The Text Big Thing?

Hi Thrill-seekers!

With all of the consolidation happening in the software industry, there’s been a lot happening to fuel the imaginations of industry-watchers who already have finely-tuned high performance imaginations. The impossible becomes unlikely. The unlikely becomes possible. One of the threads among the buzz among industry pundits kind of goes liked this:

“BI (Business Intelligence) and associated technologies focus on extracting and driving business value from structured data and ECM (Enterprise Content Management) enables organizations to manage and exploit their unstructured data and content to benefit the business in a number of ways -- combining the two would provide total data coverage and maximum value.”

Or something like that.

On the surface, it sounds like a good idea; however providing coverage for all data enterprise data assets has little benefit in and of itself.

An Admittedly Structured-Data-Centric Perspective

Rightly or wrongly, we BI folks have historically held certain beliefs to be self-evident:

· Data that has a close relationship to core drivers of the business is more valuable than data that doesn’t.

· Data that directly relates to the revenue and cost drivers of the business are most valuable to the business

· Aggregated information that reflects the structure and drivers of the business is particularly suited for effective management.

Software vendors of ECM/Text Mining solutions usually make this argument:

· On average only 20% of the total data in an enterprise is structured, leaving a full 80% unexploited and unmanaged (this is always the first slide after the title slide in their basic PowerPoint presentation. By the way).

· Many of the key indicators for performance can’t be gleaned from structured data. Items like customer satisfaction and warranty claim trends really live in the comment field of systems that are used by customer service clerks and garage floor grease-monkeys who use a variety of terms and abbreviations to describe similar problems.

· The knowledge worker’s connection to unstructured data – their email, documents, comments and web pages is stronger than the structured data because they spend more time with it and relate to it more easily.

While all of the above is true, a combination for its own sake doesn’t do anybody any real good.

The Case for Careful Combinations

There’s a case for careful combinations of these two worlds as applications merit. Many of the applications may start with either structured or unstructured data and extend the other. For example:

• Text categorization – use categorization technology to structure text into relational database tables for SQL query tools like Cognos Report Studio and Query Studio to aggregate, sort, and filter the results into easily consumable groupings and trends;
• Text mining through Performance Management (PM) artifacts – Enable the growing document, metadata and data assets generated by BI and PM systems to be “search enabled” for relating to enterprise taxonomies;
• PM as part of workflow processes – Place PM software on top of document workflow processes to instrument the health of the process. As well as BI and PM assets to support decisions required in the execution of the process;

And one more – the biggie…

Performance Management: The Killer App?

One of the reasons that convergence for its own sake isn’t important is that there are very different buying centers for PM and ECM systems. BI vendors won’t be interested in helping a publisher track down a picture of a kid eating an ice cream cone. ECM vendors don’t care most of the time if the inventory manager can list all out-of-stock items and drive demand trends for them

But…

When a company has a clear sense of strategy and has crafted their objectives to drive towards it, they assign metrics to ensure that they can track progress toward their planned targets. In the Cognos world, business intelligence is used to understand why things are on or off track, or to reveal opportunity to improve the status of the business. The reach of these PM systems ideally drives down to everyone with an area of accountability in the organization.

With that in place, you can imagine a very rich, hierarchical model of business performance that has a high degree of organizational coverage. That set of inter-related plans, metrics and report sets forms the hierarchical enterprise performance taxonomy on which to hang related text mining, content management, community management and high-value workflows.

For example, if you “own” a metric for the company, let’s say “cost per square foot of retail space” for your region, you’ll want to keep all of BI related to the drivers of that metric nearby. You’ll also want all plans, documents and search access to all related content. You’ll make all of this available to everyone in the organization who contributes to the performance of this metric and to all who make decisions using the status of this metric. This is a very real blending of the structured and unstructured worlds with clear, high-value application benefit.

Top-Down and Bottom-Up: Value at every step

Companies aren’t buying software systems they way they bought ERP systems in 1999. The days of the big-meal commitments and sketchy visibility to ROI are over. Way over. It’s unlikely that a whole-enterprise top-down implementation of a system like the one that’s described above will be bought an implemented all at once. Companies need to look for a good starting place. Here are some attributes which might help in selection:

• A clear candidate function, department or process for performance turn-around.
• A large amount of comment-field data that’s useless to traditional reporting tools.
• A clear sense of the metrics that drive a function or process.
• Lots of data and lots of documentation.
• Disparate data warehouse, content management or collaboration systems.

Cognos and Partners: Completing the Picture

As of now, there’s no one vendor out there that offers the complete story, because of the buying center issue described above. However, each world has a list of converging opportunities as implementations get more productive and look for the next step of value. When you see Cognos picking partners in the ECM/Text space, there’s a clear reason why. Our leading-edge customers get it. They get the value of some of these killer apps and are asking us to work with their chosen ECM/text mining vendor or vice versa.

So, when you hear industry pundits honk about the “inevitable convergence of structured and unstructured data management” ask them about the requirement of senior management to retrieve the ice cream picture, or service companies to report on knowledge assets rather than relate them to projects. It’s not so easy and it’s not so obvious. But there’s room for some killer apps and a ton of customer value in there somewhere.

On your marks…