HP Business Intelligence Solutions

« April 2010 | Main

May 11, 2010

Answers to a Reporter's Questions About Pervasive BI - Part 2

In the last post, I shared part of a recent dialogue I had with Canadian reporter David Hamilton of Web Host Industry Review (WHIR) who asked some interesting questions about pervasive business intelligence (BI). Here is the rest of that interview.

WHIR: One of the success story examples of an organization using BI was Blue Cross Blue Shield of Kansas City which, among other things, used its newly found insight into its data to streamline internal processes, and even expose previously unavailable data to physicians who could use that knowledge to better treat patients. It seems to me that some of the benefits of BI in this case were, perhaps, not altogether anticipated. What are some of the - I guess - unforeseen or unexpected benefits of BI for organizations? 

VF: This is a fascinating topic. People pay money to hear companies tell these stories. I’ll give some examples that I’ve seen in recent years.

• I think it was a surprise, even to Wal-Mart, when several years ago, they were able to track the spread of flu across the US more precisely and accurately than the CDC, based on purchases in their stores of flu medications.

• I have seen a few occasions where results of BI analysis done in the data warehouse (DW) had discrepancies with the operational system reports. Further investigation uncovered errors in the operational systems which had an impact on the company’s business.

• Continental Airlines’ BI system allowed them to understand and serve their OnePass customers so well that when Jet Blue was launched in the early ‘00s, Continental was able to forgo the route their competitors took when they introduced low-cost lines like Ted and Song. Continental’s analytics enabled them to differentiate themselves on their service and attract and retain the customers willing to pay a premium for better service, a move that helped them maintain a strong competitive market position even in the face of downward price pressure.

•  

• And then there is one of the most famous cases of all, laid out so superbly by Michael Lewis in “Moneyball.” During the 2001 Major League Baseball draft, Harvard grad and statistics whiz Paul DePodesta introduced Oakland A’s general manager Billy Beane to the concept of analyzing data to select the right players and the best team, for the best price. Only DePodesta’s analysis of Kevin Youkilis’ statistical data indicated the promise of the kid who the scouts dismissed as “a fat third baseman who couldn’t run, throw or field.”

The 2009 American League ranking looked like this for the Boston Red Sox player who might never have made it to the pros but for data analysis:

On-base %                          2nd

Slugging %                          5th

Runs created                        8th

Pitches per plate appearance  1st

Youkilis also holds Major League Baseball’s all-time record for most consecutive errorless games at first base (238 in 2008), and MLB’s record fielding attempts  without an error at 1B, (2,002 also in 2008). He won the 2007 AL Gold Glove award for first basemen and in his major league rookie year 2004, was only the 7th player in Red Sox history to hit a home run in his first game.

As contrary as it is to century-old conventional scouting wisdom, predictive analysis has nevertheless proven to be successful. Evidence includes, as analytics expert and my fellow Red Sox fan Thomas Davenport laments, its adoption by the New York Yankees.                               

• Technically, data mining is different from BI, but if you’re using the broad definition of BI which encompasses everything including the data warehouse, then we can talk about data mining here. Discovering the unknown and unanticipated is what data mining is all about, as illustrated in the proverbial beer and diapers example. As the urban legend goes, analysis of Osco store receipts indicated a correlation between purchases of beer and diapers, and moving them near each other resulted in a spike in sales.

WHIR: Increases in technology have only recently opened up the possibilities of collecting and analyzing massive amounts of data. Can you tell me a little about the infrastructure needed to process this data and present it in usable ways?

VF: There are numerous new and existing technologies aimed at increasing the ability to collect and analyze large volumes of data by improving performance: data compression, 64-bit computing and in-memory processing, in-database analytics, MapReduce, column-oriented database storage, massively parallel processing and flash memory, to name a few. But there is more to this challenge than just the volume of the data.

We are entering a new generation of DW/BI with an expanded set of business needs that pushes the data requirements in several ways. Just a few examples:

• Analyzing unstructured and semi-structured content along with structured data
• Integrating data from many sources, including federation of data not in the DW
• Using a Complex Event Processing (CEP) engine to analyze incoming real-time data streams, possibly joining event data with data persisted in the DW
• Lower data latency – loading data into a DW for analysis on a continuous basis (vs. periodic batch load or frequent mini-batches)
• High availability systems that support data analysis for mission-critical applications
• Automating data classification and business actions
• Maintenance of detail data – More and more organizations are looking to the DW as a system or record to, among other things, ensure regulatory compliance, which means maintaining data lineage and detail.

An organization’s data volumes as well as the number and types of users plus requirements in any or all of these areas will dictate the needed infrastructure, which will be different in each case. But for sure, an infrastructure that is built today will need to be able to process the data volumes, as well as address the above requirements, either now or in the near future, because they represent the direction in which BI is going.

I would like to say farewell, as I am retiring from the HP BI blog and this will be my last post.

Posted by HP Business Intelligence Solutions at 4:26 AM

May 6, 2010

Digital Marketing Agencies – the overlooked partner for software vendors and IT consulting companies

I'm excited to introduce guest blogger, Simone Burrows. At HP, Simone is our business intelligence solution manager for the Health and Life Sciences industry. What a privilege to have someone who writes with such passion and conviction! I Simone, I hope you can join us often! - John Santaferraro

Ripped jeans and cowboy boots mandatory.

The IT and Marketing functions of many organizations are merging, presenting a challenge for digital marketing agencies as they try to work out who they want to be when they grow up.  While digital marketing agencies have largely recognized that this key client side trend must be addressed, software and IT consulting companies don't seem to have the issue on their radar, and are missing a vital partnership opportunity as a result. 

A recent blog post by Pro Bose, VP Digital Strategy at a leading NY agency addressed the issue from the agency perspective as follows - 

"With digital marketing increasingly taking on a central role in the media matrix, technology enabled platforms and features such as personalization and real time optimization based on analytics etc are very hard to separate from strategic customer segmentation as a separate offering. Most industries have started to look at budget allocations for initiatives that intermingle the boundaries of IT and Marketing. There is no reason marketing agencies shouldn't productize offerings in collaboration with software (enablement) companies. And certainly not try to build it out internally - the past decade is strewn with carcasses of overpaid underdelivered technology products that were the excesses of ad agencies trying to build them."

So, the agencies seem to have learned their lesson and aren't trying to have sideline in marketing software development - a wise choice.  But what about the software vendors and IT consulting companies in the game of selling and implementing marketing analytics, web portals, campaign management or complete CRM suites?  Well despite their best efforts to really understand the needs of their customers, they just don't have the type of in-depth marketing 'know how' of the ad agencies, and certainly aren't in the business of providing the creative brilliance that the best of them deliver.  Yet they recognize that being able 'talk marketing' when positioning their own marketing products and services is vital.  So how do they take it up a level?

It is clear to me there's a huge opportunity for the two to work together.  Just imagine a customer's surprise to be presented with a really tightly knit marketing solution.  The customer research, digital strategy and creative content that combines to form the marketing 'vision' from the marketing agency.    Let's bring it to reality with the best information management, data integration, reporting, CRM and advanced analytics technologies.  Maybe host it as an end to end marketing solution on a super-duper integrated data warehousing platform (HP Neoview comes to mind...).  The IT consulting company ties all this together so their customer doesn't have to.  Dreamy.

So, if it's such a great idea, why isn't this happening?  I mentioned that most software vendors and IT consulting companies just don't have the agencies on their radar.  But Pro Bose at also points out a key barrier -

"Cultural context is everything and product companies and agencies have cultures that are sand and water."

Hmm, no kidding.  Just try walking into the office of an ad agency suited and booted in your best consultancy clothes, only to find yourself terribly overdressed compared to the cool agency folk in their ripped jeans and cowboy boots. 

Software vendors and consulting companies have enough problems making their alliances with other software vendors and consulting companies bear fruit.  Meetings, dinners, promises made, not so many jointly closed deals. 

The same lesson about these 'traditional' alliances also applies to making new alliances with ad agencies successful.  It's all about relationships.  You can sign all the fancy agreements and alliances paperwork you want, but unless the person at Company A likes the person at Company B, nothing will ever get off of the ground.

So, a call to the Alliances organizations at marketing software vendors and consulting companies - it's time to shed the corporate clothing, change into your best ripped jeans and cowboy boots, and start brainstorming end-to-end marketing solutions for your customers.

Check out Pro Bose's blog!

Post by Simone Burrows
HP Business Intelligence Solutions
Health & Life Sciences Industry

Posted by HP Business Intelligence Solutions at 10:16 AM