I’ve done five or six large-scale digital transformations at global enterprises and I would like to share an important lesson I have learned:
Any transformation must be organisation-wide and penetrate deep into the core systems of your business!
Many enterprises transform the periphery of their organisation, but neglect to continue their efforts all the way through to the data-centric core of their business.
This tends to result in unintended consequences that may eventually make your situation worse than when you started: delays, bottlenecks, dependencies, miscommunication and misaligned incentives.
In this blog, I’ll explain why, when it comes to transformation, overlooking the data core of your business is an error, the business value that comes from including it and how you transform your data successfully.
The number one goal of transformation for most enterprises is to make their delivery faster and more agile. Accordingly, they focus their efforts on their systems of engagement: customer-facing products and services.
But they are often too focused on optimising frictions on the periphery so that they can get from concept to customer more quickly, rather than unlocking new sources of business value at a deeper layer.
Now, you can get marginal gains by optimising your technology, how your teams are organised, changing responsibility models and so on, but there is a ceiling to what you can achieve.
The reason is that these now-transformed peripheral systems of engagement (the ‘new world’) are inextricably intertwined with a massive legacy estate of systems of record (the ‘old’ world), such as accounts, ledgers, order books and so on, which tend to be very strongly data focused.
These systems of record are still operating in very traditional, centralised, monolithic ways. And I have noticed that enterprises tend to want to modernise the systems of engagement, but they are reluctant to modernise the legacy systems of record, because it’s really hard!
You do get some benefits from modernising the systems of engagement because you can deliver customer-facing work faster and make it more impactful. But you’re still fundamentally hamstrung by your dependence on the old world!
The result is a bi-modal technological/organisational topology that has benefits but also massive drawbacks:
Your teams in the new world work in, say, two week cycles, but they then have to coordinate with teams in the old world, who are working on three month cycles! The new team ends up being greatly limited by the speed of the old team. And it requires a huge amount of time and effort to manage these dependencies.
For example, say you create a new team that wants to quickly update a feature, which will take them two or three days. But then they find out that because this requires a change in the way data is provided in the systems of record they need to wait two months for that to happen. That’s a massive opportunity cost (as well as bad for motivation and momentum).
When the people in the old world start to see the freedom and autonomy that those working in new ways enjoy, it creates resentment and friction. Yet, because they are interdependent it’s critical that these two functions collaborate! But instead you have both sides blaming each other, which creates heightened tension between them. The result is delays, bottlenecks and general inefficiency.
When you have one part of your business operating in one way and another part operating in a different way, you need two governance frameworks. It’s difficult to merge these aspects into a single operating model and the overhead of managing that is significant, both in terms of financial cost and people-hours.
This also limits your options when it comes to doing advanced data science work. When the data-centric core of the business has not been modernised there is no way you can create broad access to high-quality, trustworthy, highly-discoverable data. This means that (effective!) artificial intelligence and machine learning are not possible.
Lots of enterprises try to just make do with this bimodal setup, but it represents a massive limit on the potential business value that could come from your transformation.
What is the alternative and what’s the benefit?
The alternative is to bring your transformation efforts down a level to those core systems of record.
Because these are so data-heavy this is essentially extending digital transformation into the realm of data in your organisation, modernising your entire approach to how data flows through your business.
It’s a question of data transformation.
If you can expand the newly-transformed capabilities and operating model in the systems of engagement into the systems of record then something cool happens. Your data sources and apps start to operate in a more agile, distributed, decentralised way. Eventually, you end up with a scalable ‘mesh’ of data providers and consumers operating in similar ways and at a similar cadence.
(For a deeper dive into the core principles and benefits of the ‘data mesh’ approach you can check out our introductory article here).
So instead of the teams in the systems of record working in three month cycles, they might come down to one month cycles, which is a lot closer to the cadence of their internal customers in the systems of engagement.
What’s more, by transforming your data you make it much more widely available across your business, greatly increasing the opportunities for data-driven innovation.
There are also internal benefits. Many enterprises have important internal customers for their data that need to demonstrate compliance, do financial reporting, prepare for audits and so on. And if you don't transform how that data is managed and operated you will have massive problems with data ownership quality, lineage and access that will require a lot of time and effort to unpick.
If your data is all tangled then trying to use it, either for innovation or internal use cases, is like trying to paint your hallway through the letterbox!
But if you can untangle it, get it catalogued and make it discoverable that opens up a wealth of information for all kinds of innovative external and internal use cases!
You can paint your hallway from inside the house :)
The reason people tend to avoid dealing with the data side of transformation is because, frankly, it’s pretty hard.
But here are five ideas that will help you to execute your data transformation successfully.
Any data transformation should be led by business objectives, by what is needed.
For example, I worked at a financial services company that did this successfully. They built a product strategy that aligned with some clear business objectives. Now, some of these new products were dependent on core data systems in the back end. So they looked at which fundamental sources the new products were dependent on and would pick off the most in-demand domains. These were transformed to be more accessible, trustworthy and in line with required service levels.
So, say they recognised that out of 10 products, six needed access to market data, they would prioritise that domain and introduce new ways of working and operating.
You can achieve this from the outset by creating read views of the data that’s mastered in each data domain in the existing systems. This allows you to extract the value of data without having to change the mastering systems first, which can be a long and onerous process. The new read views (i.e. data products) are owned by the teams that master them and operated in similar ways to the teams operating the systems of engagement.
This approach returns business value much faster than “eating the whole elephant” as a single initiative and gives you time to transform how you master the data in those systems in a piecemeal fashion.
As alluded to in the previous paragraph: the golden rule is to work incrementally. You take the highest-priority domain and transform that. Then you take your lessons learned and try to do a better job with the next-highest-priority domain.
In this way, you start to build these pillars of fundamental data sources within the business, starting with the most critical and moving outwards.
Each time you are effectively decomposing the old tangled data mess into new, separate, autonomous domains and slowly expanding this throughout the business at a sustainable cadence.
Technology is the easy bit. Harder is the processes: you need to define a new operating model that suits how data flows through your business.
In the legacy IT world, these processes are managed by central IT (perhaps even at group level). In the new world, however, you need to keep data ownership within individual domains so that the teams that produce the data are also responsible for mastering it and making it available as a product/service to other domains in the business.
When you change your operating model you create friction among your people. Their job changes, the skills they need changes and they need to take on new responsibilities (e.g. build and run where previously they only had to do build).
The key is to create internal consulting functions to help people manage the transition on a personal level. These both build capability while enabling teams to participate fully.
For example, a ‘core data’ function could go into different parts of the business (e.g. pricing or payments) and help to establish new ways of working so those teams are confident and comfortable.
You are effectively breaking down traditional group IT functions and moving to a federated model where IT capability is distributed across the business domains where the actual data resides.
It’s impossible to completely avoid some kind of bimodal setup in your business. It will always be a transitive state regardless of what you do.
The trick is that you want to limit the amount of time that you spend in this state, because the longer it goes on the more detrimental the associated problems will be.
In my view, the bi-modal way of operating is so problematic that if you’re going to transform one part of your business you have to go and transform all the other parts so you can unify your operating model into a coherent, federated business.
Doing it properly is worth it: transforming your systems of record creates a stream of accessible fundamental data sources that unlock vast sources of business value and competitive advantage!
If you want to be competitive, you need to sort your data constraints, and that's where Mesh-AI can help. Identify the areas in your organisation that require the most attention and solve your most crucial data bottlenecks. Download our Data Maturity Assessment & Strategy Accelerator eBook.
Interested in seeing our latest blogs as soon as they get released? Sign up for our newsletter using the form below, and also follow us on LinkedIn.
