Digital Transformation – Processes, Data Modernisation to AI

Digital Transformation – Processes, Data Modernisation to AI


Digital Transformation & Challenges

Much has been written already on digital transformation adopting modern technology to transform existing processes to drive the productivity and reduce costs, improve customer experience, and achieve greater business agility which may lead to developing new products and services, contributing to revenue streams, and increasing profitability.

The digital transformation journey is unique to each organization based on the overall business strategy, existing technology capabilities and industry landscape. However, the common challenges many traditional organizations face on this journey is the presence of legacy technology, several siloed processes owned by different teams within a value chain, and localized data stores tied to multiple applications supporting those processes leading to various disparate platforms. These are some of the most complex issues to address to meet digital transformation goals.

This article is only focussing on automating the business processes and modernising the reliant data infrastructure as a lever to benefit from ML/AI (or other modern) technology on wider data sets to drive new data insights.

Data Modernisation & Risks   

One of the key digital transformation outcomes (Data Modernisation) is to enable consistent and reliable access to multiple data sources to business, operations & technology teams to draw insights from data and capitalizing on its full value. Data Modernisation here refers to standardising data sourcing methods, tooling, storage, mastering, data modelling, metadata, transformation, dissemination, self-service, access management and control policies. It enables development of intelligent data solutions by bringing together data from customers, operations, employees & businesses. Developing such capabilities across organisation could easily be a huge programme of work, not to mention migrating off the legacy technology, process redesign and automation, employee retraining efforts, and meeting current business priorities. Also, worth mentioning the complex and time consuming (multi-year) migration efforts, within and across multiple business units, that are required to get to the desired state. It would be difficult to accurately calculate investment requirements, plan deliverables and the efforts estimations – resources are limited. There is a constant ROI question to answer too. Then there is a risk of adding to existing legacy technology and processes if not all goes as planned. With so many unknowns and risks, how is it possible to achieve successful data standardisation – and this is only one of the steps in the digital transformation journey? Where do we begin?

Approach: One step at a time, Automate & Iterate                  

Quoting from a podcast by McKinsey – “I don’t think you’re ever done with digital transformation. It’s a muscle that you’re constantly building and you’re constantly honing to get better at”.

Traditional data infrastructures only aligned to specific business processes are difficult to maintain requiring significant effort to upgrade. However, such isolated deployments could prove to be good first use case(s) to automate and experiment. Assuming the data modernisation technology strategy and target architecture designs exist, it may be easier to tackle a couple of such use cases first. Since the traditional data infrastructure is tied to the business processes, it would also be useful to reassess the business processes itself at the same time, considering the overall transformation strategy. This approach helps build the necessary data modernisation plumbing infrastructure and standards to scale for other use cases to benefit from. Robotic Process Automation (RPA) and Business Process Management (BPM) technologies along with modern data infrastructure provide a natural progression to add ML/AI capabilities in the future. In the short term, RPA/BPM could automate the repetitive manual processes while modernising data infrastructure. In the medium and longer-term, with mature data modernisation capabilities already in place, leverage ML/AI technology to draw deeper insights across multiple data sets. This approach drives efficiencies to businesses in the short term, deliver to existing business priorities, meets ROI expectations, supports experimentation, and incrementally develops the necessary building blocks in alignment with the overall transformation strategy.

Planning the execution (Build-Measure-Learn)

Here are some of the suggested steps to follow.

  1. Make a list of opportunities

Develop a list of potential use cases that would benefit from data modernisation, business process redesign, or both. For each of the use cases, highlight the pain points, and the expected outcomes or end state.

  1. Identify & Analyse the low-hanging fruits!

It depends what teams identify as low-hanging fruit for their organisation. It would be advantageous to consider the perspectives of customers, businesses, technology, and operations. Typically, these would be opportunities with fewer steps in the business process end-to-end, infrequent data movement, simpler existing technology infrastructure footprint, well documented systems and processes, fewer dependencies overall with other systems, operational challenges – obsolescence, support & maintenance, and recent service incidents considerations if any.

For further analysis, consider end-to-end business process within a value chain. It may not be helpful to fix a part of the process or fix a redundant process. Prioritize the use cases that are aligned to planned/approved investments and business cases.

  1. Define expected outcomes, requirements & roadmaps

For the prioritised use cases, clearly define what the expected outcomes (business goals) are, and which key-results will be used to measure the outcomes (OKRs). Develop business process specific requirements and a roadmap as a high-level plan to help communicate with stakeholders.

  1. Develop platform & solution architecture

Design and document the platform architecture meeting enterprise architecture policies and overall technology strategy. Within the context of this article, moving from business process automation and data modernisation to becoming data-informed infrastructure implementing modern technology in future, the most important considerations I could think of are as follows:

  1. Remove silo data management for all the use cases and make it easy (and cheaper) to embed new data sets in future.
  2. Efficient processes with automation in place removing repetitive tasks, but also have an ability to re-design the processes when required to support the future business requirements and implementing advanced technology.
  3. Ability to efficiently onboard modern technology as new business opportunities develop
  4. Support experimentation
  5. Consider third-party technology selection criteria (features, integration, scalability, and future roadmap etc)
  6. All components of the platform can scale-up efficiently
  7. Provide the glide path (and necessary tooling) in migrating data and users off the legacy infrastructure

Develop solution architecture leveraging platform technology components to deliver the prioritised use cases.

  1. ROI modelling and business case

ROI modelling is one of the most crucial elements of the planning. It isn’t only about calculating financial benefits. It guides technology selection process by assessing and comparing the costs including comparing CAPEX and OPEX approaches and licensing methods, helps in technology procurement as per organization accounting policies, calculate ESG impact, and at time can guide technology migrations planning and feeding into overall business case development. Reuse the ROI model for future use cases.

  1. Technology development & governance

Develop the platform components and the specific solution meeting technology architecture and business requirements of the selected use case. This could be an MVP with limited scope addressing specific pain points.

  1. Measure & Improvise

Measure the solution’s effectiveness against the OKRs defined previously. Also, seek feedback from the users and improvise as necessary – develop new metrics as required.

  1. Migrate & Iterate  

To be able to use the new system, the end users are to be migrated to new infrastructure. This may also involve migrating existing data, shuffling connections to dependent systems and perhaps shutting down and decommissioning the legacy technology.

Finally, continue the validation, prioritisation, and development of the remaining use cases to build out the modern data infrastructure and automated processes. It would be useful to plan which of the existing use cases may benefit from ML/AI technology and what new use cases should be developed. Engaging customers to seek their inputs to build new products and services will certainly help.


Improved business processes and the modern data infrastructure are key to unlocking deeper data insights. ML/AI solutions only improves a part of the overall process and therefore improving the entire business process and the underlying data infrastructure is necessary to realize the full benefit. Modernising end-to-end business processes, data infrastructure and underlying technology means improved ability to respond to new opportunities by developing workflows quicker and as well as making timely upgrades to existing workflows. Scaling up the infrastructure using the automation first approach offers opportunity to learn and experiment while supporting immediate business priorities and preparing for the future.

Have Your Say: