The Fallacy That Everyone Wants (And Needs) Self Service BI
By Alan Hylands
This is part 2 of a six part series titled “Frustrations of the Data Analyst Life…And How To Fix Them“
The rise of Just Do It Yourself
We’ve grown used to doing things for ourselves that used to be done for us.
We pump our own petrol/gas.
We poke at buttons on an ATM machine which gives us cash from our bank account. Then we continuously flatten out and re-feed notes and cheques back into it to deposit money back into that very same account.
Take the kids to McDonald’s for a “treat” and you can swear freely at the multitude of options on the self-service touchscreen. Is this really progress from getting eye rolls for your indecision over the Big Mac meal or salad from the teenager behind the counter?
Don’t even get me started on the self-service checkouts at the supermarket.
“Unexpected item in the bagging area”. Sirens going off. Staff jumping over barriers to get to you and wrestle the punnet of strawberries from your hand.
(It’s a bag FFS, how could it be unexpected in the BAGGING AREA?)
Self-Service In The Workplace
Self-service isn’t a new notion then. So when did it grow into the movement that everyone should be doing their own data analysis and reporting?
When I started my first data miner job back around 2005, I became the de facto Master of the (Business Objects) Universe for a healthcare trust.
I built new universes in Business Objects and took over admin for the existing ones. I also built out reports that secretarial and nursing staff could run themselves.
– How many patients had been waiting over 3 months for their appointment?
– What was the age breakdown of our patient list over the past year?
– Which treatments did our rheumatology patients get?
Etc etc.
All relatively simple and straightforward. Either easily pulled by the user, using the drag and drop measures, or set up centrally (by me). They were then given to them to run any time they needed it.
Secure, consistent, measures set up and managed centrally.
Perfect.
It was a relatively small trust though and the set of systems we supported was also limited. The potential for catastrophic mis-communication of figures was reduced.
This isn’t always the case in the private sector though. Especially when an organisation gets over a certain threshold in the number of business users given free rein to run their own reports.
The potential for nefarious motivations to creep in is inevitable especially if bonuses and performance reviews are based on what the numbers say (or don’t say).
With that in mind, over the years I’ve moved well away from the notion that everyone should have self-service BI.
Who is it good for?
As an analyst it seemed like a magic wand to ward off the mundane, small-time MI requests that sucked up time and resources to complete Especially those that were little more than pointless busy work to keep the requester’s manager off their back and make them look busier than they actually were.
As an analytics manager, it was tempting to offer an alternative to working through these painful requests to dig out the useful requirements. The downside started to become apparent pretty quickly though.
Customers can’t always be trusted to know what they want. More importantly, the role of an analytics team isn’t to sit there like a dumb terminal waiting for the first passer-by to batter some basic instructions into us.
At senior management or executive levels with responsibility for reporting and analytics, it’s even more dangerous. I can’t see how anyone would trust any level of self-service BI reporting that wasn’t so stringently policed it might as well have came from the dedicated BI team themselves.
The downsides are stacking up here
The potential for incorrect figures to impact the business dramatically becomes exponentially greater, the more people you allow to plough their own furrow.
Some actual data on the proportion of business users who actually make use of a full self-service BI platform when it’s made available to them would make for very interesting reading.
If you build it…
Once the initial enthusiasm for the new tool subsides, I doubt you’d find even 1% becoming power users who know what they want and how to get it without any support.
A few more might need a little hand holding but be fairly competent in running their own general reports. Another smaller subset will jump straight into brain dead “just dump everything you have into Excel” mode.
The vast majority, however, will quickly just fall back into the mindset of “just send me the fucking report”.
And you’re back to square one.
The memory of some reports that were sent back to me in my early analyst days still haunt me. Those joyous ones when I was ordered to add another column to calculate a percentage because the user was just too lazy to do it themselves.
By this point Excel had been a staple of the modern office for over 25 years. If you couldn’t be arsed working out how to do a percentage column with the whole world of info available on Google, how could I ever trust you to accurately run and communicate your own reports off the whole MI system?
I’ll make a bold statement on it then:
There should always be a dedicated BI function setting up and overseeing the reporting.
They should be the internal police force to ensure consistency across the organisation. If self-service is permitted, it should be for properly skilled, experienced power users who have demonstrated they have the wit to use it correctly. And even then I’m still sceptical.
With great power comes great responsibility.
Uncle Ben knew. Disregard his advice at your peril.
-> Read Frustration #3: How To Deal With The Urgent!Urgent!Urgent! Boy Who Cried Wolf