Cognos Analytics

 View Only

Shocking Data Dashboard Confession: “I’ve Deceived You”

By Akshata Pai posted Mon February 13, 2017 06:00 AM


Shocking Data Dashboard Confession: "I've Deceived You"

Years of mistaking correlation for causation

When your data dashboard reaches full AI maturity, it will have to apologize: “Sorry I’ve been deceiving you.” Without you knowing, it’s delivered years of misinformation.  That dashboard solution that promised it would always be, ‘intuitive, interactive, and drag & drop’ will use its AI capability to say, “It’s not my fault; I was never equipped to show differences between causation and correlation.”  Odds are high that your data discovery vendor has been ignoring analytics.

That data ‘snapshot’ in your dashboard did not make you a better leader or decision maker.  In fact, it deceived you into seeing correlation instead of causation—a mistake that has sunk many careers.

“…dashboards are poor at providing the nuance and context that effective data-driven decision making demands.”
-Harvard Business Review 13Jan2017

The Harvard Business Review (HBR) offers caution to executives everywhere; your dashboard can mislead you.  The HBR article3 Ways Data Dashboards Can Mislead You” makes its point clear; you need both predictive and prescriptive analytics to get dashboards right.   Many data discovery vendors have been ignoring this for years by offering dashboards that only deal with current or past activity.

“Make your data make an impact “
– Data discovery vendor

The tagline offered by one vendor makes it sound like visualization is all you need to prove your point. But what if the visualization misleads you?  In the graphic below, you’ll see an obvious pattern between people who die from becoming tangled in bedsheets and per capita cheese consumption in the US (Source: Spurious Correlations).  Most dashboards and data discovery tools deceive you into the belief these groupings are interconnected.  There is correlation in values but it does NOT mean that change in one variable (cheese) is the cause of change in the values of other variables (death).  You won’t know, but your dashboard was misleading you.

         “It’s far too easy — and unfortunately common — for managers to interpret the groupings in a dashboard as causative when they may not be.”
-Harvard Business Review 13Jan2017

When you’re caught up in the elegance of data dashboards, it’s easy to dismiss the underlying analytics.  An attractive chart describing what has happened is undoubtedly less intimidating than an attempt to understand what will happen.   But there’s no longer an excuse for mistaking the two. Past activity is not a predictor of the future.

In the past, analytics might have been inaccessible to senior management, but with new natural-language solutions, that no longer needs to be the case. Don’t wait for your data dashboard confession.  If your dashboard could talk right now it might say, “Save yourself and see what analytics can do for you.  Start a free trial of Watson Analytics before it’s too late!”