Philosophy of Analytics, Lesson Three: When

Introduction

In lesson one I discussed the importance of keeping in mind the purpose of analytics when you are starting and designing and running an analytics project (the why of analytics).   In lesson two, I discussed the importance of understanding the who behind analytics: who is doing it and who is it for.  In lesson three, I am turning to the when question: when should analytics be used?

When: Always, Except When It Doesn't Make Sense

When should one use analytics?  Let's go back to the definition:

The purpose of analytics is to (1) justify or change beliefs with the use of evidence in the form of data (2) in order to generate true (or truer) beliefs that can then be used to make decisions related to one's goals, and that, (3) because they are more in accord with reality, are more effective in bringing about one's goals.

So analytics is useful in any situation in which one needs to justify or change beliefs by using data to generate true beliefs so that one can make decisions to achieve one's goals.  In short, whenever you have a decision to make, if you want to have a better understanding of the facts/evidence/reality surrounding that decision so that you can make a good and justified decision, you need analytics.

Now when is that?  All the time!  We are constantly gathering bits of information (data) ourselves as we go about our day regarding how the world works and how we should operate in it.  Those of us who can do this well, generally get along very well in the world.  Those of us who can't (who deny reality), don't fare so well.  Doesn't it make sense then, in a business or organizational context, to try and gather and analyze data related to that context so as to make justified decisions that are more in accordance with reality so that we can succeed in meeting our goals?  Of course! 

But lest we swing too far in the other direction by gathering and analyzing everything without limit, here are some things to consider first in making the decision to undergo any sort of analytics project:

• Costs/Benefits: what will the cost be of doing the analytics versus the benefits?  If gathering and analyzing the data will be more costly than making a (wrong) decision using the data, then it is not worthwhile to gather the data.  Simply make the decision and move on.  If you are wrong, no big deal.  Remember that you have tradeoffs to make with your time and money, and you are wanting to invest your time and money in ways that will, in the long run, save you time and money.  So only do analytics projects that will be a good investment in terms of achieving your goals.
• Order of analytics: data science is all the rage.  But should you take on a predictive analytics project before you have basic reporting?  Probably not.  There is an order to setting up your analytics framework that needs to be followed.  Why?  For starters, how can you predict anything if you data is bad?  Garbage in will produce garbage out, so you need to first make sure the data is good, which means you need a good and clean data pipeline.  And how would you even know what to predict if you don't have a good handle on which KPIs the organization is targeting?  And you don't want to have to pull together disparate data sources each time you build a report; these should be integrated in one place.  In short, do first things first to set yourself up for future success.  The following should happen roughly in order:
1. Database pipeline/database management
2. Data warehouse/data integration
3. Basic reporting from data warehouse/databases
4. KPI/metric reporting; dashboarding; visualization
5. Predictive/advanced analytics; statistical analysis
• Actionable insight: there is a story about a company that discovered that a particular variety of its wine sold particularly well on Tuesdays.  Isn't that great?  Maybe.  But could it take any action on this little nugget of insight?  No, as shelves are not currently stocked to fluctuate by day.  It could not do anything with this insight.  In designing your analytics, ask yourself, "what decision can/will I make with this information that I could not (easily) make now?"  If you can't come up with any important decision which requires the analytics you are going to implement, it is a waste of time and money.  Do not spend your time developing reports, KPIs, metrics, and predictive models that won't influence decisions and which nobody cares about.
• Appropriate scale/software:  do you need a Hadoop cluster? for most purposes, the answer is no.  If you haven't heard of Hadoop, the answer is almost certainly no.  Do you need to get a Tableau license?  If you are fine with developing reports in Excel/PowerBI, then you can save a lot of money by doing this instead (or even Google Sheets).  Should you run your reporting out of spreadsheets without a database?  Probably not.  Think about what your organization needs right now, and what it will need in the future.  Then, build out your architecture to meet the current needs in such a way that it does not preclude future needs/expansions.  Use the software you need (not more and not less) to be effective in doing and socializing your analytics.  Don't build a mansion when you only need a house, but if you foresee needing a mansion, make sure you can add onto your house to turn it into a mansion.

 There are other recommendations I could make, but a blanket statement to cover all sorts of pointers is this: be smart in your use of analytics.  Do analytics when it makes sense, is a good investment, will yield good returns, and lead to better decisions that are worth the effort.  Think about the long term goals and needs of your organization, and design the analytics framework to meet those needs in such a way that it can scale and grow as appropriate.  Be efficient.

Conclusion

Use analytics when it makes sense to do so.  Don't use analytics when it doesn't make sense to do so.  If in doubt, do some minimal analytics work to see whether you get any return on it, and then expand if it is found to be useful.