More than 10 years ago, I created one of my first robust driver-based models in Quantrix for a startup that developed a SaaS-based POS application. In 2024, the owner sold his business for a sum that would have caught the attention of Inc. Magazine or John Warrillow had they known about it.
When I use the term driver-based, what modelers really mean is that small but significant parts of the model are based on key drivers such as units and rates. For example, a unit could be the number of customers or the number of transactions. Rates could be the average deal size by customer or the average order size by transaction. Other parts of the model will not be driver-based but rather input-based (e.g., the monthly lease rate for a warehouse in the south region).
Joe (my client) and I spent the majority of our time updating the user count each month. One user was worth about $110 per month. Joe projected new additions for this three-person sales team. He also projected drops (we preferred that term over churn).
There was nothing scientific about the process. Joe’s gut feel was great because he had a sales background. Plus, we had history to support his forecast.
When I built the model, I never liked its clunkiness. Instead of explaining it, let me show you.
Modeling Tips:
Thank you, Ben Wake, for your great presentation at the 2022 Dimension Conference. I now use the pairing of the LAST and SELECTS functions in all of my models. You are a genius. If you are a Quantrix user, here is his presentation.
I glossed over the concept of helper items in the video. Earlier in my career, I worked around a gifted CTO, numerous developers, and a savvy database manager (thank you, Denise Pierce). By being a fly on the wall at many of their meetings, I learned about elegant code and solutions.
Unfortunately, I tried to apply the elegant code concept to my Quantrix modeling by making formulas in the statement editor as concise as possible while being impactful. And then I started watching Rich Lopez on his channel. About every other video includes a helper item. “Hey, isnt’ that cheating?” No, not at all. It’s clever, and it helps new modelers slow down their thinking by making it easier and even faster to build seemingly complex formulas.
In short, use helper items liberally in your models, which also improves the auditability and documentation of the end product. Also, if you are a Quantrix user, watch every Rich Lopez video. While you’re at it, add Tahir to your list, too.
