Here’s a question for you: how often do you open Facebook just to go to someone’s profile and look up a piece of information about them? If you’re anything like me then the answer is not too often. At least 9 times out of 10, I open Facebook either because I got a relevant notification that I want more information about or because I have a few minutes to burn and I want to see what is going on in my network.
Just like we no longer look people up by flipping through the yellow pages, the primary value of social apps is not the manual retrieval of specific information but rather the sophisticated curation, prioritization and presentation of information. Of course this is nothing new or profound—imagine if Facebook was just an alphabetically sorted list of links to your friends’ profile information.
Crazy. So, why does your BI solution still work like the yellow pages? BI platforms of the future will look a lot more like the consumer platforms your users already know.
Here’s what you can do today to help get there faster.
Enterprise Data is (Really) Underutilized
To measure the effectiveness of a BI solution, set a ceiling on the potential value of your data for comparison. There are two critical dimensions along which to calculate this value.
The first is data-how much of the data available is actually getting looked at?
The second is users—how many people could do their jobs better with proper access to data? Are they actually getting access?
I’m sure everyone has heard about the spectacular growth of data lately, yet the volume of data in the warehouse is orders of magnitude greater than that represented in BI solutions. The majority of BI usage remains basic static reporting on a subset of data, whereas we all know that many of the insights desperately needed are of a much more dynamic nature.
And users are overwhelmed. The explosion in data increases not only the depth of data, but the breadth. Nearly every business process you can think of is now recorded–there is data out there for nearly everything in the enterprise, and it’s increasingly of interest to everyone.
And yet BI adoption remains cripplingly low.
Humans Are Human
This problem is not just a matter of throwing more tables into your BI tool or buying more user licenses.
Doubling down on what’s not working is just a faster path to failure.
One of the most daunting facets of the BI problem is the human aspect. People have their jobs to do, and of course they have other things they care about outside of work too. They are not itching to learn how to navigate a BI tool, get familiar with datasets, or proactively use it every day.
And even if you have a really easy tool and you get past that point, there are often simply too many questions that could be asked. People feel like they are seeking a needle in a haystack. They are faced with a blank slate and do not know where to start—likely they will just go back to their normal work and carry on.
The Last Mile: Push, Not Pull
So, you have all your data hooked up to your BI tool, you’ve got it modelled properly, and performance is pretty good.
That, on its own, still does not deliver insights to eyeballs. You can reduce the friction involved for a user to proactively engage to ask a question by ironing out the authentication flow, curating and labeling datasets, and enabling easy composition of queries through a search technology. This is a good thing, but you will likely still plateau at a subset of your users.
Let’s learn from Facebook.
Facebook recognized that the vast majority of the value lies not in serving user requests, but in leveraging its troves of data to proactively engage users. There are a number of ways a BI solution can take the first step in a user interaction (we’ll cover some below), but the king of them all is to actually discover and deliver novel insights in data—the needles in the haystack.
It’s true that these system-generated insights are a fairly tall order, especially compared to the current state of BI. Perhaps you’ve seen similar claims before and been disappointed.
But this kind of functionality is not as futuristic as you might think.
Finding insights is often not a matter of having the smartest, most complex algorithms or meticulously curated semantic models. Rather, it’s a matter of just churning through a lot of possibilities—looking for anomalies across various combinations of measures and attributes.
Combing through the entire haystack.
The system can also use other data that it has access to. For example, the subset of users who are actively creating new content gives the system an idea which columns are most important. And if the system has knowledge of some group memberships (perhaps for security) then those could be used to determine similarity of interests between users. These types of insights help to target and personalize answers for individual users.
Time is also an important factor here. If some report has existed in the system for some time, the system can use previous states of that report and derivatives of it such as drill downs as a time series, then flag anomalous changes. Perhaps the most important things to keep in mind here is that the aim of this insight generation is not to run your business for you. The goal is to present something interesting and relevant that warrants further analysis by a human hand.
What Can I Do Today?
While insight generation is not as far away as you might think, it is still not quite mature, let alone widely adopted in the BI space. But there are still a number of ways you can start moving your BI solution towards the Push model and away from Pull model.
Anything that enables BI interactions to be initiated automatically instead of by the user is a step in the right direction. And while not technically Push, anything that reduces friction in interactions is great for adoption as well. Here are some good examples:
- Integration into everyday systems:
- Email reports, preferably with the ability to open in a BI tool to do further analysis.
- Chat interface in company platform: Slack, Lync, HipChat, Salesforce Chatter….
- Notifications – email or preferably mobile.
- Mobile and responsive design. This is a classic friction reducer – nothing says friction like firing up the old laptop.
- Search. Consumer product after consumer product have proven search as one of the least intimidating and most powerful interfaces out there.
- Social/collaborative features. The activity of colleagues can go a long way in extracting needles from haystacks. Think things like “Recently viewed by your peers”, and “Trending now in group ‘Sales’”.
Humans Need BI—Let’s Deliver
As the big data explosion becomes old news, the masses of information workers are impatiently waiting to see what all this data talk means for them. As BI architects, implementers, and solution providers, we are collectively straining to bridge the gap between expectation and technological reality. However, we must not become disenchanted or lower our aim. The time has come to focus on the human aspect
We must solve our BI problems keeping adoption by the masses front and center in our minds.