Predictive Analytics - There is a time and a place

Predictive analytics have been touted as the holy grail of Business Intelligence for quite some time now. However, how many organizations are actually using predictive analytics? In a recent survey conducted by TDWI only 21% of organizations have a full or partial predictive analytics deployment, while another 19% are in the process of developing it.

http://www.tdwi.org/News/display.aspx?id=8699

What does this mean to the majority of organizations? In my opinion it means that we need to apply the techniques of predictive analytics as the business case warrants. For example, if we can understand customer behaviours and predict what actions will appeal to each type of customer, we will be able to produce significant ROI for the organization. In order to increase the adoption of predictive analytic deployments, we need to take this out of the back room analyts hands and into the hands of the employees working directly with the customer. The trend towards operational business intelligence is a great way to inject targeted intelligence directly into the work processes of these individuals.

To think about it another way, does this intelligence absolutely require human interaction at all? If you have a rules based engine in place, you can directly take action by plugging in the different variables to the model, retrieve the scenario with the highest score and take the next logical step of implemention directly. If you don't require a decision from the knowledge worker (and we are nearly all knowledge workers), why add another piece of information to their already overwhelmed senses?

Predictive Analytics have a time and a place but the organization and technology need to be ready for it. If you don't have ready access to data, business stakeholders, and modelling expertise take 1 step back. If you don't have a way to take direct action from your predictive analytics solutions take 2 steps back!

Mark

SOA & Rules - Great Blogs

Increasingly as I look at where Business Intelligence is heading, we talk about integrating BI into business processes and pushing BI to the front lines. Ran across the following blogs that discuss harnessing a Service Oriented Architecture and rule engines to integrate operational decision making into applications.

Smart Enough Systems: http://smartenoughsystems.com/wp/
Enterprise Decision Management - a Weblog: http://www.edmblog.com/

As if you didn't have enough to read...8)

Mark

Enterprise Search - Foundation for integrating Information & Context

For a Business Intelligence guy it is a little strange dipping my toe into the world of Enterprise Search, but for some strange reason it keeps surfacing to the top of my search queries on Google. I'm starting to think that the typical web browsing session will closely mirror how information workers should be able to access corporate data.

It really should be that simple, open up your browser, enter in a couple keywords that describe what information you are looking for, hit the enter key, and presto you have access to the accumulated knowledge of the organization on your screen. Maybe you are describing something that is called something else... Your browser gently suggests a couple like terms (synomyns) that you can click on to expand your search.

According to studies, information workers can spend up to 25% of their time trying to find information. If we could easily quantify what the value to the organization of reducing this time spent by only 5% we would have an ROI that would convince anyone to fund an enterprise search initiative. Worse yet, up to 40% of the time the information worker can't find all the information they require.

The true value of Business Intelligence is unleashed when it is integrated as much as possible into the processes and daily tools of today's professionals. This means that we need to provide a 360 degree view of the business task or problem at hand to make the people turning this information into action as effective as possible. This is the promise of Enterprise Search, the Google-like functionality that makes information a tool rather than something you collect.

Until next time,

Mark

Dashboards by Example

I love this site! Nothing like looking at lots of sample dashboards!

http://www.enterprise-dashboard.com/

Operational Reporting Meets Business Intelligence

One of the interesting debates we have had over the past couple years is the difference between true Business Intelligence and Operational Reporting. This discussion started around who within the IT organization was accountable for developing "reporting" for applications. We have come from a place where Business Intelligence and reporting were considered synonymous, and all reporting requirements were considered the domain of the organization's Business Intelligence team. The argument was that reports that are more operationally based needed to fit into the overall information architecture, along with more analytical style of reporting. The issue became that no single team was able to have the subject area knowledge of every operational system within the organization.

Ok, phase II, all operational reporting requirements would be handled by the development teams most familiar with the application. This makes a lot of sense since a report can be considered simply an output of the application (definition of an app: inputs into a process, a process that happens, and then outputs come out). The major issue here is that this can lead to silos of reporting being developed that are not integrated into the enterprise information framework. As an end user, do you really want to have to go to multiple applications / portals to get all the reports you need?? This doesn't make any sense, as a consumer I want the "one stop" shopping approach.

The future... I really see that we have an opportunity to provide one integrated portal to knowledge workers that contains all the information they need to do what knowledge workers do best. It shouldn't matter what application / database the data is contained in, it should all fit into one portal, one integrated set of data assets. All the major BI platforms are now able to support basic reporting, slice & dice, predictive models and dashboards. We truly can make a one size fits all platform, even if we end up with a couple technologies being stiched together. As long as they can integrate (meta data, meta data, meta data) we can still call it a single platform.

Ok so the technology is there, but how do we make this all work? That is the million dollar question, by definition we would need to account for the following:

• Multiple development groups (IT & Business) contributing to this single platform
• Decisions on how to present data for reporting (source system, operational data stores, data warehouses) - Architecture decisions to ensure appropriate reporting architecture are key
• Providing a set of best practices and standards across all teams
• Potential conflicts between groups on business definitions and rules

The concept of the Business Intelligence Competency Centre (BICC) can be the glue that holds this distributed approach together. The BICC can set the standards, build the processes and provide that central view of the overall information architecture that is so key to leverage a single integrated environment.

Here is a link describing the BICC concept, there is a ton of information out there, most of the major BI vendors provide information as well.

http://www.olapreport.com/bicc.htm

No one said it would be easy...8)

Mark

Costco - Operational BI in Action

We were recently at our neighbourhood Costco on our bi-weekly expedition to buy bulk and ran into a situation where a very simple application of operational Business Intelligence was successful with me and my wife.

We were standing in line at the checkout when a Costco employee came up to upsell us their Executive membership. Normally we aren't interested in paying more for a membership, especially when we already pay $50 a year. The executive membership is another $50 a year for a total of $100. However, on this day we were feeling charitable so thought we would give him a chance.

First of all, he scanned our current card. After scanning our card he was able to show us that if we spend the extra $50 to upgrade the membership we would save at least $65 a year! This would pay for itself! Now, lets think about the BI process in the background:

• Rep scans our current membership card on his handheld / wireless device
• Last year's purchases are accessed online, and then a formula is applied to determine the rebate we would of gotten if we had the membership last year
• Rep lets us know the savings and is able to sign us up for the Executive Membership on the spot!

As BI goes more mainstream and is embedded in more applications, we will see many other examples of Operational BI in action as consumers. How many memberships has Costco been able to get because of this? Probably a lot considering their is always a line up for the checkouts!

Until next time,

Mark

Business Intelligence 2.0 - Overview

After doing some initial research on the Internet it became readily apparent that a good deal of the concepts I had in my mind were captured under the heading of "BI 2.0". This seems to be taking the Web 2.0 social networking concepts and bringing it into the Business Intelligence world. Business Intelligence really is the perfect implementation of the Web 2.0 technologies, the core business process of people looking at information from different perspectives and collaborating on what they see to make business decisions is really what Business Intelligence is all about. There are many other aspects of the next wave of Business Intelligence innovation outside of the Web 2.0 realm, but for now I'll focus on the impact to the user experience. In the end our client's needs rule!

Before digging in to any one topic in depth, thought it would be best to provide a brief overview of what each area is at a high level.

Intelligent Search

Searching all relevant information (structured of unstructured) to return results in the context of what is contained within the documents. Very similar to web search engines, but focused on enterprise information assets. Idea is to search on a topic and return back both flat reports, dashboards, business process documentation, social network (discussions, blogs, groups, etc.) information and anything else relevant. This information can be easily bookmarked into our portal views to enhance and build on information exchange relationships.

Tagging

Adding context to information through the subjective experiences of end users. Would work in the same way web sites like flickr allow all viewers to tag their images, interpreting pictures in ways that computers can't. This is key to enabling Intelligent Search. Users would be able to rank information to make relevant content known to larger groups.

Discussions

Enabling discussion board style functionality within information portals. Common discussion can build more cross-functional relationships amongst users, building the informal networks so crucial in today's fast paced business environment. I can picture people annotating and highlighting sections of reports / dashboards right on the portal and being able to share this with their network.

RSS Feeds

Integrating Business Intelligence content into RSS readers would truly bring control over how users information is categorized by putting it right into their email clients. All the features of RSS readers will be available including refresh frequency, archival and labelling. Anything we can do to help organize the information avalanche will make our users that much more efficient.

Mashups

Allowing users to take more control of how they work with information is key to mashups. Rather than having to open multiple reports and try to come up with how they all correlate, imagine being able to pull pieces of these reports together on one dashboard and having it all just work? Add in visuals like placing performance metrics on a diagram of a business process to see analytically and cognitively how everything fits together. The possibilities are limitless here!

Blogs

Provide trusted Subject Matter Experts with a forum to share their knowledge. If you think about it, there is someone in every team that is the guru and has great expertise tempered by experience. These people and other holders of specialized knowledge like to be recognized for their area of experitse, give them a forum to share this with the larger community will increase their engagement and their value to the team.

Wikis

Now wikis, at first thought I'm not sure how this applies (might be my control freak nature), but having content able to be edited by anyone gives me the shivers! Nevertheless, there could be something here. When we have such great tools such as blogs and discussion forums, a great deal of knowledge is hidden in those threads of conversation. This is great for collaboration but maybe not so great for other people searching for this knowledge. If you can convince your business users to do it, using Wikis as the ultimate source of authority on one topic can assist with best practices and findings from Business Intelligence projects. Capturing all that great knowledge capital into an asset that is of value to the company and helps share that knowledge.

This is just my first pass at what is out there, will continue to dig!

Mark

The Journey Begins

It is pretty embarassing when you come back to a blog and the last posting was in November of 2005. Well given the new Web 2.0 world it must be time to start collaborating and sharing knowledge rather than being a scrooge and keeping it all in your head.

After about 7 years working in the field of Business Intelligence it is a scary prospect to stand up on your hind legs and sniff out what is happening on the horizon. It used to be so simple to do BI. Here was my typical project

1. Hear about some Business Unit that was looking at new reporting requirements. Well, reporting??? That is what we used to do, it is kind of embarrassing to think about how many reporting solutions were branded as "business intelligence"!
2. What's the first thing you do now?? Call the user? Are you kidding, what the heck do they know? I know what to do, lets look at the source systems, who cares how they are used!
3. Oh boy, data analysis, a BI geek's dream. Here we go, warm up your SQL coding fingers. Yee haw, lets do some group bys, order bys, distinct values until we curl up at night dreaming inner joins and histograms.
4. Once you know that data model and the data inside it you are off to the races (who cares how the data gets populated into the database, that would mean talking to the users, big no no)
5. Ok, now the fun part, lets design as many star schemas as are needed to use every single element in that source system. We don't know if these subject areas are going to be useful, but gosh darn it we are going to give it to the business. BTW, they call customer's accounts? But the source system column name was customer_id? (pesky users)
6. Now that we have our destination schema, now the truly fun part. And wasn't it fun? Nothing like hand coding hundreds of stored procedures to do your extract, transform and load routines. It was truly a dream for the hyper-active coder, to be pulling your hair out encompassing complex business rules directly in your code.
7. Oh the sweat and tears when you try to make this all work together, like pulling out finger nails with pliers. But eventually you get the spaghetti factory moving, you feel like the next BI messiah.
8. Hmmm, hey did anyone ask those pesky users if they wanted any reports off this beauty of complex systems?

If was fun wasn't it? The fact that your business users look at you like you have two heads didn't phase you at the time...

Thankfully, we have matured and have put in place some extremely sophisticated tools to make the above artwork of luck and raw perserverence something that is much more accomplishable. We are now poised on the next wave of innovation within the Business Intelligence and Corporate Performance Management industy, the new world of the collaborative web is going to take all this great information we can provide and place it within webs of users communities that can turn this information into action and results.

As I embark on this journey to see what is the right Business Intelligence strategy for my employer, I look forward to sharing what I find with whoever runs across this blog. If you have any comments or contrary opinions to those I state please, please comment, comment, comment. I'm wrong a lot, but it sure is fun to try! 8)

Mark