??Ask this *one* question to save heartburn (and millions of dollars) in your data science journey

Hello!

I'm excited to bring you my newsletter on LinkedIn! I've been publishing it for over a year on Substack and the response has been great. It will now be available on LinkedIn as well.

A quick lay of the land. Each edition will have three sections:

• Spotlight idea:?This will be the story in focus, the main act.
• Industry roundup:?2-3 useful things I stumbled upon, and why you should care
• My updates:?Quick snippets on happenings from my world - articles, talks, and videos.

I will keep these short and snappy to make the newsletter byte-sized. As always, I'd love to hear your feedback in the comments and strike up a conversation. Let's begin.

I. Spotlight:???Ask this *one* question to save heartburn (and millions of dollars) in your data science journey

“I want my organization to be data-driven,” said the CEO of a medium-sized manufacturing organization. This was the early stages of the pandemic, and we were on our second Zoom meeting.

His organization had begun its digital makeover when COVID hit. With manufacturing plants spread across the country and a customer base that was increasingly digital, the CEO understood the importance of data.

His enthusiasm was evident when he asked me, “How can my team use machine learning to improve business outcomes?”

While executive interest in data was music to my ears, I simultaneously sensed alarm bells going off in my mind.

In my earlier conversations with his team, I found that most of their data were still on paper. Their small digital footprint was handled manually in spreadsheets. But, the business teams had little interest in excel reports or using them for decisions.

The organization lacked quality data and the maturity for simple descriptive insights, whereas the CEO was aspiring a big push into data science.

Such disconnect is widespread in organizations.

How do you resolve this unrealistic expectation to avoid the likely failure and disillusionment with data science? This framework will help.

(Picture: Data quality vs Analytics depth - 2x2 chart)

This chart shows?data quality?on the x-axis and?analytics?depth?on the y-axis. Here’s what the four quadrants mean:

• Zone of stagnation?(bottom left):?Here, both data quality and analytics depth are?low. This is where all companies get started. Unfortunately, many stagnate there when they don’t prioritize data engineering and investments in analytics.
• Zone of disbelief (top-left):?Here, data quality is?low, but analytics depth is?high. Imagine running predictive algorithms on questionable data. This always leads to user disbelief.
• Zone of complacency (bottom-right):?Most large organizations live here:?high?on data quality but?low?on analytics depth. They use gigabytes of neatly curated data to churn out elementary MIS reports while refusing to step up their analytics maturity. What a waste of potential?
• Zone of nirvana (top-right):?This is where all organizations aspire to get to. They have high availability of good quality data and can analyze them for deep insights. Getting here is a marathon.

The manufacturing CEO attempted to get to the ‘zone of disbelief’ - acquiring machine learning ability without raising their data quality.

To highlight the disconnect, I asked him, “Let’s say a machine learning model reveals a groundbreaking insight about your production process. Will you make a high-stakes decision based on this finding?”

The CEO thought for a while and said, “I’d be cautious - it will make me curious but not confident enough to act on the recommendation.”

When probed further, he finally said the magical phrase, “I wouldn’t trust the data.”

When a hidden analytics insight fails to raise our curiosity but instead raises doubts about the data, we might have just put the cart before the horse.

I politely told the executive that they might be early for machine learning.

So, the next time you find an organization in the ‘disbelief’ zone demanding advanced analytics (or AI), ask them this one question:

"Will you make a high-stakes decision if your algorithm stumbles upon a ground-breaking insight?"

Unless they say "hell yeah," they are not ready for advanced analytics. Save them time and money by putting them on the path to cleaning up their data first.

—

How can a firm starting in the ‘zone of stagnation’ reach ‘nirvana’ without slipping into the zones of ‘disbelief’ or ‘complacency’? No, they need not hit the brakes on analytics—more in the next newsletter.

II. Industry Roundup:

1. Article: How to make your Chief Data Officer’s life a misery

9 minutes | LinkedIn | Ian Thomas

Though the title of CDO is steadily gaining prominence, the role continues to be poorly defined. This post does a great job explaining the common mistakes that lead to short, ineffective tenures of leaders who take up this position. It concludes by sharing why there is hope in sight for companies and how things could settle down.

->?Read the Article

2. Article: Top 12 Strategic Tech Trends for 2022 and Beyond

5 minutes | Forbes | Peter High

Gartner has identified 12 broad trends that will act as force multipliers of digital business and innovation over the next five years. 6 of these 12 trends fall directly into the realm of data and analytics. The research projects the highest growth rate for IT budgets in the coming years. Exciting times indeed. Check out these technologies.

->?Read the Article

3. Article: Why is data missing from the balance sheet?

3 minutes | CFO University | Prashanth Southekal

Today everyone understands the value of data, but why isn’t it in the balance sheet of companies? This article lists five reasons why it’s tricky to list data as an asset - 1) difficulty in valuation, 2) inconsistency in depreciation, 3) needs heavy context, 4) may not be an acquired asset, and 5) its value flips based on compliance or the lack of it.

->?Read the Article

III. From my Desk:?

1. Article:?Are you guilty of these 3 cognitive biases in decision making?

8 minutes | Forbes

We make around 35,000 decisions every day, but many of these are biased. Why do we end up making such poor choices? "Our irrational behaviors are neither random nor senseless," said Dan Ariely. I share how we can guard ourselves against making such systematic mistakes.

->?Read the Article

2. Article: How to do data science when you don’t have big data

5 minutes | The Enterprisers Project

If you think that a large, tidy data warehouse is a pre-requisite for value from analytics, think again. I share four ways to pursue analytics, even with small spreadsheets of data.

->?Read the Article

3. Guest Lecture at Penn State Univ: Business forecasting in real life

40 Slides | Penn State University - College of Info. Sciences & Tech

In this guest lecture for the Data Science Capstone at Penn State, we covered the business applications of forecasting techniques. Using case studies from our work at Gramener, we explained the machine learning lifecycle, emphasizing practical challenges faced, where we messed up, and how we recovered. I enjoyed interacting with the bright students at the College of IST.

->?View the Deck

Found this useful? Please hit subscribe here to receive future editions. If you'd like this delivered to your inbox, please sign up for the newsletter on Substack .

Thank you for subscribing and reading the newsletter. I appreciate your attention,

Ganes.

My Website ?|?Twitter ?|?YouTube |?Gramener