t let your ROI suffer because you have to pay for something you don’t really need. Look for use-case-specific application templates with built-in best practices so you can strike the delicate balance between starting from ground zero with a big consulting project and putting up with rigid box cutter implementation.
 
5. Protect Your Investment With Proven Scale And Performance
One of the three Vs of big data is volume. You can count on the volume of your data to keep skyrocketing. As your business continues down the path of being a truly data-driven organization, you can also expect the number of users performing self-analytics to increase. It is critical that your data analytics platform easily scale to address the growth in data volume, data variety, and user community. The ability to deliver the right insights at the right time will help ensure your sustained success.
 
How Do You Get There Faster And Safer?
According to a recent Accenture research report, “89% (of respondents) say that companies that do not adopt a big data analytics strategy in the next year risk losing market share and momentum.”2 This is hardly surprising given the hypercompetitive market, there is no time to waste. Technology adoption takes skills and experience. To shorten your time to critical business insights, make sure you engage the right service provider to help you realize your big data analytics strategy. They should have expertise in big data analytics and it is highly desirable that they also have strength in big data management. Analytics, while very important, is one of many stages in the big data lifecycle. Big data, if taken care of properly, will take good care of your business.
 
1 https://instagram.com/press/
2 How the industrial Internet is changing the competitive landscape of industries, February 2015, Accenture.

 

 

 

JOE LEUNG  Marketing Manager Big Data Analytic Solutions, Hewlett Packard has over 16 years of experience in technology marketing. He started in Japan as the country's product line manager for Hewlett Packard's high frequency electronic design automation software. His career continued across various businesses within HP resulting in a diverse background encompassing hardware, cloud, software, direct and indirect Go-to-Market models. Prior to his current role, he worked on cross-portfolio solutions marketing with a focus on the small and medium business segment. Joe has an MBA from Duke University, MSEE from University of Texas at Arlington, and BSEE from University College London.

Contact Joe Leung

Listen To This Interview with Joe Leung