Building an Effective Analytics Strategy

I wish all the readers a happy, healthy, and prosperous 2023.

In response to my article titled?“The first 90 days as CAO”, I had several people asking about ways to build an Analytics Strategy. Every organization would have a different Analytics strategy based on its industry, size, technology infrastructure, business priorities, culture, and many other factors, so there is no ONE way of coming up with an Analytics strategy. ?However, there are guidelines that can help one build an Analytics strategy.

1)?????Technology Strategy is not an Analytics strategy: I have seen too many roadmaps that talk about consolidating enterprise data into a modern data warehouse, with BI and AI/ ML stack to support it. Kudos to technology firms for selling this as an Analytics strategy, but it is not.

2)?????Analytics is needed wherever there is a selection. Very early in my career, I had Shrikant Dash , tell me that one needs Analytics only when a selection is being made. When a selection is made, there is an automatic rejection taking place. When a selection is made, adverse selection happens as well. Making these selections better is where Analytics comes into play. Simple and powerful advice that has helped me in every role I have been in.

When building an Analytics strategy, one needs to focus on making better the key “selections” the organization is making.

3)?????Go for the jugular. A typical organization would be making 1000s of decisions, from product design, pricing, product placements, marketing promotions, supply chain, and customer support to everything in between and after. The key is to identify the decision points that need analytics intervention. Not all decision points are made equal and neither do they have a similar impact. I would always recommend that one focus on the most important selections the organization makes.

I have seen folks looking to pilot with smaller problems or less impactful selections, but the effort needed to get a working solution for any problem is almost the same, as one is mostly fighting against process inertia.

4)?????Focus on incremental improvements to decision-making. Not every business problem needs machine learning models and AI from the get-go. A selection process can be made better even with specific business rules and visibility into past patterns etc. So, every selection process needs its own roadmap. Often additional data elements may make the selection better. One may eventually need to build machine learning models, but that is mostly after the easy pickings have been done.

5)?????Exploit Unstructured data: Most organizations today are not using 80% of the data in their decision-making because it’s in an unstructured form (text, images, videos, etc.). Exploiting this data would open new insights that the organization would benefit from.

6)?????Align with Operations: As an Analytics organization, the Operations or Business teams are the internal stakeholders. The candidates for Analytics intervention need to be championed by Operations for any chances of success. Working closely with the Ops teams would give great insights into how to make the business process better.

7)?????Focus on simplicity and scale: Today’s workforce is often saddled with legacy tools that are complex and manual. To be able to influence change, the Analytics tools need to be super simple to use. Ideally, Analytics should be pervasive in an organization but invisible to the business users. Embedding Analytics to business workflows/ application is probably the easiest way to get Analytics into action.

8)?????Closely partner with the data organization. CAO needs a close partnership with CDO or the data organization. It is prudent for the CAO org to define what is needed and leave the how to be defined by the CDO. All data consolidation/ technology modernization conversations could be relevant here, but only after the end game is defined.

This partnership is often the most important but the most fragile partnership that defines the success of an Analytics strategy.

With so much focus on Analytics across organizations, I would welcome feedback comments, and alternative ways of looking to build an Analytics strategy.

Shrikant Dash #analytics #datascience #datastrategy #analyticsstrategy