Skip to content

Alok Mohindra

Chapter 1 - The power of models

Novel, Highlight4 min read

The power of models for solving hard problems

Models are everywhere and yet we hardly ever notice them. They exist in everything mankind has ever conceived. Whenever he needed to make a decision based on his understanding of the behaviour of some complex real-world or man made system a model was integral to the decision. Always the model was wrong but sometimes the model was useful - occasionally very useful.

A model is a simplified representation of some underlying system. The level of simplification one applies depends on the primary use-case of the model. Typically a model is either a working model or a theoretical model.

  • A good working model helps highlight the major parameters that drive changes in results while minimizing complexity by simplifying higher-order effects so decision-makers can focus on understanding the tradeoffs required to achieve the primary drivers of success.
  • A good theoretical model is designed to enhance one's understanding of the underlying system of interest. Theoretical models help us understand with greater nuance the higher-order effects of changes to input parameters on the system stability, productivity and health. As such theoretical models generally articulate greater detail and more sophisticated understanding of an underlying system under study. This greater accuracy comes at the expense of a higher cost of design, development and maintenance and also at a higher risk of making the fundamental model logic opaque or difficult to understand
In every case a good model ..
  • is defined in an easy to understand structure that unambiguously describes the detailed parameters of the model as well as appropriate inputs and logical relationships
  • is defined in such a way that a non-technical domain-expert would be able to manage a model with minimal training in only a few self-service sessions.
  • captures the attributes and relationships of a system in sufficient detail to represent similar behaviour to the underlying system under various inputs and loads.
  • captures necessary detail for communicating behaviour while ignoring details unnecessary for understanding the behaviour of the system.
  • clearly and unambiguously defines attributes and relationships of the system and their contribution to system behaviour
  • tracks data provenance for all facts and assumptions provided as inputs to the model, by recording who provided the data, with what rationale, at what time, on what IP address, with which credentials - as well as tracking any subsequent changes to input data
  • aims to continuously improve to help ability to clarify decision-making and make models useful for gaining insight
A good model is useful

Good models represent a system worthy of understanding. Definining the edges of a system is another blog post, but I'll attempt to explain by way of an example:

In most cases your team are already using models to manage each of these business functions independantly but since it is ad-hoc the results are ephemeral and inconsistent.

By embracing modeling at the strategic level of your organization you empower your team to think and act strategically at scale allowing you to leverage your strategy on a global level all at the speed of internet.

How do I get started creating a model-driven organization?

The first step is to clarify the models you currently use to manage your organization and more specifically how that relates to your current role. Imagine what it would look like for everyone in your organization to be able to clarify the value they contribute in a meaningful way, such that it rolls up to a relevant strategic operational activity or supporting function driving the success of the business.

  1. Make a list of all key input variables that affect your current business model at the corporate level, using as fine-grained resolution as possible (e.g. Number of unit sales of product A for $1.20, Cost of $0.80)
    • Work with finance or accounting to develop a 3 year cost and revenue projection, clearly separating assumptions affecting input cost and revenue drivers to make it easy to communicate business logic
  2. Make a list of all key business functions and consider prioritizing the development of function-specific models based on overall business impact. Ideally developing these business models should happen using an iterative approach with each function being further developed as a model on each iteration.
    • Leverage the help of your finance peers, they have been building quantitative models for years.
    • Don't settle for spreadsheets. They are an archaic technology that makes it virtually impossible to make really great models. In 2020 it is time to as for more than a spreadsheet. Make a better model instead.
Dedicated to Lord Ganesh

This is a the first post of the rest of my life. I devote this post to the lord Ganesh who brings good fortune to those undertaking a risky but otherwise worthwhile adventure. He is the god of entrepreneurs in India and around the world. Without a spiritual purpose it is difficult to find a true vision for a better world.

This is an H2 Header

And this is some text underneath

This is an H3 Header

And this is some text underneath

This is an H4 Header

And this is some text underneath

This is an H5 Header

And this is some text underneath

This is an H6 Header

And this is some text underneath

This is an example link.

This link has no title attr.

This is an example reference-style link.

single asterisks

single underscores

double asterisks

double underscores