Digital Transformation/Modernisation, It is a Continuous Journey

Fahad Najeeb, Head of Data Platforms and Engineering, Latitude Financial Services

Digital Transformation is an exercise undertaken to uplift an organisation’s processes and Modernisation is about upgrading the systems & technology to meet current demand and requirements. (I will use the term Modernisation to address both in my article).

Most organisations today take an approach of modernising their processes and systems as a once off exercise i.e., Transition from State A to State B. Question is if this always work and is the best approach? Or perhaps should we ask does State A always need to be non-existent for State B to exist?

"Aim for small changes – key all your data on a singular ID transactionally, operationally, and analytically"

Modernisation focus can easily be at decommissioning of current state which depending on the complexity of the processes and systems is usually expensive, extensive and time consuming. How about if the focus is increased on co-existence (or transitive but for longer term) for current State and future State while achieving the modernisation targets.

There are many factors that make modernisation complex:

1. Time and Cost of discovery to analyse current state being high.

2. Commercial licenses of current state locked in for years.

3. Dependencies of security, infra, network between old state and new state.

4. Value proposition usually after complete transformation.

One approach is to go through each step one by one while another approach is not to disrupt the current state and stand-up a parallel state. This parallel state may exist in digital platforms, digital channels, Data ecosystem, middleware, and applications.

How would one pay for a parallel state, immediate question comes to mind?

New state needs to be built on a high value use case or a critical go to market initiative that will help adoption and drive usage across the organisation. Think about an acquisition your organisation might have done, this could be the best time to leverage this approach and build modern systems. Adoption, usage and modern systems gravity takes on eventually and helps for complete transition from base line to target.

Coexisting states bring disparity as well, you will need to have processes and patterns that are forward and backward compatible. For e.g., an organisation might stand-up a modern data platform alongside its legacy warehouse, modern ecosystem will drive adoption of multiple high value use cases and increase adoption.

You will need to have mechanisms (Process/Patterns) in place that make the parallel usage frictionless in legacy and modern ecosystem so consumers can marry up the data in from both worlds. We use AWS DMS heavily to continuously mirror data assets from old world to new.

Use a framework such as fitness functions (based off Building Evolutionary Architecture by Neil Ford, Patrick Kua and Rebecca Parsons) to design triggers based on criteria’s that will kick off either manual action taking, thinking or automated transition from one state to another. Result is your organisation won’t be static and solely dependent on complete human intervention to modernize/transform your landscape.

Some practical thoughts on various topics when we talk about modernisation below:

Technology:

Confidentiality, Privacy and Right to Forget – These topics are widely discussed and need appropriate handling, depending on what industry you operate in, always good to have some capability made to address these concerns for your structured, semi-structured and unstructured data. We recently implemented a state of the art AWS ML based Product by partnering with AWS that addresses the very topic.

Realtime – Always strive for Realtime data movement as opposed to batches where you can, push for current system on how they can move towards being real time, the ability to do a decision, the ability to reach out to a customer in real time, the ability to handle fraud in real time; all of this is more beneficial than to be able to do same activities in days.

Less File Movement – Avoid, if you can any or all large batch file movements, this with due time increases data inconsistency, silos, data quality issues and majority of the time are very point to point and non reusable.

Services and Platforms – Build reusable services that are enterprise grade and can accommodate more than one use case or initiative. You will have a better value proposition when you need funding and the service will pay for itself sooner.

Centricity of Customer – Build and strive for any shape and form of a customer 360 view, if you can’t have a complex state of the art customer centric view. Aim for small changes – key all your data on a singular ID transactionally, operationally, and analytically. Make sure to propagate these changes in all future systems while slowly chipping away legacy challenges.

Enrichment – Make reusable store of Customer Profiles which are enriched in real time fashion with fast moving facts about customers, these facts can be coming from any upstream system or could be based on ML based insights, You can then utilise these enriched profiles to take further informed decisions on customers and their segments. Think about CDPs.