To say we are living in the data economy would be a huge understatement. Most organizations these days understand that the data they’ve been gathering is not only currency, it’s among their most valuable assets, especially if they are willing to wring every last ounce of potential out of it.
To that end, many businesses are turning to data science, a technology that has the ability to help companies better understand and analyze their data and, in turn, make better decisions, faster.
Layton Construction is one of those businesses. The Utah-based commercial contractor with projects all over the country recently passed a new milepost in its progressive, three-year data analytics journey by implementing a data science solution. The reason was simple.
“We had a lot of information in one place (Domo) already,” Jeff Metcalf, the firm’s VP of information systems, said during a Get Leverage webinar earlier this month, “and we wanted to see what it could yield us in terms of predicting business events.”
So far, so good. Since choosing forecasting as its first data science use case, Layton Construction has shortened its month-end close cycle and saved countless man hours on reporting.
“These are two really big wins for us,” said Jeff, who has worked at Layton for almost 20 years and has also used data to decrease the company’s incident rate by a whopping 50%, which undoubtedly has saved lives and reduced injuries.
Layton Construction was able to score quickly on the data science front because its use case “checked all the boxes,” said Domo data scientist Vishakha Shenoy. “They had a meaningful business question, quality data, a great action plan, and a person in Jeff who had both expertise in the business and an understanding of data science.”
Jeff and his team also understood from the very beginning that they had to have patience.
“We didn’t go into any of this thinking, ‘We have to be done by such-and-such date,’” Jeff said. “We knew it was going to be a long process and that we were going to have to go back multiple times to refine the model. That’s just the way it is. There’s always another question or another anomaly that you have to account for.”
Patience is doubly important when the goal is a good production data science solution, which is exactly what Layton Construction was looking for.
“Run-of-the-mill data science is mostly about building a good model,” Vishakha explained. “Production data science, on the other hand, is about automation. It’s about building a data science model that not only fits your training data, it fits your live data. And that requires a very clear link between business acumen and data science expertise. All this takes time.”
For Layton Construction, the results have been worth the wait.
“We now have the ability to analyze the data a lot more quickly within the system,” Jeff said. “We can push a button at the end of our close process and have those calculations done for the forecast.
“Also, the model is actually learning as we’re going, so we don’t have to retrain it. And as project complexities change over time, and maybe different scopes of work are added or different businesses are spun up within our company, we can know that the model will be able to adapt and adjust to that.”
For those looking to follow in the footsteps of Layton Construction—which is to say, apply data science in an attempt to solve complex problems and gain separation from the competition—Jeff has three pieces of advice:
- Make sure you have the data to train your model off of;
- Make sure your efforts are going to provide a meaningful way for you to improve the performance of your organization, and;
- Make sure that everybody involved understands what you’re working toward.
“That second piece is really important because a lot of times people make decisions without knowing if the juice is going to be worth the squeeze,” Jeff said. “Data science projects are a process. So, make sure that what you’re going to get at the end of that process is going to be useful.”
To learn more about how Layton Construction is leveraging data science, and to hear why Vishakha opted to build a causal model for Jeff, watch a replay of the webinar here.