James HolincheckAs an analyst, I get to see a lot of product demos (I started to count but it started to scare me how many I see in a year). One of my growing pet peeves is around reporting and analysis. Almost all that I see is what I call a “field of dreams” strategy. If we build a solution that encompasses all of the data required, that calculates all of the potential metrics that users might want, and provides them the ability to drill down and around all of it, then users will flock to use the solution because they will gain so much insight. I think that is why most BI solutions do not get used beyond a select set of users today.
At the end of the day, you analyze data to gain insight so that you can make better decisions and take appropriate actions. The starting point should not be the data analysis, but should be the types of decisions and actions that a particular user needs to take. Let’s look at an example. Let’s say that I am a regional bank manager and I have twenty bank branches in my region. What I might be concerned about are revenues, deposits, labor and facilities costs, profitability, customer satisfaction, etc.. These are business metrics that hopefully are tied to broader corporate objectives. So far, this is relatively straightforward. Let’s now say that customer satisfaction ratings have dropped below a particular target level. I know as a regional branch manager that this drop could be a leading indicator to revenue and profitability (assuming I have a good corporate performance management solution, then I should know the correlation/causality). That is not good. It is getting close to the holiday season and if that metric does not improve, I might not get my holiday bonus. I need to take action to bring customer satisfaction levels back up or I am toast.
This is a moment of truth for business intelligence. What would that regional bank manager do? At best, I might login to the system and see which branch(es) have the problem (or I might even get an alert that tells me). However, do you think they are going to go into a Business Intelligence system and drill down through all of the data to find what might be the potential root cause of the issue? Do you think that the business intelligence system will have drill down paths to all of the potential root cause data? I do not. This user does not have the time (remember, this person is overseeing 20 bank branches) even if they had the skills to do so.
Let’s move to my home turf, Human Capital Management, and assume that I know which branches have a customer satisfaction issue. What employee-related factors might be causing this issue? Most regional bank managers would do a different type of drill-down analysis to figure this out – pick up the phone and call the branch manager(s) in question and ask them what is going on (they might use more colorful language than that) that is causing the drop. The actions taken based on these conversations are not likely very data-based and frequently may be inaccurate (or make the problem worse). That is not good if I want to get my bonus.
In my idealized world, what a business intelligence solution should do is tell me which employee-related factors might be the likely cause of the problem. Has employee engagement dropped? Has 90/120 day voluntary turnover increased? Do the branches have open headcount (are they understaffed)? Has the number of part-time workers increased (maybe they are unhappy about not being able to work full-time)? What is the tenure of the bank management (are they new)? Do exit interviews show that management is a problem (if you saw “Undercover Boss” this week, you probably know what I am talking about)? I think you get the drift. There are many, many more questions that might highlight potential root causes.
What I want the system to do is help me understand which ones are the most likely root causes in this situation. Let me reemphasize this. I do not want to have to do a lot of analysis to figure it out. I do not want to have to call up HR to do some analysis (though that is better than doing it myself). I want the system to do the leg work. I want it to be intelligent enough to ask all of those different questions, look at the metrics associated with those questions and see if there is correlation/causation with customer satisfaction. Then, I want it to present me the results so that I can see the most likely candidates.
Guess what, even at that point, I may not take action though. I likely still will want to discuss the results with the branch manager(s) because the data may say one thing, but reality could still be something different. Once I am comfortable that I know the root cause, then I need to figure out what action (or actions) to take. Maybe the system can suggest potential actions. Maybe there is an online community of senior level bank managers where I could get advice and insight (yes, there could be a social software angle here)? Maybe I would talk to my HR business partner? Maybe all of the above? This is a discussion for another blog post.
The next time you watch a vendor demo of reporting and analysis. You will likely see them do a drill down to find a root cause (in fact, the data is set up so that they will easily find a pre-determined root cause). In reality, they are just applying the intelligence manually. Vendors, consultants and HR professionals have the knowledge to know where to drill down, where the potential root causes may be. Why can’t that knowledge be captured in a system? Why can’t someone design a business intelligence system with actual intelligence about the business domain built-in?
That is the end of the rant. Am I being too harsh? Am I asking for the impossible? What do you think?
Category: Uncategorized Tags:
12 responses so far ↓
1 Kevin Yapjoco February 18, 2010 at 9:15 pm
What you’re asking actually borders on Artificial Intelligence. I’ve never seen Business Intelligence tools do that.
We’ll probably see something like this in a couple of more years as demand for smarter systems increase.
2 TechSphinx February 19, 2010 at 8:20 am
The first commenter, Kevin, is correct. What you are asking for truly depends on the specific context of the using organisation and its particular data. The best any software vendor can possibly do is to provide a rich toolset to allow users to get to a rich set of data. That, plus a number of pre-built data views/reports/dashboards to show how it is done. True, some fail at even providing this, but I wish consultants would actually explain to user organisations that this responsibility is theirs, not that of the software provider. Consultants should also build practise areas to help users create meaningful BI from whatever software they are using. I am not sure why they don’t do more of this.
3 Sability WFM February 19, 2010 at 9:26 am
ditto, ditto…. In a recent position paper, we also recently identified Business Intelligence as one of the top 4 unrealized ROI’s from Enterprise Workforce Management implementations.
My biggest complaint about the vendor BI presentations isn’t that they show prepared data, its that they imply that the BI tools bundled with their software are really easy to use. I haven’t met many department managers who can create anything more than a list report of historical data using these tools. Again, companies should go into this with their eyes wide open. You will need to train IT resources on these tools or hire consultants to build the reports.
http://mailimages.sability.com/e0003/pdf/Sability_WFM_The_ROI_You_Didnt_Get.pdf
4 Jim Holincheck February 19, 2010 at 10:45 am
Kevin,
I do not think it is artificial intelligence (though the system learning part definitely is) as it is like an expert system plus a lot of statistical analysis (to understand linkages and correlation/causation). The point I was really trying to make is that the starting point for BI has to change. Thanks for your comment.
5 Gilles February 19, 2010 at 10:47 am
There is something called “guided analytics” where the vendor consultant talks to the customer to identify possible root causes. These talks can result in guided analytic paths. If x happens, than possible causes could be y and z, take a look over there!
But it requires extensive talks with the customer
6 Jim Holincheck February 19, 2010 at 10:48 am
Techsphinx,
I agree to some extent, but there is a starter set of questions that would be common to most roles in most industries. You could start there and then add to the framework over time more job and industry-specific questions and metrics. I did not mean to imply that it would be a fast journey to get to this type of solution, just that the current journey really is not heading in this direction. Thanks for your comment.
7 Jim Holincheck February 19, 2010 at 10:50 am
Gilles,
Yes, that is more of what I am trying to get at. However, I do believe that there are a common set of questions that can be used as a starting point with customers. Thanks for the feedback.
8 Jim Holincheck February 19, 2010 at 10:51 am
Sability WFM,
Totally agree. That was really the point of the post. There is lots of room for improvement. Thanks for participating in the discussion.
9 Gilles February 19, 2010 at 11:00 am
@ Sability WFM always, ALWAYS ask for proof of concepts, product demos with your own dataset.
Demoing is easy with your own clean and prepared data, but real data is different!
10 Beth N. Carvin February 19, 2010 at 1:40 pm
I’m so glad that you mentioned digging into the exit interviews because as I was reading the scenario, I kept thinking, “Look into your exit interview reports, Mr./Ms. Regional Branch Manager and you’ll find out exactly why your Customer Service ratings are down.”
We see organizations using exit interview management systems doing exactly this. These sytems don’t get a lot of attention in the HR technology discussions (maybe because they are offered by non-VC backed companies?) but there are many companies getting the kinds of benefits that you speak of. Including suggestions for solving the identified issues.
True some companies do a better job at integrating them into their corporate problem solving than others. And also true we have a long way to go before all companies are using these systems as a matter of course. It is interesting, though, to learn from the organizations that are doing root cause analysis with this kind of tool.
*Note: In the spirit of disclosure, I have a vested interest in this field as CEO of Nobscot Corporation.
11 Naomi Bloom February 19, 2010 at 5:39 pm
Jim, as you know, I’ve been pushing the idea of embedded intelligence throughout the HRM delivery system’s platform for what feels like forever. At the core, your post, which is excellent, suggests the possibility of embedding the kind of investigative journalism which is at the heart of root cause analysis as well as the solutioning that good consulting methodologies contain right into the HRM BI, and I believe this is a very doable objective, obviously without limits. In my HRM Business Model “Starter Kit’s” metrics component, there are a ton of suggested metrics, but the real guts of the needed embedded intelligence comes within the process model component. We’ve got our turnover stats, but how do we decide if those are good or bad numbers? Are low performers choosing to leave because we’ve made it clear via a great performance review process and frozen salaries that low performers must either up their contribution, face stagnant compensation or, worst case, they’ll be counseled out? Are high performers choosing to leave because we haven’t made a big enough compensation distinction between low and high peformers? While consulting — really good consulting — can be a terrific help with getting to the story behind the metrics story, I think we can do a MUCH better job of capturing that intelligence and embedding it with the metrics manipulation and drill down. If TurboTax can do it, so can HRM software. Have a great weekend, Naomi
12 Mike Smitheman February 22, 2010 at 4:13 pm
I agree that there needs to be an element of “human thinking” within any successful analysis, however, predictive analytic tools (which is what I would call what you are referring to) can and should help managers highlight obvious exceptions within organizations. For instance, if employee absence is above average, a good analytics solution should show me the top 5 areas ( or employees) as it relates to absences this month, the types and the reasons and I can then ascertain, through the data, why it’s happening. The BI solution can help me sort out the “why” with scorecards, exception reporting, charts and dashboards and above all else, make it simple and fast to get to the information needed. I think BI nirvana may be a mixture of a manager’s own “field of dreams” with the use of scorecards and predictive analytics to pinpoint the problem and charts and graphs at their fingertips that answer the questions they suspected in the beginning. One can’t happen without the other.