A while ago, I posted some thoughts on a problem which I see throughout the industry (and elsewhere). Management Information (MI) is, in effect, a broad church of statistical inference from which management make decisions. That’s pretty much it.

In the previous article, I looked at some of the logical flaws which can really screw up that decision-making process. My main point was that most management don’t have the necessary training nor the motivation to really dig beneath the surface to understand the calculations and whether they actually have any valid meaning whatsoever.

This article follows on somewhat from the original. If you haven’t already done so, I’d recommend you read it to get an idea of where I’m coming from. Done that? Okay, I’ll begin.

A Complete Tool

Once upon a time, around eight years back, I was a technical consultant working on a consultancy tool for a major accountancy firm. Much effort had been expended in creating an application which would allow HR departments to assess role profiles and utilisation. However, this application was founded on some very ropey logic. You see, there are skills that people have and others that people don’t have. Most people don’t understand statistics, yet our lives are governed by the use and interpretation of statistical data. The problem is that (in my opinion) the very people who have the strongest grasp of statistical validity and ‘appropriateness’ are the very people whose opinions are ignored by the often rather clueless middle management (also known in many larger firms by the collective, shady notion of “The Business“) and as a consequence nothing useful can be derived from the MI.

As an aside, this application was subsequently used widely, and the initial marketing proposal for said system had the slogan: “XX XXXXXXXX - A Complete Tool” - which, where I come from, means something entirely different, but somehow it was quite apt!

No Time Like The Present

As a techie-by-trade, I’m often asked to help sort out fiddly little problems. I was approached recently by a friend whose role is basically MI. He was performing an audit of a contact centre KPI MI tool, which was basically a commercial application which had been hacked as if it were attacking one’s family. There was a calculation determining working time periods from logs. Like a crucial cog in a complex machine, this calculation was critical to many other calculations, which assumed that the values returned would be meaningful. However, it contained a Severely Erroneous Assumption.

The objective was to aggregate employees’ shift times, but took into account the fact that in any given shift, there was an automatic 1/4 hour break. The calculation went something like this:

NetShiftPeriod = (EndTime-StartTime)+0.25

…and this figure was summed for each individual for each day or somesuch.

However, the Severely Erroneous Assumption was assuming that 0.25 represented a quarter hour. It did not.

In Oracle, the particular field-type expresses times as their fractional part of a full day. Thus, one hour would be 1/24. or ~0.042 days. A quarter hour would be one quarter of this amount, i.e. around 0.01.

A typical shift being 8 hours, or 0.33 days, what was happening was that the calculation effectively skewing the results so that dependent calculations (e.g. agent bonus, KPI etc.) were way out. And they’d been using this as the basis of performance related pay, bonuses, promotions and so on, for at least eighteen months.

Scary biscuits, as they say…

Breeding Like Bunnies in Excel Hell

These days, the corporate world is run on a flimsy, fragile infrastructure of ill-defined, bug-ridden Excel spreadsheets. These nasty creations breed like rabbits, quickly undermining any real MI strategy. They are normally cobbled together in lunch-hours, with untested logic, false assumptions and are normally of extremely dubious design. However, although often Evil Incarnate, they are sadly unavoidable and another item of evidence supporting the MI Lie.

I recently had occasion to help a small team out on a staff movement audit tracker spreadsheet. Written, of course, in Microsoft Excel, and underpinned with its partner-in-crime, Microsoft Access. Originally written by Blind Io on stone tablets, this spreadsheet had the feel of a holy relic. It was revered and respected, but also feared. It had been around since day dot, and had stopped working. I’ll spare you the details (lest you go blind like dear Io) but in a nutshell it was the usual story:

• Junior Staff Member (JSM) tasked with creation of spreadsheet;
• JSM starts spreadsheet, but does not complete;
• JSM leaves for other role;
• Other Staff Members hack it about a bit depending on the whim of the day (repeat, rinse);
• Spreadsheet is a mess, but is Depended Upon, management in a pickle because it ‘disnae work nae more’;
• Someone like me gets called in to fix it, end up in therapy for months as a result.

Yes, that’s what I call Excel Hell. It’s a dark place. Avoid!

Summing Up

The MI Lie is a worrying conclusion to an inevitable trend. Too many chiefs, not enough injuns; but what can we do?

Sadly, there is no answer; as a wiser man than me once wrote, “There are no silver bullets“. We must tackle each and every battle in turn. Management, and the nebulous Business, need to be informed that the utmost care must be taken, and educated in the flaws that will result otherwise.