Analytics Across the Spectrum: Defensive to Offensive Analytics

Data Analytics Overview

It is important to understand two major differences in terms used in the data analytics space. The first two terms are very commonly used, the second two are less commonly used in name, however, the themes are often used without naming them explicitly. 

The first two are Data vs. Data Analytics.  The second two are Defensive vs. Offensive Analytics, as outlined by Leandro DalleMule and Thomas H. Davenport in their seminal work, 'What's Your Data Strategy?'.

Data vs. Data Analytics

In many organisations, you will find a CDAO - a Chief Data and Analytics Officer. Data and Analytics are often grouped together, and in fact, are heavily reliant on each other. My personal, and not a common view is that these should be distinct roles. Indeed, I've been the Head of Data Analytics, and I believe I do a better role focusing on Analytics. There are better-suited personalities to each of these roles, specifically governance and data management specialists for the Data focused roles.

Defensive vs Offensive Analytics

In a bank, Defensive Analytics will focus on topics such as Collections, Capital Optimisation, improved credit processes and general data quality improvement and data management. A large part of Defensive analytics that can be transformed, with the right positioning, into an Offensive Analytics play, is Financial Crime. Financial Crime analytics together with Fraud Analytics is typically a defensive play for banks, however, this can be a key client offering. Forward-thinking banks will externalise this offering to clients, and promote it as a selling point.

There may be slight differences in Defensive Analytics between investment and retail banks, however, the core themes are similar:

1. Preventing loss
2. Preventing fraud and financial crime
3. Ensuring client and data privacy
4. Reducing cost and process optimisation
5. Working closely with Legal, Risk and Compliance to help them achieve their objectives and also oversee analytics projects
6. Ensure regulations around data and analytics are adhered to
7. Data quality and accessibility

An automated data quality system is the core of Defensive Analytics. Data Quality should be one of the most important metrics your CDO is measured on.

From Defensive to Offensive

Offensive Analytics is perhaps the more sexy side of analytics (and probably often misunderstood), as it is often client-facing and more aligned to data monetisation. Offensive analytics will also be quite different between an investment and a retail bank; however, the core themes will be similar. Offensive Analytics will focus on:

1. Gaining new clients
2. Retaining existing clients
3. Data monetisation (Direct and Indirect)
4. Improved decision making (often dashboards surfacing previously unseen data are used effectively here)
5. Harnessing data from platform plays to improve client experience and expand the bank.
6. Helping business and client-facing teams ensure data is being used to drive value for clients and the business.

Defensive vs Offensive Analytics can also be seen in terms of defending the core vs. defining the edge of the organisation.

Defensive Analytics is largely focused on defending the core of the organisation and ensuring the bank continues to improve on and refine its core offering to clients.

Offensive Analytics, meanwhile, defines the edge of what analytics can do for the organisation. It looks for the use cases that will wow clients and make people realise the true power of data and analytics. However, this is only possible due to the work done by Defensive Analtyics on improving data management and quality, often classifying data and ensuring data is ready for Offensive Analytics use cases.

Data vs. Data Analytics

The differences between Data Management vs. Data Analytics can be categorised in the tabs below.

Privacy Data Classification Model Building Model Deployment Privacy

Data Management: Focuses on design and implementation of privacy principles.

Data Analytics: Ensures that analytics use cases do not breach privacy guidelines and legislation.

Data Classification

Data Management: Ensure processes are in place to manage and monitor data quality, report on exceptions and make changes.

Data Analytics: Feedback on data quality issues picked up in the analytics process, assist the data management team to resolve issues where possible.

Model Building

Data Management: Ensure quality data is available for model building. Surface data via data mesh or data lake.

Data Analytics: Core focus of the data analytics team.

Model Deployment

Data Management: Processes for identification of data drift.

Data Analytics: Monitor and ensure mitigations are in place for model and data drift. Work with ModelOps engineers to deploy models.

Data Analytics in a Bank

In a bank, one wants to take big data, and in the words of Hemant Taneja, "unscale" the analytics so that each customer, no matter if that customer is an individual or a massive corporation, is obtaining highly personalised analytics to best help them with decision making.

Initially, this will likely be done via bespoke analytics and highly skilled data scientists working with those with business expertise to build solutions for clients. Over time, the most appealing solutions will need to be deployed in order for them to be scaled and sustainable, being created regularly in an automated fashion.

Other analytics should be offered as an on-demand service to business users. This is the true meaning of self-service analytics, where tools are deployed as a no-code front end environment for business users to request analytics on an ad-hoc basis. Once we achieve this, data scientists work to continuously improve the analytics offering and build new tools for the business.

Those in charge of data will also look at data architecture, master data management and data quality, It is clear that these areas are too large for one person to also focus on analytics and how all this data is being used. This is another reason why I propose that the CDO and CAO are separate roles to look at data management and data analytics respectively.

The golden thread between these two areas is data engineering and even the relatively new field of data analytics engineering. Simply put, these people ensure that data is available at the right time and at the right place for analytics to be performed. They will also ensure that when productionised, analytics can run smoothly.

Good luck in your analytics endeavours!

Matthew

Share this post