Data driven decision making (DDDM) is easier said than done — Decoding the belief system
In Short
- Decision-making is simply a process of making a choice. Data-driven decision-making (DDDM) means analysing and interpreting data to make the right choice. DDDM can also be interpreted as establishing a seamless connection from data -> insight -> decision -> action.
- It is a valuable skill and most organizations advocate for its implementation. It enables all levels of the organization, from the CEO and senior management to business analysts and engineers.
- But then, establishing DDDM in practice can be more complex than it appears. While most technical papers and blogs focus on technology, tools, and processes as bottlenecks, I am writing this blog to throw some light on underlying beliefs, attitudes, or perspectives that have higher influence and often overlooked aspects of DDDM.
Earliest use of data-driven decisions
Before we dive into the specifics, here’s an interesting fact that explains the earliest use of data.
The ancient Egyptians developed a sophisticated system to track the Nile’s water levels, primarily for agricultural planning and resource management. The method involved using markings on stones or other structures along the riverbanks and these markings served as indicators of the water levels during different times, especially during the crucial flooding season. They even trained pigeons to communicate the data. Interesting: Read more
Although the industry has made significant progress with processes, tools and technologies from ancient times, we still have belief systems that hinder DDDM.
Let us decode the major ones.….
#1. DDDM IS NOT JUST ABOUT NUMBERS
It is much more than just being good at numbers. In the world of measurements, each number corresponds to a specific term: Measure, Metric, or KPI. Knowing what these terms mean is important to understand and interpret the underlying data.
- A Measure is a value with a unit resulting directly from a measurement. Ex: # of Miles
- A Metric is a combination of one or more measures that evaluate the performance of a specific process. Ex: Mileage: # of Miles / # of Litres
- A KPI is an essential metric used to measure something significant. Ex: Mileage as a metric measure the engine's performance. A potential KPI.
In the entertainment industry, the Q-Score measures an actor's popularity and recognition by converting ratings into a numerical value. Over time, they named it the “Tom Hanks Q-Score” because Tom Hanks is like the most-loved actor. His name became the gold standard for this metric. So, when you hear “Tom Hanks Q-Score,” it’s like they are essentially asking how much audiences love that actor and whether their movies are likely to be successful. Interesting: Read more
Similarly, in the business world, comprehending these measurement terms is like understanding the story behind the numbers, which predicts the business's success. So, It is crucial to have a solid grasp of Measures, Metrics, and KPIs if one wants to make informed decisions based on data.
#2. NOT A FAVOURITE KPI
How many times have you found yourself disliking a specific Key Performance Indicator (KPI) for no obvious reason? Often, the reason for the dislike is a lack of understanding. On the other hand, you may not want to understand a KPI because you don’t like it.
KPIs make sense over time when actively in play.
No university or schools taught us the real world KPIs like mileage, speed, heart rate, Interest rate, steps count, accident rate, and many other but we still made sense out of it over time and understand them through experience.
Currently, you may not like a KPI because you aren’t sure how it will operationally influence the outcome. However, with time spent on the KPI, you will eventually learn what factors influence it.
Consider the number 35. As a standalone figure, it doesn’t hold much significance unless you know the KPI it’s related to. For instance, if it represents Mielage as 35, which means 35 miles per litre, then you can make a well-informed decision.
Some of us may dislike certain KPIs due to past experiences, as they may encourage negative behaviour.
As you gain experience, you will learn KPI’s help provide feedback on the potential things that we can improve; and are not actually the driving factor to hit our targets. They are measurement and evaluation tools to track progress and efficiency. Depending on the project’s context, a KPI may be effective or ineffective in bringing about the desired change.
Finally, It’s okay if you don’t understand a KPI at first, but don’t let that make you dislike it.
#3: YOU ALWAYS HEAR THE CREATOR'S VERSION OF THE STORY
When the people who collect data and measure KPIs are not the ones who own the data or the KPIs, the interpretation of the data can often be shaped by the storyteller's narrative, rather than the actual meaning of the data.
It is a symptom of absence of Data governance (data policy).
It is critical for an organization to have indisputable ownership of data and the corresponding measurements. It cannot be left to individuals who have access to the data to perform these calculations on their own.
Additionally, KPIs and data owned by a specific team may not be taken seriously outside of that department. This lack of trust and consideration arises from a siloed mindset, where data outside one’s domain is viewed with scepticism.
#4: DDDM IS PERCEIVED AS A SENIOR MANAGEMENT SKILL
Decision-making is often perceived as a responsibility that solely rests with an organisation's top management or dedicated data scientists. However, this belief is not entirely accurate. In reality, decision-making takes place at all levels of the organization, ranging from senior executives to junior staff, across both management and technical domains.
Ex: Frontline Engineer & Mid level lead Decisions
Defects, bugs, code coverage, code complexity, static code violations, security vulnerabilities, Cycle time, Test failure rate, Response time, Memory usage, CPU usage, Availability rate are just few examples which needs decisions at developers and lead level in software development.
Decentralized decision making is key.
You may be surprised if you assess how often frontline engineers engage in data-driven decisions and their skill level in driving such decisions.
One question to Ponder is: Are leaders making more decisions than they need to?
Yes, if the next level does not make enough decisions.
#5. DDDM — A CRUCIAL SKILL OFTEN OVERLOOKED IN TRAINING PROGRAMS
An organization’s training calendar typically prioritizes domain, product, and process training. However, Decision-making and problem-solving (DDDM) skills are essential for all employees, regardless of their position. Unfortunately, this crucial skill is often overlooked.
Most of us assume that when data is available, people will make decisions based on it. However, this is not true unless people are familiar with the measurement terms and know how to interpret the data. It is a skill that needs to be groomed.
Data literacy is the ability to read, write, analyze, communicate, and reason with data. It’s a skill that allows individuals and organizations to make better, data-driven decisions. Figure out whether your training calendar covers “Data Literacy”.
It is a pressing need to change the mental model and embrace a training approach that focuses on the development of essential skills in employees, rather than relying solely on the skills checked during recruitment.
Conclusion
Leadership, Data Governance, Single source of truth, and Tools & Technology are well-known DDDM hindrances and worked actively across organizations.
However, understanding DDDM from your organization’s belief system & culture is even more crucial. No single heroic or departmental effort can eliminate these mental models. A systematic and collaborative effort across organizations is required to address them.
Remember, the biggest obstacle in change management is comprehending “how people perceive” the change.
As an end note, With AI becoming more prevalent, it will be interesting to see where data-driven decisions will go from here as AI models do not come with preconceived belief systems.
Credits:
I would like to thank Michael Backhaus & M Bubesh Kumar for providing your valuable insights. It was a great collaboration with both of you.
