The value of high-performance analytics

Like Vince said a few days ago, you don’t have to be Goliath to see how the intersection of big data and high performance analytics creates competitive advantage. Let’s also be realistic though: compared to someone like Walmart, pretty much everyone’s David! While Telstra and the Commonwealth Bank may not be the largest companies by global measures, they’re still great examples of how high performance analytics and big data can drive groundbreaking results.

Understanding why means stepping back to consider the fundamentals - to me, big data is both an absolute and relative definition. The absolute definition is the one people usually focus on and, more often than not, is used by detractors as a reason not to pay attention. That may seem counter-intuitive - after all, if it’s real and measurable, why ignore it?

Put it this way - how often have you heard someone say something along these lines shortly before they dismiss it?

  • “Our warehouse has only gigabytes of data, we just don’t need to worry about it at the moment.”
  • “We’re only starting out - we just need to focus on the basics for now.”
  • “Our warehouse is scaling just fine, we’ve got big data under control.”

Take advantage of what you have, not just what you want

It’s true that one aspect of big data is how best to manage and capture the increasing amount of information that we’re generating. Volume, velocity, and variety are important, but they’re not the whole picture. We’re in this game for value: without achieving an outcome, we’ve wasted time and money.

Most of the conversation about big data is about the mechanics of capturing new information rather than the outcomes from using the information we already have. It’s true that not every organisation has the transactional volumes of CitiGroup, the retail spend volumes of Catalina Marketing, or the market basket data of Walmart.

However, most organisations are sitting on data that they just aren’t analysing for fear of opening Pandora’s box - a typical telecommunications company with tens of millions of customers can easily be working with billions of call detail records, a positive goldmine for identifying networks and relationships to help drive targeted marketing and retention.

Most ignore this data simply because they don’t believe they can practically analyse it in any meaningful way. They may not think they’re missing out on the value of high performance analytics, but they are. It’s the same for any reasonably-sized retailer, bank, or insurer - the opportunity is there, it’s just a case of taking advantage of it.

Big Data: The case of ignoring the obvious

Let’s go back to our detractors - their statements highlight some fundamental misunderstandings, namely:

  • It’s about value, not data. Capturing information is critical, but it’s only the input. You need to do something with it to actually create value, otherwise you’ve just added cost. And, you typically need high performance analytics to actually do something with all that data.
  • It’s about continuous value creation, not a point in time. The variety aspect of big data ensures that there’s almost always a new way of delivering value or a new way that existing information can be leveraged to solve new problems. Big data isn’t a problem to be a solved, it’s an intermediary step to becoming a smarter organisation.
  • It’s about what you can do, not where you are now. There isn’t a maturity curve with big data - it’s just another information source. With the right tools, mindset, and approach, it’s no different to any other form of business analytics. Waiting to capitalise on big data means suffering significant opportunity cost and competitive disadvantage.

People are already doing this. Today. Yesterday, even; this isn’t something to be aware of as an obstacle threatening somewhere down the track. It’s here and it’s been here for years. The starting point may be different depending on the company, but one thing is common - high performance analytics drives real value.

Back in 2010 Telstra was acknowledged at SAS Global Forum with an Enterprise Excellence Award for their use of analytics. This would have been impossible without their ability to scale their processes to deal with ever-increasing amounts of information. The productivity boost was staggering in some situations - they managed to drop processing times down from 11 hours to approximately 10 seconds. They support multiple areas of the business including the contact centre, the retail network, and many other groups. And, by doing this, they saw real improvements - a great example was a 15% lift in their customer retention activities.

On the other hand, the Commonwealth Bank of Australia decided to focus the charge on managing fraud. Most companies align fraud management to various lines of business - credit, mortgages, and so on. The Commonwealth Bank decided not to follow this approach and instead establish a single platform to handle bank-wide fraud management. This would have been impossible without the use of high-performance analytics; the volumes and process challenges were too great. While this chutzpah alone was impressive, the really amazing thing was how well this approach worked - they not only doubled their detection rates in managing cheque fraud but improved their detection in Internet fraud by 60%!

Dare to be different and don’t take things for granted

This all brings us back to the subjective aspects. Big data and high performance analytics can also be seen as relative concepts - it’s about encouraging the attitude of taking advantage of the things you’re not leveraging. It’s rare that an organisation really exhausts the value of all its information assets; more often, it falls back time and time again on the information sources it feels most comfortable with.

A critical aspect of high performance analytics is re-examining those fundamental assumptions in the light of new technical capabilities and asking:

  • How could high-speed visualisation change the way I view current and future performance, driven by in-memory processing?
  • How could I improve productivity by orders of magnitude, driven by in-database analytics?
  • How could I improve operational outcomes by augmenting them with dynamic and powerful predictive insight, driven by real-time analytics?

High performance analytics is a great wake-up call. Too often, we get stuck in doing business as usual, achieving the same outcomes by doing the same thing over and over again. Business analytics is a discipline of change and there’s no greater potential for change than a disruptive technology that allows you to do what was previously impossible.

Take the time to re-examine what you’re doing with fresh eyes. I guarantee you’ll be surprised at what you could be doing.

tags: analytics, big data, business analytics, high performance analytics

4 Comments

  1. Posted April 26, 2012 at 5:50 pm | Permalink

    What advice would you give to companies who don't know where to start with big data? In my experience, the old way of thinking about analytics will just have you chasing the wrong data.

    • Posted May 3, 2012 at 12:24 am | Permalink

      It really depends where the real problem lies. Big data creates two challenges:

      * How will you capture and store everything that's being generated
      * How can you use it to drive economic return

      A lot of the time, I think the first challenge is big enough that teams forget about the second. That's obviously a bad thing - they eventually realize they're seriously exposed because despite having massive amounts of information, they still don't know what to do with it.

      As far as driving return goes, it's helpful to think in terms of evolutionary and revolutionary improvements. The "simple" way to leverage Big Data is to take existing approaches and either augment them or scale them. Augmenting them involves extending them to include different types of related information - if credit scores are currently based purely on structured data, start incorporating unstructured data as well. If your retention models use contract length as the primary driver, extend them to include relationship-based information so you can include viral churn.

      Scaling them involves making them more granular. Rather than using seven segments as behavioral classifieds, move to a micro segmentation approach where each segment is further decomposed into 20 sub-segments. Then, use these to help drive greater granularity and accuracy in attrition or cross-sell models.

      Revolutionary improvements are harder - there's no easy answer apart from having a lot of interesting discussions and creative speculation! Still, they're fairly easy to identify when you stumble over one - it's just a case of asking yourself what you can't do at the moment and seeing whether those assumptions still apply given Big Data and High Performance Analytics.

  2. Posted May 3, 2012 at 11:31 am | Permalink

    Hi there,
    I think evan has pretty well summed it up. If I was to be on my bigdata journey I would love to think about my own definition, the one trhat makes sense for my organisation. Although the "old" way of thinking about analytics may sound bad I don't think it is. My issue has always been that the so called "old" way is often not the "classic" way, it't the way organisations that have not understood the real value of analysis. Next month I will be blogging about something called the OODA loop, you should take a look at it on the internet - I think you will finnd the approach interesting. I would still look to seeing where the most value lies economically as he mentions. I still think you have to look at what hypothesis you are looking at and looking to collect the data required. Of course that may mean you have to start storing additional data over time to create an even more accurate view/model etc.. How to get started is to get started :-)

    • Posted May 7, 2012 at 9:01 pm | Permalink

      Thanks, Greg. I'm excited to read about your views on the OODA loop. I think Evan's hit on some good points, which also highlight the challenges of companies looking to do something with Big Data--finding where the most value lies economically is difficult for a company that has never undertaken such a task before.

      In my experience, this is where most of the roadblocks to implementation lie. That lack of understanding makes the goal of demonstrating value especially difficult. That's what I'm referring to when I talk about the "old" way of thinking.

One Trackback

  1. By Big Data Seats CIOs at the Boardroom Table on May 8, 2012 at 11:52 pm

    [...] timely, reliable, trusted and accurate data.  While we have been recently discussing the power of high performance analytics it is clear that data governance, data quality, master data management and data integration are [...]

Post a Comment

Your email is never published nor shared. Required fields are marked *

*
*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <p> <pre lang="" line="" escaped=""> <q cite=""> <strike> <strong>

  • About this blog

    The Asia Pacific region provides a unique set of challenges (ความท้าทาย, cabaran) and opportunities (peluang, 机会). Our diverse culture, rapid technology adoption and positive market has our region poised for great things. One thing we have in common with the rest of the world is the need to be globally competitive while staying locally relevant. On this blog, our key regional thought leaders provide an Asia Pacific perspective on doing business, using analytics to be more effective, and life left of the date line.
  • Subscribe to this blog

    Enter your email address:

    Other subscription options