Measurability operates over an incredible range. It’s the ‘gravity’ of the metrics universe. Why do I say so? Because it is a quality attribute that manifests itself at business-level decision-making, without being constrained to that realm in the least.
Measurability permeates everything; it goes past the business level, down through engineering and development activities, and even further still, to actual infrastructure.
And what exactly is measurability, you might ask? It is the ability to quantify the behaviour of a system along with its most important outputs or results. This ‘quantitative’ aspect suggests the use of statistics and tabular data, the gathering of which exposes a well-known-yet-oft-poorly-solved problem: with a probability that is almost certainly ONE, your system has NOT been designed to fess up the numbers you need.
Beginning with some almost-universal truths
First of all, dealing with analytics means wading into chaos. It is for this reason that you are employed at all, O analyst, so whinge not. Instead, consider the following truths:
- Noone thought about metrics when your system (eg a software application or even an entire business system… such as a franchise) was built.
- Also, noone thinks about metrics during the opportune moments, like when the system is about to undergo routine change, or gets updated.
- Even, now, when it has become abundantly clear that business owners have some very pointed questions to ask about the business, which questions can only be answered by joining the dots between several sub-systems; which sub-systems are hell-bent on providing staggeringly different accounts of the same logical transactions… even now, noone does anything about metrics.
- And the final truth: You have a report to put together before monday. On metrics.
Are all these things not true? Do I lie to you, analyst friend? Of course not. Now that you know that I know what you’re going through, perhaps you’ll trust me further:
Retro-fitting measurability after the ship has flippin’ well sailed
PHASE 1 : Lick your wounds. Remember what it tastes like.
Do nothing in this first phase. That’s right, I said do nothing. But find patterns, dammit. Each time you sob and sweat over a spreadsheet, take note of:
- what data elements keep recurring, regardless of the business question(s) ?
- what calculations keep recurring?
- what report structures / views keep recurring, and which ones communicate insights the most effectively?
- which subsystems agree well with each other and which ones don’t? Which ones can you bypass entirely for certain types of data? Which ones are crucial?
- How often are you being asked to garner key insights and metrics? Is there a cycle? Could you build a schedule around it?
- Is there anyone in the business team who always catches your trip-ups? Are there people who easily find holes / inconsistencies whenever you hand them a report? Make a note of them. They must be assimiliated become the test harness for your fledgling effort in the war on confusion. Give them every opportunity to kick your ass.
PHASE 2 : Get organised, get automated
List all the things that you have to keep doing and then automate them where possible. If it’s not possible to automate them, then find the fastest way of doing them. A little creativity goes a long way here:
- low hanging fruit: create useful report / view tables in your database
- save repeated sql queries and automate them where possible
- create reusable document templates and worksheets set up so that you just need to populate it with the right data
- become an “export as…” junkie. If you come across any tools, subsystems or dashboards that give you the opportunity to export to a useful format, make this your preferred way of transferring data between systems. It also minimises transcription errors.
I’m sure there’s many more timesavers you can think of. But remember your ‘test harness’ folks from STEP 1? Good. Always make sure they see your deliverables FIRST. Keep feeding them your evolving analytics… When you get to the point where they start to purr contentedly (instead of tearing a strip out of you and your reports), you may proceed to the next phase.
In my next post in this series, I’ll go over the remaining phases of getting up to speed with measurability.