1. Enabling Technology
2. Applications to Business Analytics

Enabling Technology

One of the key aspects to the ParAccel message is that traditional enterprise data warehouses (EDWs) have always had the ability to build analytical logical models.  But it was moving those analytical models to the physical implementation that cause issues for traditional row-based DBMS applications.

Now with ParAccel’s “software appliance” architecture for columnar-based DBMSs, EDWs can grow from the theoretical into the actual because now those models are possible due to the power and speed of the ParAccel database.

The particular area that I enjoyed was the ability to provide PCI compliance for implementations in finance and in particular telecommunications.  In the past, I have seen IT departments “encourage” (read demand) PCI compliant level encryption for entire data tables based on the need to have a single column encrypted.  ParAccel offers that ability on a column by column basis which allows for EDW implementers to make better decisions on where they would like to allocate their DBMS resources both in terms of disk space and processing power.

NOTE – I would liked to have seen/heard more about specific size/speed metrics from customer implementations on the PCI compliant encryption, but the answers I received provided comfort that the answers weren’t simply theoretical. 

Business Analytical Applications

One of the topics brought up both “in the room” and “in the twitter-sphere” was about how it was entirely possible that the Voltaire quote, often often attributed to the late Senator Ted Kennedy, had direct application to perceived success of traditional EDWs:

The perfect is the enemy of the good. 

This quote implies that many EDWs in striving to be “perfect” in terms of data model design and being able to understand everything from “the things we know we know” all the way to the “unknowns we don’t know” often missed the opportunity to provide a very specific analytical ‘services’ to the business stakeholders of the EDW.

ParAccel aims to correct that issue with EDWs by using their enabling technology to solve some of the analytical issues facing businesses today whether that be associated with the EDW or the operational platforms that hold the raw data.

Again, a telecommunications application caught my eye.  LatiNode’s Communications used ParAccel technology to prove out an active analytical business model for optimization in telecom network routing selections.

NOTE – In the interest of full disclosure… Despite LatiNode’s compelling application of telecom route optimization analytics, it appears that at least some of their success at telecom network optimization came via unscrupulous routes ( …no pun intended… ) and is now longer operating.

The ParAccel featured customer page provides more information on other specific analytical applications.

In Summary

ParAccel, in my humble opinion, is approaching the data analytics market in the right way.  They are cautious in their customer “selection” and their product development.  They aren’t necessarily attacking the ‘big boys’ of the DBMS world, but rather providing a clear alternative in particular areas to build their reputation and qualifications.

I look forward to hearing more about the customer successes of ParAccel and the “impossible” tasks that ParAccel enables as many industries move from traditional EDW implementations to more targeted, and potentially operational system oriented, analytical applications.

• We are burdened with all the baggage of EDW infrastructure, politics, etc.
• Business demands extreme agility in deployment new analytic applications
• EDW trends to shift power away from end users, creating tension

1. Intrinsic speed of the technology is now supporting any thinking style
2. Analytic capabilities allow us to ask the 'unaskable' questions
3. Elastic scalability ensures full scope of required infrastructure

Often times software companies have a solution in search of a problem or an elegant solution that "only" really works on the fringes of business problems.  The recent presentation from Composite Software to the BBBT showed me that Composite is not only elegant technology but relevant to direct business issues.

Out of the Gate

Composite brought one of their clients - Compassion International - to present ( note i believe that this was a first for the BBBT .... ).  What Compassion International has been able to accomplish in a two year timeframe with Composite for their BI/DW needs is really amazing.  It shows what happens when you remove the "empire building" from the usual Business Unit vs Business Unit; Business vs IT; Everyone vs Finance games that usually cripple the current era of IT implementations... not just BI/DW.  I highly recommend that any organization take a look at what they can do when they stop thinking about protecting org charts or budgets and start thinking about what the company is around to do... in this case helping as many children as possible with the available funds.

Composite's solution allows Compassion International to perform some very elegant technology solutions that solve direct business issues.  And again, this is accomplished without much of the turf wars of who owns what data, where and how...

Sizzle... and Steak

The initial aspects of the Composite presentation from Compassion were almost too good to be true.  Great combination of client and technology.... The second half of the presentation told the BBBT crowd more about what Composite was doing for solutions that may not have been as tailor made... And Composite ( ... in my humble opinion... ) did not disappoint.

• Simple data federation among a diverse of sources
  
• Extension or augmentation of data that is mostly in their enterprise DW
• Sharing or distribution of enterprise DW, especially in Web2.0 compatible formats
  

What is analytics? According to Aha!, it is embracing the "actionability" of data and statistical based models to drive business decisions. They feel strongly that companies differentiate themselves based on the level of analytics being employed. (see the graphic below)

From my standpoint, the question is whether making it easier to do predictive analytics means that low level analysts will able create viable models? Can people create a sophisticated model without understanding the statistical analysis behind the models? This is one of the limitations that traditionally has kept analytic applications complex and the adoption rate low.

We were treated us to a demo of their product which was based on a case study from Coventry Health Care. Their customer wanted to reduce their churn rate and develop an integrated retention approach. According to Aha!, based on the project they were able to improve their member retention by 7.5%, versus their target of 3%. For the demo, they used a flat file of data that had previously been run through a predictive analytic engine.

The user interface provides both aggregate and detail information. The detail screens contain 'sparklike' graphs that provide a window into the performance of each particular KPI. The summary screens provide some basic statistics of the KPI along with a larger line graph, which represents the historical trending of the measure. The graphs, while not flashy, do build incrementally on the screen. The business metadata is fairly sparse, which implies a user base that truly understands their data and the metrics being presented.

Unfortunately, I was unable to stay for the entire demo. From what I saw, I was impressed with the scope and organization of their application. Like most analytic applications, a great deal of the value in their application appears to be reliant upon the upfront development of the statistical models. While they don't solve all the upfront work, they are passionate about exposing complex analytics to the 93% of the market that is currently only being served by BI and reporting.

What to Like:

• I liked the fact that a SaaS company was moving to link strategy analytical models with operational processes.  This is a bold approach for an organization with data access challenges outside the corporate data center.... Then again SalesForce.com was a bold challenge to move the sales funnel outside the data center as well.
• I also liked the fact that Aha! views analytics as being much more than something that you can accomplish in MS Excel or with the SUM() function in SQL.  Along the continuum of analytics, Aha! looks at the highest levels of optimization and predictive analytics and how to put those analytics to use within the above mentioned operational business processes.... Not just models that "operate" in isolation among the guys/gals in the "white lab coats".
• Finally, I really appreciated that not only does Aha! take the improvement concepts of Six Sigma and TQM to heart in their preamble... Aha! brings that into their stated methodology to link the modeled results to their brand of collaborative analytics.

What not to Like:

• The demo provided lots of good information to the user.  However, I would have expected a more robust dashboard environment.  But this is a slight disagreement with the presentation layer.  It was agreed at the BBBT that it is/was much more important to have the data accurate than pretty.  In this Aha! provides the confidence that the numbers presented are accurate and faithful to the model.  Enhancing the 'human factors' is a relatively easy exercise.

In Summary:

I really like the features and functions of the Aha! SaaS Axel Services Platform and the methodology of the Aha! implementation.  In combination, they represent a good proposition for many organizations.  However, there are refinements to the human factor aspects and the overall high level messaging that Aha! need to make before the Axel platform can really move from the niche opportunities, that it has been filling, to a mainstream product offering that moves beyond those niches.

But again, SaaS pioneer Salesforce.com must have seemed like a niche player to many at its start to link sales performance with operational data.  While Aha! may not be in SalesForce.com's class in the future; Aha! is making strides to take advanced analytics from the lab to the frontline.

Today, we got to hear from Aha! at the Boulder BI Brain Trust. 

Once I disconnected my brain synopses that were connecting the company name to the 1980’s song and innovative video “Take On Me”, I wrapped into the ideas they were presenting.   They are taking on the “magical, mystical, overloaded” term of “analytics” - a market expected they expect to grow 10% next year.  They are targeting the “93%” of business users that don’t use analytics today.  Aha! defines their space as the “Business Embedded Analytics to operationalize predictive and other analytical models.”  This 4-year-old company is part of the growing trend of software-as-a-service providers.

After spending some time defining analytics, which remains elusive, Aha!’s definition is “Analytics are models and model-based decision making.”  Companies can differentiate themselves with the output of analytics.

They are attempting to bring, to pull a title of a talk I used to give, “data mining to the masses.”  This is a noble idea and has been pursued for quite a few years.  Microsoft, for one, has made great strides here.  The idea relates to the concept of “Attention economics”, which is an approach to the management of information that treats human attention as a scarce commodity, and applies economic theory to solve various information management problems (from its Wikipedia entry.)  

I have said this before, but it bears repeating here where Aha’s form of analytics is like the “data mining” definition we have used and they are addressing these problems with data mining.

“Data mining has long been a means to attaining high business value from a warehouse.  As the means of automating discovery to explore and identify new business insight, it stands alone as an access method.  Interactive Query or OLAP presents the measures of the business organized around their logical dimensions.  Hierarchies in the dimensions allow for organized grouping and lead to drilling up and down in the data to find what you’re looking for.  Mining, however, makes you aware of situations that may represent new market opportunities or business problems that have yet to surface to the level of notice through standard interaction methods.

However, much of data mining has been relegated to the domain of a special breed of expert, often holding a Ph.D. in a statistics discipline.  The mining process currently deployed in many organizations is not only time consuming due to the challenge of the tools and the semantic gap between the front line and the statisticians, it is also non-iterative in nature.  Discovered nuggets flow from the miners to the front line and are only selectively interesting and actionable.  The feedback loop is missing.  It’s like having a luxury car but keeping it parked in the garage at all times.  Mining tools that are interactive, visual, understandable, well-performing and work directly on the data warehouse/mart of the organization could be used by front line workers for immediate and lasting business benefit.

The advanced techniques deployed in a lot of mining programs are generally well beyond the understanding of the average business analyst or knowledge worker.  If this advanced level of analysis is reserved for the few, instead of the masses, the full enablement of the warehouse in the organization cannot be realized.  If those whose analytical interests stay well within the complexity of computing sales commissions are shut out of mining, mining is not nearly as effective as it could be.”

Cut to Aha!’s take on this problem… via a long-awaited demo…

In the healthcare example, they created a large flat file pulled out of the client’s databases, to assess propensity to disenroll.  In the example, they delivered improved retention rates at a measurable NPV of $43 million.

Customers of Aha! (product: Axel) are able to interact with a model until it is perfected.  The model runs on continual updates of data sent to Aha!.  Overall, including some things seen under NDA, it’s impressive.  Some of the challenges include that the collaboration is occurring only while the customer is reviewing the output, though the metrics are numerous and interesting and the interface obscures some of the value.

Aha! is bringing together an old world that needs change - data mining/predictive analytics - and the new world of software-as-a-service.