Recently, I listened to a webinar that featured a business intelligence vendor talking about BI self-service, and...
I had a major gripe with a key part of the presentation. The guilty party will remain anonymous because it wasn't just the fault of the one vendor. It's an industry-wide problem that affects BI users in general.
During the presentation, the speaker declared that "business users must become more technical" to use self-service BI tools successfully -- a mantra that he repeated in multiple variations as he went on. Wrong!
The people who run BI vendors and develop their software tend to be smart, but they also tend to be very technical. They create fancy new algorithms and data analysis functions, and then think the reason they're not selling enough software is because business users are slow, incompetent people who just don't have the knowledge to comprehend such brilliant features. Also wrong!
This might come as a shock to some, but the business community includes a lot of very smart people, too. However, they're usually busy focusing their intellect on issues in manufacturing, sales, marketing, finance and other business operations. Just as it's not the IT or BI team's job to understand tax regulations, the legal details of a sales partnership or other business matters, it's not the job of business executives to understand analytics applications with poor user interfaces -- even ones labeled as self-service BI software. Nor is it their job to understand cryptic information in database fields -- for example, whether db_gross_rev and st_rev_total mean the same thing in different places.
BI self-service involves more than just handing business users some tools and saying, "Voila!" Calling software self-service is only valid when you deliver something that people can actually use to do the things they need to do. That means understanding the different types of users, and their needs, better than many in the BI industry have done to date.
Meta layers matter to BI users
Metadata is a good starting place. BI vendors routinely talk about metadata, but I'm not sure they all understand the full impact it can have on a UI. The main focus seems to be on intermediating between multiple systems to ensure that data is organized consistently. That's absolutely critical, but it's not sufficient. When the use of metadata is capped at presenting a consistent view, we're limiting self-service BI tools to business analysts and a few guru end users; people who understand data structures.
It's not up to other business users to become gurus. As a result, it's necessary to create a metadata layer that does more than interpret db_gross_rev as revenue, while still leaving large tables of information for business users to wade through on their own.
There's also a lot of talk about how broader use of artificial intelligence (AI) technology is going to change things in business environments, as well as our personal lives. In relation to BI, I've heard predictions that it will enable better semantic analysis of loosely structured data, plus other advances. But one area I haven't heard much about is AI's potential role in the problem discussed here: better facilitating BI self-service processes.
To create a BI user interface that a typical business user can leverage, semantic analysis must be used for more than looking at data that has been captured in analytics systems. AI must be incorporated into the user interface itself.
When different business managers in a company ask for information about revenue, a self-service BI system needs to do more than parse the appropriate access level and data definitions for each user. A built-in AI application that has basic knowledge about individual users and what they're looking for can assist in delivering the right data to them in a timely manner.
Don't ignore the full scope of BI needs
I can see everyone on IT and BI teams emphatically saying, "Of course I don't ignore business users! I need to know the business to build BI systems!" The problem is that's only half of the equation.
It's one thing to understand the nuts and bolts of business processes enough to provide relevant BI and analytics applications. It's another to understand the flexibility and -- two things many technical people think are evil -- the intuition and qualitative experience that go into business decisions.
In a business sense, intelligence is more than the result of queries and technical routines; it's more than an automated analytical process. Too many in the BI community, both at vendors and inside user organizations, have ignored the broader view of intelligence that would help them provide full-fledged BI self-service capabilities to business managers and other end users.
It's time for advances in metadata management and AI technologies to be brought to bear on more than the back-end portion of BI systems. We need to focus on truly delivering self-service business intelligence -- to the fullest extent possible.
More from David Teich: IT still has a role to play in self-service BI projects
Tips for business users to boost adoption of self-service software
Middle ground on autonomy, governance needed in managing self-service BI