Organisational decision-making using DMN

Introduction

“How did you arrive at that decision?”

Decision models describe how decisions are made in a given situation. They can be seen as formal and documented heuristics. 

This paper examines why defining a decision model as a separate entity from processes (actions) is great for companies (and for you!).

• We start with explaining the advantages of using Organisation level Decision Models (ODM). 
• “Getting things done” is governed by a process. We examine the Decision-process relationship, later explaining how processes will improve when decision models are fully leveraged.
• We explain a typical decision hierarchy and how ODM fits into that – with examples. 
• Decision simulations are  fundamental, but rarely leveraged to answer those “What if” scenarios. We look at the advantages of decision simulations.
• Finally, how will organisations evolve? We examine both distributed and auto-pilot models. ODMs form the foundations of these approaches.

So let’s begin by looking at what’s great about Organisation Decision models.

Why use Organisation level decision models 

Established Organisations are based on decision models. It is part of their “success formula”. As the organisation grows the number of models grows. However these models are rarely explicitly defined. As  in “This is our organisational decision model”.

Instead they are defined “indirectly”. Usually through process models in larger organisations. Or in most cases via KPIs (“Sales targets”,”Share value”), training or culture (“Principles”, “Codes of Conduct”). In summary : Decision models mostly exist within employee minds.

Clearly defined organisation decision models have numerous advantages. Some of these will be expanded upon in later sections. These are :

1. Resilience – Moving away from being held beholden to single points of failure.
2. Transparency – A better understanding on how things are done
3. Simulation friendliness –  Being able to easily test the model under different scenarios. “What if”
4. Future readiness – Moving towards autonomous and autopilot organisation. 
5. Crisis composure – Making decisions in stressful and chaotic scenarios becomes far easier and safer.
6. Generally improved execution quality and performance

In summary, ODMs allow quicker, relevant and safer decisions. Open standard Decision models enable common understanding of those models.

Decision Model and Notation (DMN) is an example of an Open Standard decision model published by the Object Management Group . See here for more info. It will be used as the basis of this blog. Next, we will examine how a decision model fits within the decision-process relationship.

Decision-Process Relationship

Organisations can be viewed through the lens of decisions and processes. Decisions are made which set processes in motion. This “decision-process” relationship is recursive, with one decision leading to another. For example, the decision to “improve customer experience” will spawn many process conceived decisions including:

• How should we change the way we work?
• What new capabilities are required? Can we use existing capabilities?
• How do we update our governance process?

This blog’s focus is organisational decisions. Actions, guided by process,  “get things done”. See the “Taking Better Action” section which explains the  Decision model Action relationship. 

We next examine typical organisational decision structures across roles.

The Organisational Decision hierarchy 

A decision hierarchy can be described as levels of decision abstraction. The higher up the hierarchy the more abstract the decision. Theory gradually transforms into practice as decisions traverse down the hierarchy. For example, 

Top level theory : Marketing automation should increase margins by 10%. 

Lower level Practise : Requirements analysis, Vendor selection, Project lifecycles, increased hiring, Go-live, System maintenance.…etc

Decisions within a hierarchy can be categorised at three levels: strategic, tactical and operation. This reflects “Why” a path has been chosen. “What” is the best course of action. “How” to turn the chosen course to reality.

 As organisations embrace digital, artificial intelligence and automated technologies the quantity of decisions required to be made grows exponentially at all levels.  The role of DMN is to improve the quality of these decisions.

Response latency describes the time taken to yield the results of a decision. How successful was it in achieving its goals? This feedback is fundamental in gauging the KPI impact of performing the decision. The table below summarises the response latencies within each of the decision types:

Response latency times will continue to fall as artificial intelligence and automation is more widely incorporated into the decision-making process. Dynamic organisations make high-quality decisions faster than their competitors, because their organisational structures are geared towards optimised decision-making, as well as embracing AI and automation. 

Using an example of “improve customer experience”. Why? Reduce churn, improve reputation and increase revenue. So within a typical decision process, the initiation (and funding) of such an initiative starts at the “strategic decisions” level. Tactical decisions facilitate the planning and organisation of the process. Operational decisions transform the vision into reality. At every level, decisions are made, and acted upon. The following sections examine each stage in the pyramid.

Strategic

These decisions steer the strategy, the direction, and ultimately the culture of the company. They are concerned with the whole environment in which the firm operates, the resources and the people who form the company and the interface between the two.

These decisions can be categorised as follows:

Although these decisions appear ad-hoc and unstructured they can still be modelled using DMN. For example, the strategic decision to adopt “people centred marketing” follows the model shown below:

The KPI which underpins this example would include customer satisfaction and churn. 

Mapping ad-hoc strategic decisions has several benefits:

• Provides a framework which other decision makers can use to perform similar decisions
• Transparency – Provides an audit trail of how decisions are made.
• Encourages the measurement of the decision’s effectiveness
• The model becomes part of the company’s Intellectual Property
• Enables generic “principle” models to be defined which handle multiple scenarios

Established companies usually have processes to validate strategies as part of their governance framework. These processes would have decision points (approve or decline, and why). These strategic decisions would then act as inputs into other strategic decisions, such as delivery and integration. For example, “is our delivery approach in line with the overall organisation strategy”? If yes, continue, if no, revise approach. In fact, strategic decisions will feed into every decision made at all levels.

Most decisions at this level are binary, i.e. approve or decline. When these choice outcomes are combined you get a clear picture of the organisation’s strategy and culture. Digital music is a binary format (data essentially consisting of trillions of ‘0’s and ‘1’s.) When combined and transformed you hear a continuous flow of music. So too with the binary nature of strategic decisions. They steer the organisational path.   

Although Decision Modelling Notation is geared towards automated and repeatable decisions, the methodology applies to all decisions. DMN asks the question “What are the general ingredients of this decision?” Even at a higher level there are types of decisions which are repeatedly performed.

• “Should we acquire this organisation?”
• “Should we invest in this software platform to become more competitive?”

After the decision is made additional rounds of governance are performed which require additional decisions to be made. A governance decision model is also required. 

The model used to arrive at a decision would differ from company to company. Decision models at all levels would be part of the organisation’s intellectual property and competitive advantage.

Tactical

Tactical decisions act within the constraints set by strategic decisions and are related to the medium-term (usually over a period of one week to a year) response rates. Tactical decision areas include marketing strategy, expense control, training planning, logistics management, etc. 

The scope to automate tactical decisions is higher than strategic decisions with numerous case management, human resource and workflow platforms performing these tasks and adding value. There are many ways to model tactical decisions with DMN, which can be implemented (i.e. auto-approvals for low-cost expenses or holiday requests during Christmas) or used as a “ways of working” guideline.

Operational 

Decision Model Notation operates mainly in the operational decision domain, where the decisions have a greater impact on customer experience. “Is this customer eligible for this product?”. “What is the likelihood of this customer accepting this product?”. These are the types of operational decisions which need to be made millions of times per second.

Operational decisions are also performed millions of times per day. The marginal revenue achieved by such decisions will have a big impact on the bottom-line.

An example of an operational decision model (above) examines the decision around what “next best action” the customer service agent should deploy for any given incoming call. Without assistance the agent’s own marketing knowledge and experience” would be used to determine the best course of action – cross-sell, up-sell or retention. The role of an AI-driven Next Best Action platform is to provide the best possible assistance to the agent. In addition, NBA would drive relevant and synchronised content in non-human assisted digital channels such as chatbots, mobile push-notification and e-mail.  

Governance

The DMN model is part of the overall governance structure. As such its development would be governed by the following factors (which would need to be agreed upon):

• Engagement of stakeholders
• Establishment of roles and responsibilities
• Conflict resolution
• Feedback and response handling

For example, all things being equal, should a particular customer be offered a cross-sell communication or a service communication? To resolve this potential conflict, the correct stakeholders need to be engaged and come to an agreement.

Note though that the higher the impact of a decision, the greater the need for oversight. 

Critical simulations

Simulations are fundamental in improving decision quality. 

• Weather forecasters employ supercomputers to predict multiple weather patterns – impacting lives and the economy.  
• The military performs war games and exercises. Do organisations perform scenario drills? 

In  the same vain, simulations enable an organisation to

• Refine the best course of action for extreme scenarios
• Build confidence and reduce stress based on having an actionable plan in place
• Turns theory into reality. The brain is tuned into the potential of multiple extreme realities.

If Simulations are so great why are they not used all the time? 

1. Unknown scenarios.  We only know what we know.
2. Cost.  The more realistic, the more cost. However, the better the decision. 
3. Ego. Certain people do not want to be proved wrong – and would rather not see or accept the results of simulations.

Having a model enables simulated scenarios to be tested against them.

Asking those “What if” questions.

• “What if” one of our key supplies increased subscription prices by 20%
• “What if” we won the largest contract in the organisation’s history.

How would the decision model handle it?

Remember – the resulting decisions from these scenarios will impact employee lives. “Bonus payment”, “Promotions” as well as “Cuts” and “Job losses”.

Computers have been used to simulate weather, scientific experiments, finance and other fields for decades. “Organisational Decision making” is less mature and widely used despite the huge advantages. Cost, expertise and mind set  are usually the limiting factor. 

Simulations are the key reason for adopting organisational decision models to improve decision quality and productivity. As digital-twin and metaverse technologies improve, creating realistic and interesting simulations becomes more feasible for smaller companies.

Crisis Composure

“I included this section at the last minute – I had to do it” 

In a crisis, decision quality is heavily impacted. Why? Stress, Worry and Anxiety. 

Making coherent decisions in a crisis is the hallmark of a “star player”.  Experience has taught them to change their perspective and remain calm. Their outlook is completely different to those who panic and freak-out. Hence the saying  “the rule book goes out the window”. 

Having a decision model which covers crisis scenarios greatly enhances decision quality under stress for “non star-player”. Macro resilience improves.

• “What if Sales dropped 50%”
• “What if Sales increased 50%”
• “What interest rates increased 3%”

Notice that a crisis can also mean very good news. Over-excuburents can be regulated with pre-defined decision model pathways. 

For example, Sales are up 60%! Crisis (excitement) snap Decision : increase workforce by 40%. One year later interest rates increased 8% due to inflation. Sales drop 50%. Crisis snap Decision : cut Workforce 30%. Remember, families and dependents are impacted by such decisions. Thoroughly simulating such scenarios would reduce their negative impact. 

From an operational perspective. Mini crises can occur once to many times per day. Ranging from outages to mass churn events.

For example, Call centres were the initial use-case for Next Best Action. Why? The stressful and high-turnover Call-Centre environment can cause inexperienced staff to “give too much away” to retain customers – discounts, equipment, etc. 

Future ready

Realistic Simulations will play a large role in future “decision-first” organisations. Future structures will also evolve from the traditional hierarchical constructs.  Interdependence and greater democracy will be how they evolve. 

Decisions in a Distributed Organisation

How can Decision Model driven organisations evolve?

As they scale they become more decentralised and distributed with greater degrees of departmental interdependence. Resilience and agility increases combined with an overarching belief and understanding of the organisation’s vision. Decisions speak louder than words.

 Decentralised and interdependence. What else could this mean?

Interdependence

Organisations will also move towards a more distributed way of working where decisions at all levels are made closer to the action feedback. Employees at all levels would have the capabilities and responsibility to make decisions at all levels. They will also share in the value generated by their decision. However, employees would need to prepare for such an empowering operating model – the technology already exists. The imagination and courage required, will soon catch-up.

Democratic decisions

There is a historic pattern where strategic decisions are made by C-suite executives, tactical decisions are made by middle management and the front-line workers make operational decisions. Moving towards a more Decentralised Autonomous Organization (DAO) approach in making strategic and tactical decisions should lead to greater organisational resilience and greater loyalty.  Employees at all levels can be involved in shaping key decisions within the decision models framework.

Decisions in Auto-pilot Organisations

If you believe the AI hype, then more and more levels of operations and management will be assisted or replaced by it. First by AI driven processes (classification and automation) . Then AI driven decisions (Next best action). Organisations with clearly defined decision models would have a strong basis to move in this direction.

Taking Better Action

Remember, without action ( real action – not simulation) we cannot progress and learn from decisions made. Taking decision-model driven actions will yield better results. 

The actions taken will automatically take on the authority and confidence of a researched and simulated decision. Actions driven by predefined process management would be even stronger. 

The judicious use of decision management and process management sets the organisation on the path of autonomy and then auto-pilot status.  

Summary

Decisions are indeed the foundation of an organisation’s success. Performing rapid and effective decisions ensures that the organisation adapts to challenging situations and moves forward. Operational decisions are where DMN typically  lives. The millions of decisions performed at this level form the tapestry of the culture of the company from a customers perspective. However those decisions are driven from other decisions which are not explicitly modelled. Strategic and Tactical decisions.

Impedance mismatch is a term used in computer science to describe the problem that arises when two systems or components that are supposed to work together have different data models, structures, or interfaces that make communication difficult or inefficient.

The role of organisational decision models using DMN  to reduce Impedance mismatch in decision making.

Finally, do organisational decisions matter to customers? YES at every level. Better decisions mean better service, product quality and greater product longevity.