Strategic Analytics and AI

This blog is about strategic analytics and AI. So I thought I would start off by writing the first article about that.

What is strategic analytics?

The term strategic analytics has been around since at least 2008 although the work in this area goes back several decades. Strategic analytics refers to the application of mathematical and analytical techniques including statistical analysis to answer questions that are of strategic value to the business. This includes enabling decisions in areas such as revenue management, product development, competitive positioning, cost efficiency and profitability.

Levin and Mcgill suggest that strategic analytics encompasses analytics utilized to answering two types of questions – one-off analyses to answer big questions such as strategic alliances, partnerships and M&A or the automated analysis of numerous small decisions such as inventory control and pricing, promotions etc.

Strategic analytics refers to the application of mathematical and analytical techniques including statistical analysis to answer questions that are of strategic value to the business.

Strategic analytics would also include the analysis of data to support strategy based on frameworks such as industry analysis, five forces analysis, value chain analysis or resources and capabilities analysis. Another identifier of strategic analytics is the use of data to evaluate high level financial metrics for an organization such as an NPV analysis or free cash flow analysis. Other such analysis to evaluate investments based on metrics such as return on investment (ROI) or return on ad spend (ROAS) also constitute strategic analytics.

What are the tools and techniques for strategic analytics?

The most common tool for strategic analytics remains Microsoft Excel. However, tools including Python, R and SAS are also being increasingly used. A wide range of techniques are used in strategic analytics including statistical analysis, predictive modeling and machine learning, forecasting, optimization, simulation and scenario planning. Here is a brief description of each of these techniques:

Integrated machines that execute strategy hold the biggest promise.

Statistical analysis - an important part of strategic analytics remains some basic descriptive analytics and exploratory data analysis. Techniques including averaging, bivariate analysis or cross-tab, correlations are essential for understanding the characteristics of some of the variables involved.
Predictive modeling and machine learning – techniques including classification and regression which fall under this category are essential for strategic analytics. Problems such as churn or upsell prediction, delinquency in loans prediction, fraud detection, acquisition and cross-sell are cases for classification techniques. Problems including investment level prediction, ad spend prediction, inventory level prediction etc where the predicted quantity is continuous are cases for regression techniques. In most cases, such predictions are incorporated in and investment or cost analysis equation with trade-offs to answer a strategic question.
Forecasting – predictions involving time series measures generally fall under this category. For instance, predicting the data usage in a telecommunications company by month, week or day involves a time series. Other forecasting problems include forecasting calls to the call center, forecasting churn numbers or acquisition numbers or investment levels or profits on a monthly or some other periodic manner. Such forecasts could also be used along with optimization and scenario planning methods to answer a strategic question.
Simulation and scenario planning – Monte Carlo simulations are the most common type of simulations. This could be for generating scenarios in a financial model to explore what if situations. More sophisticated simulations may also be used to generate data and scenarios.
Optimization – this is generally used to maximize or minimize some objective given constraints on a few variables. Knapsack algorithm, integer and mixed integer programming are some of the techniques used for optimization.

What does AI have to do with strategic analytics?

Most organizations are grappling with formulating an AI strategy, that is coming up with a strategy to apply AI to serve some business objectives. They are trying to come up with an approach that evaluates the use of AI to deliver business value while transforming the way they do business. They are also trying to understand the impact of AI on their business strategy while assessing the risks due to the use of AI as well as the risks if they are slow to adopt AI.

While most organizations are in the stage described above, BCG has outlined a view of where AI is applied in some of the leading organizations and where it could be used increasingly in the future – namely in formulating and evolving business strategy. In what they call strategy machines, they outline how some organizations such as Amazon have employed AI to continuously learn from their data and execute on their strategy where there is a machine that learns, formulates and executes on the business strategy that is best for the organization. For instance, Amazon has integrated its fulfillment, order management and inventory control systems which function like a well-oiled machine by executing on the strategy of maximizing profitability.

I believe that it is these kinds of integrated approach to the application of AI to execute strategy that will hold the biggest promise in the future. However, most organizations are a little way off from this kind of integrated approach and must take small steps on their road to this ideal end state.

For more information on strategic analytics and strategy machines:

About this blog

This blog is about strategic analytics and AI. I will write about topics that are relevant to strategy but also about analytics, data science and AI. That said, I plan on writing articles on those topics as well as on data-driven and managerial decision making and digital marketing. Some future topics include causal inference, managerial decision making and data science, AI trends and data science for digital marketing.

Photographs by Unsplash.