“You cannot predict the future without having a very stable, validated, reliable view of what happened yesterday,” Jennifer Vanderveldt, CFO of Mammoth Holdings says in the first six minutes of our interview.
That statement sets the tone for our Zoom call. Vanderveldt is former head of strategy, consumer insights and analytics, at craft retailer Michaels, where she created a center of excellence for customer analytics. At Mammoth, which she joined in May of this year, she’s heading up the build of an automated data warehouse on the Snowflake data cloud platform. The project involves organizing and structuring disparate data from multiple locations and multiple point-of-sale vendors.
The data comes from all the local car washes (more than 100 to this point) being acquired by Mammoth, a private-equity-owned multi-brand organization. Giving the car wash business scale lets Mammoth centralize procurement, customer service, human resources, and back-office functions.
“Great operations” is the key to driving growth the in the business, said Vanderveldt. “It’s what the customer primarily sees first from us. … A lot of folks don’t know the name of their car wash, they just know they get a clean, shiny, dry car.”
But it’s the data generated by the operations, purchases, and customer behavior that will help guide decisions on where to look for those growth opportunities. Among the topics Vanderveldt and I discussed were the CFO’s role in building a unified source of truth and where the analytics function should sit within a company.
CFO, Mammoth Holdings
This interview has been edited for clarity and length.
JENNIFER VANDERVELDT: We’ve used the most cost-efficient approach to building a data and analytics platform, which is a partially offshored model. I brought over a director of data science. And we use a data analytics shop in India because highly trained data engineers are hard to hire in the United States right now. … I needed someone in the [director chair] who understands data and analytics and who understands the application of all of that to the problems I’m trying to solve as a CFO. The woman in that chair is a data scientist, but she got her MBA in finance and entrepreneurship. So, she thinks about business questions strategically, and then about the right methodology from a data standpoint. The person [in her position] also [has to be able to] manage an outsourced group of highly trained data engineers to do some of the hands-on keyboard work. She’s a purple squirrel.
VANDERVELDT: The first major win is when we start having daily sale reports and scorecards [across all the car washes]. I want to empower my chief operating officer Corey Joslin by putting the right data at his fingertips so he can drill down into site-level information on cars run per labor hour — that’s a key metric. And I need to provide that data on a daily and even potentially hourly basis.
The CFO needs to become a data champion. A data champion wants to understand enough of how the clock must be built such that the clock consistently tells the right time.
If the data structuring has been done correctly, then we can start predicting off of that data. I can drive better efficiency by using the data to build a regression model. And then, in time, have an [artificial intelligence] model that is self-correcting, taking what happened yesterday and modifying a prediction for tomorrow without a lot of human intervention.
By hooking into our cloud-based [human resources information system], we can also determine how much labor at each hour of the day a site manager will need. A dynamic labor model will get us closer to the efficiency frontier of peak labor utilization, which should drive points of margin.
VANDERVELDT: Exposure to data analytics and intellectual curiosity. A reductionist CFO might say, ‘I just need to know what the answer is.’ Just give them data and a fancy business intelligence tool, and they’re going to trust that it’s right.
But the CFO needs to become a data champion. A data champion wants to understand enough of how the clock must be built such that the clock consistently tells the right time. The CFO needs to ask questions like, ‘How is this data being inducted into the database? How is it being extracted out of a source system? What types of transformations or normalizations of that data are being effectuated such that when I want to look at weekly comparable store sales, I know one of my sources of error is not how the data was structured in the database?’ … The CFO needs to be the champion of the right way to do something, not the most expedient way.
VANDERVELDT: I want to make sure that anything I put out there I can defend. I think it is incredibly important for the office of the CFO to own the data [and] own the data stack and data instrumentation. And to own all of the data models, data forecasts, and financial forecasts. They are important to how companies are valued externally.
In younger companies with fewer controls, I think that the IT department, the systems, and the analytics powering data, should roll up to the CFO. … But if I were to hypothesize around C-suite roles 10 to 15 years from now, I think mid-cap and larger companies will have a chief analytics officer. I think that [position] should report directly to the CEO and be completely unbiased by other departments [that may] manipulate data and try to provide vanity metrics. The analytics folks wouldn’t own a P&L. They would just give the CEO and the CFO a very unvarnished opinion on what is really driving the business.