#tldr: wrangling a complex project often means starting in the middle

When we approach projects in the workplace, we tend to have a very linear approach: start at the beginning, make each piece, usually serially, put it all together, and ta-da, the model is complete. While this approach is perfectly suited to standard projects, it completely falls apart when there is a fair amount of complexity in the model or when working with an existing system that needs to be significantly altered. Most planning models are not created from a blank slate, but rather, from pieces together existing, disparate pieces of a system that may include many platforms, a lot of manual intervention, exceptions, and other non-standard, but very real world, conditions.

We don’t work in a pristine, academic environment, but, rather, a complex, messy, real environment where things don’t always fit neatly together. This can cause a lot of frustration and communication issues between the business and the developers of the planning system, since standardization is not always desired by the business, but is obviously a by-product of building a robust planning system. How do we harness this complexity, and respect it, without losing our minds?

I often think of the models I design as bridges – connecting disparate parts of the organization into a unified whole, a bridge that brings information across the river to different parts of the organization. For example, in Biotech models, we are frequently connecting the supply of various types of skillsets in staffing (chemists, researchers, project managers) to the demand for those skillsets by project. Another example: matching sales forecasts with production capacity to ensure orders are anticipated and met. Building a bridge is, in itself, a complex task, and best accomplished with planning and foresight.

When working with an existing system, one that may or may not be working the way the business wants it to, we often are lacking full information and lacking the foresight necessary to proceed in an orderly fashion. For example, I was once asked to improve upon a model with hundreds of moving parts – cubes, data sources, connections between different parts of the model, etc. This model was a fairly standard system with financial revenue and cost information, as well as product build information, royalties and licensing, and the other usual complexities of a real-world model. But no one knew how, or even if, it worked.

What to do? In this type of situation, I employ what I call the “Rocks in the River” approach to building that bridge. Picture yourself standing in the middle of a busy river, waist-deep in water. What you want is a bridge, but what you have is wet clothes. So, you start feeling around the riverbed for rocks, and start collecting them together in a pile that you can stand on. For our FP&A model analogy, this is like starting with one module, say, Royalties & Licensing, and understanding how that bit works, if it works, and what it is connected to. How are Products defined? Where does this flow into the P&L? How are new licensing agreements encapsulated? Now, we have our first pile of rocks and can at least stay dry!

As you continue to build out your understanding of both what the model is doing, and what the folks using the model would like it to be doing, your bridge grows. As you move along the path, following data sources back to their origin, following meta-data structure updates, following calculations that are dependent upon other parts of the model, you can start to see how the whole system works, and have at least a rudimentary bridge across the turbulent river. Well done!

It can be uncomfortable at first, because our brains don’t love starting in the “middle”.  But before you know it, you’ll have a fuller picture, and be able to install the building blocks of a real foundation to your model. Hopefully, the fog will start to clear and you’ll be able to see more clearly such things as:

·       How the business activities flow through the model

·       How Actual and Forecast data interact, including the different levels of granularity for each

·       Where the sub-models tie into the bigger picture

·       What is required for inputting from users in order to “make the model go”

·       What user interfaces will make using the model intuitive

For users of Planning Analytics, there are a couple of tools Cubewise offers that help greatly with this:

1.     Pulse: this Planning Analytics/TM1 monitoring tool with many features. In the case of this article, what’s most relevant is its ability to tell you which objects are connected to which other objects – cubes, dimensions, processes, data sources, the whole ecosystem. And, one of my favorite features is the ability to search for, say, the name of an attribute, to see if its used in any Rules or TI, for example. Learn more about Pulse here.

2.     Apliqo UX: This is a front-end for Planning Analytics/TM1 that allows you to create a more guided experience for your end users, thus taking the guesswork out of things like: which version to enter forecast data into, where the interdependencies in the model exist, and how to summarize the end results. Learn more about Apliqo UX here.