Digital transformation demands a new automation framework

Chris Gardner of Forrester outlines steps on laying the groundwork for the next catalytic revolution

Comments

Before automation can do its job, data needs to enter the system

Today, half of all companies are engaged in digital transformation efforts and laying the groundwork for the next catalytic revolution — automation.

Automation involves a broad set of tools, from artificial intelligence (AI) to robotic process automation (RPA) to orchestrating software-defined infrastructure to physical robots and beyond.

Enterprise architects and infrastructure and operations leaders will play a key role in implementing automated systems that control the operation of tasks, processes, and systems — often replacing or complementing human labour.

All automation follows a common sequence of events: gathering stimulus, evaluating it, sequencing a response, and then executing it. You can compare different types of automation across nine dimensions in three groups: process dimensions, enterprise dimensions and people dimensions. When you compare technologies using these three groups of dimensions, you end up with eight main categories of automation: 1) engagement; 2) sales; 3) marketing; 4) decision; 5) process; 6) industrial; 7) infrastructure; and 8) development. The combined dimensions and categories encompass the automation spectrum.

The process dimensions focus on how automation kicks off and runs

Before automation can do its job, data needs to enter the system. In the past, this happened through highly defined mechanisms that only accepted (and worked in) code. More recently, unstructured data acquisition and processing is valid. In the process dimensions:

Data acquisition describes the acceptable input. On one end of this scale are tools that accept highly structured data. On the other end is automation that accepts highly unstructured data, such as intelligent assistants (e.g. chatbots) that must accept a large variety of variable, human-created input.
Comprehension outlines the rules by which automation executes. In coded comprehension, humans set the rules and automation follows them. Deviation is not expected, and all processes within the system are human-readable.
Determinism is how you get from point A to B. Once data is in the system, you must process it. A highly deterministic system follows a specific, predefined workflow: Do A, then B, then C. The opposite is a nondeterministic system, where workflows aren’t predefined and the automation itself has flexibility in handling business rules.

The enterprise dimensions outline the impact to your company

Once you decide to implement a particular type of automation, you need to vet its impact on your company and its employees. This requires a thorough examination of the skill of your workers, a justified business case, and a governance plan moving forward. In the enterprise dimensions:

Robotics quotient (RQ) indicates your automation readiness. Some automation requires a thorough understanding of the internal gears to use it, while just about anyone can utilise others.
Business case viability verifies that it’s worth the effort. Every implementation requires a justified business case, and automation is no different. Automation with low viability hasn’t yet been proven and has no immediately impactful business cases.
Governance/auditability indicates how well you can manage your bots. Once you implement automation, you need to manage and — in some cases — flag it for compliance. On one end of this dimension, humans have complete control and are able to control all decisions and audit as necessary.
Chris Gardner of Forrester: 'Once you implement automation, you need to manage and — in some cases — flag it for compliance'

The people dimensions make vividly clear how automation will affect humans

As you automate, you change humans’ experiences. We can judge this in three ways: effects on human-machine interaction, broader social effects, and the impact of work as a whole.

To be clear, these are fluid constructs that change with time. As technologies and processes mature, what doesn’t affect people today could very much affect them tomorrow. In the people dimensions:

Human-machine interaction outlines whom (or what) the bots are talking to. Not all automation has a direct connection with humans — some work directly with other types of automation through backend channels.
The human effect goes into the overarching social impact. Automation, in general, has a wide social impact — it affects just about everyone, from the products we buy to the services we use. However, particular types of automation have a broader effect than others.
The future of work effect describes the labour and financial impact. As you implement more automation, you redefine work itself as well as how to pay for the work. Automation that has more of an individual work effect remains largely manual. Tasks and labour stay relatively static..

Categories of automation show connectivity and disjunction patterns

All types of automation can fit into eight major categories. When we map these categories across the three groups of dimensions, patterns emerge. Patterns across a category sometimes exhibit a wide spread. A wide range indicates that different automation engines may be necessary, even among similar use cases. Conversely, a narrow band suggests that technology reuse may be appropriate. Here’s how the eight categories and the three groups of dimensions intersect:

Engagement automation heavily affects people. These technologies engage with employees and customers, either directly or through agents (human or virtual). This includes chatbots, customer service robots, and workforce automation. On average, these technologies take a mix of inputs and handle them in a more deterministic way. They have moderate business case viability and are transparent to audit. The effect on people is profound because many of the technologies are human-facing and have a direct impact on workers.
Sales automation technologies are tightly coupled. Once marketing reaches the customer, sales automation helps seal the deal. Technologies in this space include sales readiness, engagement platforms, and partner relationship management (PRM). In contrast to marketing automation, the technologies in this space bare tightly coupled around clearly defined averages. The business case viability is high with this group, and most of the technologies have a good degree of human-machine interaction.
Marketing automation has the widest variance across its technologies. Marketing automation includes any technologies used to find and reach customers. These technologies vary significantly but take in unstructured data and process it with a coded approach. They have reasonably high (in some cases very high) business case viability.
Decision automation runs the gamut on all dimensions. Decision automation focuses on making educated decisions based on insight. This includes technologies such as predictive analysis/machine learning, simulation, and streaming analytics.
Process automation takes a very structured approach. Process automation focuses on driving efficiency around business processes. On average, technologies in this space lean toward structured and coded processes. They can be somewhat opaque to govern and audit.
Industrial automation has great potential for workforce disruption. Industrial automation features technologies designed to create, inspect, and deliver products.
Infrastructure automation has a similar model to development automation. Infrastructure automation focuses on managing back-end computers and systems. Technologies in this category include configuration management, identity and access management, and security analytics platforms. Many of these technologies share characteristics with development automation, so they focus on code and determinism. They have a moderately high business case viability due to their focus on efficiency.

Development automation is evolving with AI. Development automation focuses on creating and maintaining software. It includes continuous delivery, different types of test automation, and software composition analysis. AI is evolving this space from determinism and coded comprehension toward nondeterministic and learning processes. It has moderate business case viability, and governing the technologies is transparent. The people effect varies based on the automation used.