Understanding the Analytics Maturity Model

Milind Desai
6 min readMar 5, 2022

--

Data Analytics is a buzzword of this century since the last decade. The big companies like Amazon, Alibaba, Coca-Cola, Facebook, Google, IBM, Microsoft, SAP have been early adopters of data analytics. These companies have realized great value by integrating data analytics across all functions from Strategy to Daily operations. Today several companies, big and small are leveraging data analytics and are showing a great interest in investigating how to advance their analytics adoption journey.

Gartner’s Analytics Maturity Model

In this article, I will be discussing Gartner’s Analytics Maturity Model. This model consists of the following four phases:

  • Descriptive Analytics
  • Diagnostic Analytics
  • Predictive Analytics
  • Prescriptive Analytics

These four phases represent different maturity levels at which organizations can be during their adoption journey in analytics. The ‘Descriptive Analytics’ looks backward at what has happened in the past (hindsight), whereas the ‘Diagnostic Analytics’ focuses on gathering insights from the past data and using it for decision making. The ‘Predictive Analytics’ and ‘Prescriptive Analytics’ predict the future and focus on ‘foresight’. The human input in decision-making reduces as you progress from Descriptive Analytics to Prescriptive Analytics. The decisions are automated in predictive/prescriptive analytics. The realized value of the analytics is much higher as you proceed into the advanced phases of the analytics. Let us discover these four phases in detail.

Figure 2: Gartner’s Analytics Maturity Model — Author

Descriptive Analytics

This is the first phase and addresses the question “ What happened” in the past using historical data. It addresses the questions such as how much, what, when, and where looking at the past data. For example,

“How many iPhones did Apple Stores in the city of New York sell last month?”

“How many customers downloaded the newly launched health app from the play store yesterday?”

“How many customers visited the store in the last 3 days of the SALE?”

“What are the regions which have less than 20% coverage of COVID-19 vaccination program?”

Photo by Adeolu Eletu on Unsplash

Case Study Example:

Descriptive Analytics helps a retail industry to discover information about the top-selling brands/products, top searched products, areas which have a high demand for your products, etc. This analysis will help the retailers to decide which products need to have higher inventory, which stores need to place frequent orders on fast-moving items, manage additional manpower, logistic partners and additional infrastructure needed to handle the rush of customers during the SALE period. These actions are mainly driven by the human interpretation of data.

Diagnostic Analytics

This phase is the next advancement in the adoption of Analytics and addresses the question “Why did it happen”. In this phase, you go deeper into the understanding of data by using statistical techniques such as statistical summary, the correlation between the numerical variables, bi-variate and multi-variate analysis to discover patterns in the data and discover trends in the data. The Diagnostic Analysis focuses on deriving the hidden insights from the data which are not captured during the Descriptive Analytics phase.

You can find answers to the following questions:

“Is there an impact of gender on the sale of the newly launched iPhone 12?”

“Does plant operations costs differ across the two locations significantly?”

“Is there a positive relationship between the age and the sales revenue from the protein bars?”

“Which of the following factors are responsible for the increase in the number of COVID-19 fatalities in the state — Age, comorbidity, the number of vaccination doses administered, gender, and lifestyle?”

You can also perform various hypothesis tests to prove/disprove your assumptions using statistical tests such as t-tests, ANOVA, Chi-square tests.

Case Study Example

Photo by Arlington Research on Unsplash

A popular bank claims that the average waiting time for customers is not more than 3.15 min. You can collect the data and analyze the waiting time using hypothesis tests and prove or disprove this claim. Analysis of this data will help the bank to evaluate the underlying factors which are influencing the waiting time for the customers and take appropriate measures to improve their operational efficiency.

Predictive Analytics

This is an advanced phase of analytics that focuses on data mining and uses various advanced analytics techniques such as machine learning, and artificial intelligence to build models that make predictions by using historical data. This phase addresses the question “What will happen in the future”. Predictive analytics studies the hidden pattern in the data and establishes the relationship between the outcome variable — in the case of supervised machine learning — and the predictor variables to build a mathematical/statistical model which can be used to predict the outcome variable for new data. This phase involves various stages of data preparation, data cleaning, feature engineering, model building, and model evaluation. Predictive Analytics helps automates the decision making process using these models that forsee the future and equip you with the right information to take the right actions.

In this phase, you will be able to find answers to the following type of questions:

“How many iPhone 12 smartpones will I be able to sell in the next month?”

“Will John purchase the new car in next quarter and do I need to follow-up with him and add him to the list of prospective customers for next month?”

“What is the probability of being hospitalized for a 62 year old male weighing 69 kg who is having a fatty lever and is staying in Texas, North America, if he is infected with COVID-19 virus ?”

“Is the color of the suit preferred by middle-aged men (35–50 years) in London dark-grey?”

Case Study Example

Photo by Maxim Ilyahov on Unsplash

For example, you can build a predictive model to predict if the customer is likely to purchase the annual subscription Pre-paid plan from the telecom service provider by understanding the pattern of his usage, frequency of recharge, recharge amount, amount of data consumed, his/her profile, etc. Telecom companies can then do a promotional campaign and target, such customers, where the propensity of purchasing the offer is significant and the ROI on the marketing investment is maximized.

Prescriptive Analytics

The next advanced level in Analytics is Prescriptive Analytics which addresses the question “How can I achieve it? — What should be done ?”.

Prescriptive analytics enables you to choose the optimal analytics solution to solve the business problem by enabling you to evaluate various mathematical models, compare their performance on the available data for a specific industry, and pick the most optimal solution.

In this phase of analytics, you can get answers to the following types of questions:

“How much ground staff will I need to operate the volume of passengers on 25-Dec in Mumbai?”

“Which are the top 3 combo offers that will boost the staple sales in next month?”

“In which Cafe-Coffee Day outlets across India should I open the mini book stores to do a cross selling?”

Case Study example:

Photo by NordWood Themes on Unsplash

Imagine that you are searching for a video on a specific topic such as “How to play Chess” on YouTube. YouTube’s search engine uses an algorithm that sorts thousands of videos to find the most relevant and useful results to your search query on the basis of relevance, engagement, and quality. Further, the YouTube search engine improves the relevance of the recommended videos by considering your search and watch history. Hence your search results might differ from another user’s search results for the same query.

Summary

In this article, we have seen the four phases of the Gartner’s Analytics Model and discovered the type of questions each of these phases address.

If you like this article, please feel free to share it on Twitter, LinkedIN and Facebook.

P.S. Enjoy reading another interesting article Data Analyst Vs Data Scientist.

Follow me on LinkedIN

--

--

Milind Desai

A blogger in Data Science, Artificial Intelligence, and Business Analytics