Towards data-driven productivity
Image by Antti Pikkusaari

Towards data-driven productivity

The era of analytics point-solutions is over. Welcome to the era of ubiquitous AI enabled by data products in a mesh.

To stay competitive in digital age is to embed data, analytics and AI in all aspects of value creation.

Use cases and data products need to be integrated with all digital capabilities. Settling for isolated analytics use cases won’t cut it. Instead, we need Scale, Speed, Agility, Quality and Innovation. And we need Managed Complexity.

Enterprise is to transform into a digitally enhanced and data-driven value creation machine that hums with internal consistency and coherence. Nothing less is sufficient.

This article kicks off an article series set to explore how to acquire, build, organize and develop business capabilities for the next phase of the digital age. Phase with unprecedented scale and impact of AI deployment. Phase that will set digital winners and laggards apart.

Exploration will revolve around themes like architecture, operating model, enterprise applications, engineering and platforms. Data Mesh inspired distribution is now applied to all digital capabilities – leading to new concept: Digital Distribution.

Digital innovation at productivity frontier

Success in digital age calls for continuous innovation. Leveraging digital technologies for productivity gains becomes business management prime objective.

But what exactly is productivity? In simplest terms, Productivity equals Value divided by Effort. Numerator translates to customer value and customer experience. On the denominator side, operational efficiency is the main factor. In short, productivity is about more with less.

So, digital innovation is about deploying digital technologies to increase customer value, improve customer experience and to enhance operational efficiency. On this macro level, things do not appear complicated.

Deployment of digital technologies moves productivity frontier in each and every industry. Productivity frontier is basically about competitive landscape. Leading innovators move the frontier themselves. Followers stay at or near the frontier. And finally, laggards fall behind. They fail to innovate and therefore they get marginalized.

However, digital innovation is hard. How to facilitate that thru good design and strong leadership is the overarching theme of this and later articles.

Digital assets and capabilities

Technology, productivity frontier and competitive landscape are all company external concepts. When the focus shifts to company internals, we start to talk about digital assets and capabilities.

The difference between assets and capabilities is significant. For example, data as an asset refers to data sets and sources while data capabilities would be about IT infrastructure, processes, tools, organization and human skills.

Overall, digital assets and capabilities can be categorized as follows:

  • data assets and capabilities
  • business applications of all types
  • IT infrastucture for computing and storage
  • software development and engineering
  • human skills and competences for all of the above

From business perspective, access to capability is what matters. Ownership is often of less importance. Cloud computing has made that very clear with multiple SaaS applications established as the norm. Correspondingly, software development outsourcing can be the most feasible option. However, data as an asset is different: Ownership actually matters. A lot.

Embedding analytics in all aspects of value creation

To maximize productivity is to embed data, analytics and AI in all aspects of value creation. That is, to integrate them with other digital business capabilities. No longer it is enough to settle for a couple of machine learning models to boost marketing operations. Let alone to treat analytics as a monopoly of BI-assisted decision making. To stay at the productivity frontier, we need to aim higher than that.

Think hundreds of concurrent analytics and AI use cases. McKinsey global survey The state of AI in 2022 provides good perspective and yardstick. And even that is far from end-goal.

To maximally boost customer value, customer experience and operational efficiency with analytics and AI means integrating them with products, services and processes. And to systematically build towards AI Organization.

Embedding analytics and AI into business processes implies integration in business applications, specifically ERP modules as part of the overall Composable ERP – as coined by Gartner. In this way, analytics and AI start to feed back into business applications to further boost corresponding business processes. This is significant addition to traditional way of using IT as business process aid.

For example, machine learning model deployed to optimise manufacturing process thru integration with MES module. As an isolated use case, this is not new. What is new is that such integration becomes the norm with use case volumes increasing fast. Furthermore, data will now be deployed as managed data products sourced from distributed data mesh.

Digital capabilities design criteria

To avoid point-solutions, digital assets and capabilities must be treated holistically as a system consisting of technology, processes and people. Before kicking off system design, it’s good to take a moment to think about design criteria. What is it that we require from such a system?

Scalability comes first. To achieve the overall goal of hundreds of concurrent use cases embedded in all aspects of value creation – integrated in all products, services and business processes – we want to avoid bottlenecks emerging as road blocks.

Speed is next. Time-to-market is often the deciding factor with regards to keeping customers happy and capturing market share. Hence, our multifaceted system needs to be fast in coming up new digital solutions.

Agility is somewhat elusive concept that can mean many things at the same time. At its core, it’s about flexibility and responsiveness with regards to fast-moving customer needs. Agility is about flourishing in complex and dynamic environment. System of digital capabilities without in-built agility is not competitive.

Quality, especially lack of it, is way too familiar to us working with digital capabilities and data assets. Achieving good quality is inefficient and costly as an add-on based on manual controls and intervention. Instead, good quality needs to be in-built system characteristic with high degree of automation.

Innovation is the key competitiveness ingredient in the digital age – needed for high-performance processes, winning customer solutions, differentiation, and ultimately business renewal. System of digital capabilities is in crucial role in catalysting and enabling innovation.

Finally, system of digital capabilities is prone to get complex. However, total elimination of complexity does not appear as feasible target. What is needed is Managed Complexity. A system that allows itself to be managed and developed while fulfilling the requirements listed above.

Luckily there’s a design principle that serves all our needs.

Distribution as design principle

Domain-Driven Design has its roots in software development where it was introduced to help to design, develop and maintain large complex systems. DDD has led to proliferation of microservices as superior alternative to monolithic application designs. As the the name implies, distribution to domains is at the heart of DDD. More recently, DDD principles have been applied to data management, leading to Data Mesh. See Data Mesh book review and beyond.

Decision to extend DDD principles to be applied to all digital capabilities is based on simple observation on underlying drivers: Need for scale, speed, agility, quality, innovation and managed complexity is still there. Nothing really changes when moving from application development to data management and further to digital business capabilities as a whole.

Digital Distribution is the last phase in the evolution that started from software development with DDD and later expanded to data management with Data Mesh. The context changes but key principles and underlying drivers stay the same.

Digital Distribution applies DDD and Data Mesh principles to digital capabilities as a whole

Data mesh introduces the principle of Business Domain as the owner of data products. Setting business domain to become the most important distribution building block in the overall digital capabilities system design is not only natural extension of data mesh but comes with significant benefits.

Business Domain becomes the most important building block in Digital Distribution

First, business domain as Bounded Context has very high utility. That enables shared semantic understanding by all parties involved. Things become meaningful, tangible and understandable. Language and used concepts and terms become common. This is sometimes called ubiquitous language or domain language. In this way, business domains become areas where knowledge, activities and creation come together. Indeed, it is these human cognition and communication aspects that must be set as primary system design paramaters rather than any technological factors. In this way, business domains become digital innovation and productivity engines – which is the point of the whole exercise.

Human cognition and communication aspects set as primary design paramaters – leading to Business Domains as digital innovation and productivity engines

Second, assigning ownership of digital assets and capabilities to business domains has profound merits. Product thinking with lifetime ownership is one. Business applications are optimised to serve domain specific processes and activities. In case of data as an asset, this leads to data products that bring not only value but also quality. Software development is no longer about temporary projects but about permanent applications, services and products – with continuous iterative improvement to better serve organizational and customer needs.

Third, cloud computing service model flexibility fits perfectly with federated business domain ownership. Prime examples would be numerous SaaS applications and Composable ERP with the ecosystem of business application modules that can be either off-the-shelf applications when standard is enough or customized when differentiation is needed. Customization may involve AI-boosted business processes for efficiency and/or differentiation. Furthermore, with regards to taking the ownership over data and software, cloud based platforms, tools and automation options provide a way to significantly reduce cognitive load and operational overheads in business domains. Things become manageable.

But how to identify and define business domains? Two rules of thumb apply:

First, business domains are chosen to follow natural organizational boundaries be they functional, geographical or both. Bounded context and shared semantic understanding are used as guidance. Because the main design goal is to secure efficient human interaction, collaboration and innovation, it is no surprise that traditional business functions like Marketing are a good first guess.

Second, in identifying and defining business domains, we need to live with certain fuzziness but we do not allow ambiguity. By allowing fuzziness we keep our minds and options open when striving for optimum configuration. By removing ambiguity we ensure organizational alignment and eliminate waste due to communication failures. Nick Tune describes the difference between fuzziness and ambiguity in his article on domains.

Reuse and network effect make data different

All digital capabilities benefit from distribution to achieve scale. However, with data as an asset the potential gains can be even higher. This is because of unlimited reuse combined with network effect.

Data mesh consists of data products connected to data sources, data consumers and to each other thru well defined input and output ports. Put together, they form a mesh of potentially tens or even hundreds of data products.

As discussed, embedding data in all aspects of value creation leads to hundreds of concurrent analytics and AI use cases. A real value explosion occurs when large number of data products are utilised by even larger number of analytics and AI use cases. This is the pinnacle of data-driven productivity.

Large number of data products utilised by even larger number of analytics and AI use cases is the pinnacle of data-driven productivity

But there’s a caveat. To capitalize on network effect, data products need to be reusable. That can only happen when strong dependencies between data products and analytics use cases are avoided. In other words, each data product is to be designed for extensive reuse rather than tailored for individual use case. Flexibility and loose coupling are the applied design principles here.

That then leads to multiple operational requirements, including data product management as organizational function, good understanding of market and data consumer needs, data product life cycle management, data contracts, and ideally co-creation with combined innovation process between data producers and consumers. From another perspective, all of this results from taking ownership over data assets and from the set objective to maximize data-driven value.

The imperative of data product reuse leads to multiple operational requirements

Merging of operational and analytical planes

Embedding analytics and AI in all aspects of value creation has yet another consequence. Telling analytical and operational data apart is becoming increasingly difficult.

When analytics was predominantly about enabling business intelligence, analytics was separated from actual business operations and definitely isolated from business critical processes and IT systems. Operational data and analytical data used to be different worlds or “planes”. Well, no more.

Analytics and AI playing increasingly important role in customer value, customer experience and operational efficiency implies merging of operational and analytical planes. In practise, that calls for integration of analytical data and machine learning models with business processes and applications, digital services and connected products.

Merging planes is about integrating analytics and AI with business processes, services and products

This does not mean that all data becomes analytical. Instead, a significant portion of all data remains transactional by nature with no need for analytics. However, this does mean that business operations, including business critical processes and applications, will have analytics and AI component with a fast increasing role.

All that constitutes significant technical and organizational challenge where optimium choices on architecture, IT infrastructure and operating model become critically important. One thing is clear: There’s no way that this could be implemented and maintained as centralized monolith – not from technical nor from organizational perspective. Distribution with business domains taking ownership of their digital assets and capabilities appears as obvious path forward.

Digital strategy dualism

Traditional role for strategy has been to provide answers on how to succeed in the business environment. For example, how to position the company in relation to markets and competitive landscape, or what to offer to customers and how to best serve their needs.

Digitalization does not change those earlier fundaments. Indeed, digital strategy is to outline how digital technologies are to be deployed for higher customer value, better customer experience and enhanced operational efficiency. Or in more concrete terms, how digital assets and capabilities are used for better competitive position to gain pricing power and achieve better margins leading to profitable growth. For example, how analytics and AI use cases are used in business processes and services for winning customer experience. Strategy is not about detailed plans but it does need to show the relationship between digital investments and business benefits.

However, that is no longer enough. Digital strategy has another role to serve with equal importance: Guiding the decisions and choices on how to acquire and develop digital assets and capabilities. That is, how to implement Digital Distribution. This role is closely associated with the need to manage complexity discussed earlier. Business domain ownership does not imply anarchy – on the contrary. Managing the system of digital assets and capabilities cannot be based only on operational daily routines within the organisation. Instead, as so much is at stake, decisions on digital assets and capabilities enter the domain of strategic management. That is, business management being hands-on with strategic decisions on digital assets and capabilities and how to utilise them for long-term business success.

Journey of discovery

With these thoughts and guiding principles, let’s now start a new journey of discovery: An article series exploring digital assets and capabilities and how to acquire, build, organize and develop them.

Upcoming themes include:

  • Architecting Digital Distribution (published article)
  • Digital Distribution operating model
  • analytics and AI boosted enterprise applications
  • software and data engineering for data-driven productivity
  • platforms of various types for data-driven productivity
  • Digital Distribution as strategic option
  • digital strategy
  • change management in digital capability build-up
  • and many more

The overall objective will be to maintain strategic business perspective while diving deep enough into individual digital capabilities. The overarching theme is how to realise the vision of analytics and AI embedded in all aspects of value creation – without scalability limitations and while building for speed, agility, quality and innovation.

Kingsley Uyi Idehen

Founder & CEO at OpenLink Software | Driving GenAI-Based Smart Agents | Harmonizing Disparate Data Spaces (Databases, Knowledge Bases/Graphs, and File System Documents)

1 年

Great post!

Ionut Gabriel Stanca

Senior Sales Manager at Antenna Entertainment CEE

1 年

This an excellent article and many Chief Data Officers should have it in their inbox. Two other things are needed to implement what you described: 1. The right tool. 2. Human Talent. 1. The right tool. Many companies live under the wrong impression that they are data-driven only because they bought best-in-class software for each department. But data sits siloed inside those departments, and they don't have a complete picture. 2. Getting people with the right skillset became very difficult and expensive. Companies are also looking at reducing their operational expenses, so many have a hiring freeze. I'm looking forward to reading your next article.

Raj Grover

Strategic Visionary: Architecting the Data-Driven Digital Transformation Roadmap for Value and People Centric Excellence

1 年

The title itself speaks for the importance of data, analytics and AI in the productivity and value creation. A highly comprehensive note, Antti Pikkusaari. Thanks for sharing, Sir. Earlier this month, leadership at a corporate house was discussing his #digitaltransformation framework with me and every essential element was mentioned in the framework, but data and analytics were missing. When I pointed it, he said, "Data is there in every element of this framework." I said until you won't mention it explicitly, everyone will NOT understand its importance in decision making and data will remain useless and your digital transformation strategies and roadmap will not achieve any success without strongly pushing it across the organisation. Without embedding data, analytics, AI into every aspect of the organisation whether production, value or innovation, you can't see the desired outcome and success. I look forward to your next note in the series. Wish you all the best!

要查看或添加评论,请登录

社区洞察

其他会员也浏览了