There has been a great deal of interest in analytics in the recent past. Technology and algorithms have evolved dramatically in the past decade but they are still being applied to the improvement of old processes. The best analytics outputs are useless if the organization isn’t ready for it!

Analysis [uh-nal-uh-sis]

1575-85, Greek, equivalent to analȳ́(ein) to loosen up ( ana- ana- + lȳ́ein to loosen) + -sis -sis

Statistics [stuh-tis-tiks]

C18 (originally “science dealing with facts of a state”): via GermanStatistik, from New Latin statisticus concerning state affairs, from Latin status state

Very often people tend to confuse analysis with statistics. Talk about analytics and immediately the discussion turns to trends, statistics, big-data and these days Machine Learning and AI. The tendency is understandable – data analysis is an important part of analytics and often gets mistaken for the entire process itself.

Showing people how fast you can create a pivot table is way cooler than just be seen as staring at the data, wondering what you are trying to do in the first place.  Statistics plays a significant role in the analytic process, but those calculations only make sense in the context of our understanding the interactions and relationships of the elements in play.

Gartner has put up a rather nice image of what drives the success of Data Science and right up there (as a happy coincidence) are (a) Ask good questions and (b) Know the constraints


HR Analytics:

Analytics, when applied to humans at the workplace, have to deal with data from disparate sources – employee records, evaluations, surveys, social interactions and the list keeps on growing. In many organizations, the process gets stuck right at the data wrangling level since disparate systems operate in silos and just reconciling the information to get a complete view of the employee becomes a nightmare. Metrics reporting and preparing rudimentary dashboards then are made to pass for analytics. These so-called ‘analytics reports’ are then little more than a collection of pretty graphs and tables with no obvious value addition or insights.


The analytics value chain:

Adapting Michael Porter’s concept of a value chain to the analytics process: A value chain is a set of activities that are performed to deliver value (product/service) to those that intend to consume it. We could then say that there are five core links in the analytics value chain.



Moving past the challenges of Data Collection and Data Wrangling we get to the primary three levels of Analytics

  • Descriptive Analytics: Honestly in most cases, this is just a fancy term for reporting of metrics. These are usually trends of historical data and some extent of correlational observations. This forms an important starting point for establishing an analytical mindset – because after all “what doesn’t get measured doesn’t get fixed”.
  • Predictive Analytics: This gets a little more interesting here. We now start building models using statistical and probabilistic techniques. Predictive modeling is now buttressed with ML algorithms to improve accuracy (that is for another post). At this stage of analytics capability, depending on how good your past data is – you should be able to develop models for good employee fit based on CV patterns, models for resourcing/attrition etc.
  • Prescriptive Analytics: This is the new frontier. At this level it is no longer about making models to predict outcomes based on past behavior, it is about predicting based on new inputs/alternate scenarios that are unknown as of now. A good example would be how alternative employee engagement models would drive productivity in the workplace.


The value of analytics:

We looked at the value chain of analytics, but honestly, you wonder – why all this trouble (and it is not easy for sure)? What value does analytics really provide?

Well aside from making you cool and a hit among the colleagues for being a data-warrior (or ninja or whatever martial art form excites you) precious little (kidding!)

Most of the people who set out to be data warriors end up torturing the data till they get what they want to see. (Some of the ludicrous conclusions published, based on flimsy data or worse, are no less than war-crimes if you think of it)

The true value of analytics (when done right!) is essentially of two forms (and that too they eventually blend).

  • Economic: Insights from data analytics can lead to improved customer service, better processes, market reputation among other things. These are difficult to correlate directly to financial impact but eventually, through these, there is a positive business outcome (brand reputation leading to more sales, increased stock price, cost savings in the long run through connecting intangibles etc) which translates into financial impact.
  • Financial: Financial impact are those that have direct improvement of financial metrics through insights gleaned from analytics. Finding hidden insights into causes for attrition could lead to solutions that stop the bleed which translates into an immediate impact on the top line and bottom line through increased productivity and cost savings in hiring and training of new resources.

The biggest challenge when working with people analytics are those “little things” called “emotions” and “free will.” If a server in the cloud farm behaved in one way today, chances are likely it will behave exactly the same tomorrow. Not so with people!

Descriptive analytics can only tell leadership about what has happened in the past. To be able to predict future outcomes and to establish linkages to seemingly unrelated events makes HR Analytics both an art and a science.

To repeat a cliché’ – change is the only constant when it comes to the realm of human capital. Your recruitment and retention strategy for last year might be rendered totally ineffective this year because of moves made in technology, labor markets, company dynamics, business focus or a hundred other reasons.

So how can an HR organization develop an analytical mindset?

Three essential steps:

  1. Get leadership buy-in and establish strategic direction: Analytics cannot be done effectively in isolation. To get insights you need to share and receive data from other sources (other departments, internal/external social media streams, market information, technological information etc.) To break down the silos, you need to ensure that top leaders and BU heads are on board with the strategic intent of this transformation very clearly established. Rome wasn’t built in a day, and neither will a data-driven analytic mindset. Make a roadmap, get buy-in and dig in deep for the long haul!
  2. Know the constraints and Ask the right questions: People often get excited about the technology. Big Data, ML, AI, Data mining, Deep Learning. Stop! It’s not about the technology. Technology is an enabler, not the solution. You need to know what you are looking for. The insights you get from your data depends on your knowing what you are looking for. And then comes the twist – of knowing your constraints. It would good to know everything you want to, but be realistic. There are several constraints you will work with at any given point of time – data sanity, technology, privacy concerns, time, budgets. Knowing where to draw the line and being aware sometimes ‘good is good enough’ can take you a long way ahead in your journey.
  3. Bridge the analytics capability gap: As you move up the value chain, both the complexity and the value derived go up (exponentially so). But don’t be under any illusion of how difficult this journey can be and hire the right resources. Just because someone is “passionate about people” doesn’t make her a right fit to run an advanced analytics program. Putting the wrong people and not giving them clarity about what they are trying to achieve will set you back on the transformation (sometimes fatally so)


There is little scope to short circuit the move of an organization up the value chain. Organizations are tempted to throw money at the problem but that is not really going to help in the long run. You could buy systems that promise the Sun and the Moon (and indeed some of them are pretty advanced) but always remember the notion of GIGO (Garbage In Garbage Out).

Till you know what you are looking for (the right question), what you need to answer the question (your data sources), how you intend to get what you need (the data wrangling process), analytics will leave you feeling meh!

And sometimes the biggest challenge to a transformation comes from the leadership. If there isn’t complete buy-in, they might just end up being the biggest obstacle and you might end up sending the same message as O.H.Perry did in the Battle of Lake Erie

We have met the enemy and they are ours.

Two ships, two brigs, one schooner and one sloop.



We love discussions! Please do leave a comment on what you think about this post.

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.