One of the overlooked benefits of AI is that every enterprise now has a Big Data beach. A “Big Data beach” forms when AI FOMO encourages management to allow endless waves of operational data to wash up on enterprise systems. Some might think terabytes of 1s and 0s will form a sunny expanse like those in a Corona commercial, where you settle into your beach chair, dig your feet into the warm sand and enjoy an iced tea (or beer) in the comfort of knowing that “all our data is somewhere on this beach if I need it. Ahhhhh …”
A recent experience makes me think a better metaphor might be a Great Lakes agate beach. The agate beaches on Lake Superior are made up of plain-looking, wave-smoothed little nuggets of basalt, granite, chert and other indistinguishable stones. Agate beaches are not particularly attractive and are difficult to traverse barefoot, but their ordinary-looking stones have been foundational materials for roads, building facades and the manufacturing of sandpaper, if that happens to be your passion.
What you might not know, however, is that every year these nondescript expanses also attract hundreds of enthusiastic children and rockhounds of all ages who know them as great places to hunt for and find the highly prized, multicolored, banded gemstones that everyone loves to have on their desk: agates. “Agates are way cool,” says my grandson. “Can we go back and find some more?”
Operational Data Isn’t Just for Operations Anymore
I came to the agate beach metaphor while helping a startup team that is using AI-based video generation to create daily construction safety training videos in more than 75 languages. When I started digging into how they might measure and prove the value of an easy-to-use, multilingual video training platform, I discovered that not only do the safety teams of construction companies enthusiastically embrace all types of data, they also are very mature in the use of that data because worker safety is critical to a construction business’s operational performance. Consequently, just as most established contractors have FICO or PayNet credit scores to help them work with lenders, they also have a workforce safety score called EMR — the experience modification rate — that helps with operations. The EMR score is the metric used by insurers to determine what the contractor pays for workers’ compensation insurance. It’s also a go-to prequalification metric that project owners and general contractors use to decide whether to allow a contractor to bid on a project. EMR consolidates a company’s historical workers’ compensation injury claims to predict future injury risk. That sounds familiar.
For an industry known for being slow to adopt digital technology, I was surprised to discover how extensive and sophisticated the data behind EMR is. Calculating an EMR requires workers’ compensation claims and payroll data classified by the type of work and severity of injury. The measure is normalized for company size and compared with similar employers doing similar types of work. A company’s score typically is based on three years of data, and the calculation requires ongoing data collection, maintenance and analysis. Construction company leaders are highly motivated to get and manage their EMR because of how insurers and project owners use the score. Indeed, I found that construction company executives take great pride in a low EMR score as a measure of the company’s discipline and performance. I thought to myself, “This little agate feels a lot like a credit score.”
A Safety-First Culture Is a Better Credit Risk, and EMR Measures It
The credit score connection is more than a feeling. Workforce safety is a coveted cultural discipline for a construction company to help deal with a critical operational risk. Operational risks create financial risks. Character, or culture, is one of the five Cs of credit. If EMR can affect the company’s access to revenue and indicate costs that reduce margins, then it certainly could inform underwriting.
Consider two construction excavation contractors with similar financial statements, credit scores and years in business. Excavation contractors are notable in this case for two reasons: They are the keepers of the yellow iron on a job site, and when I asked a general contractor site manager which subcontractor trades he worried about most from a safety perspective, he said, “Excavators: They can be cowboys.”
Imagine the first excavation group is not made up of cowboys and has a consistently low EMR and a history of disciplined operational performance. But the other has a deteriorating EMR from a pattern of more workplace injuries. The second contractor faces not only higher direct workers’ compensation claims expense but also a safety multiplier effect from the indirect costs of lost productivity, schedule disruptions and replacement labor caused by the injuries. These costs can erode margins and jeopardize future cash flows. Further, a higher EMR will likely limit access to future projectswhere project owners have safety qualification standards. The first contractor may be better positioned to qualify for larger, time-sensitive projects like the growing number of data centers. More and bigger projects create additional equipment demand for an equipment finance partner. Traditional credit analysis might consider these two contractors comparable, but they are not.
So why haven’t I seen EMR scores in underwriting data sets? I spoke to several lenders serving construction customers, and none of them had heard of EMR. I spoke to the safety leaders of several construction contractors, and none of them were aware of their finance peers using company EMR scores to help negotiate lending. I asked a contractor association leader whether any members use EMR scores with financing partners. The leader pointed to negotiating with workers’ compensation insurers, “obviously,” but was not aware of anyone using EMR in lending negotiations. When analysis stays in its lane, interesting data relationships are missed.
Agates Can Only Help Those Who Covet Them
Big data resources in any enterprise are nonhomogeneous, difficult to navigate and hard to mold or shape into useful constructs. An equipment finance data warehouse schema is a complex tangle of asset dates, borrower data, funding data, industry parameters (North American Industry Classification System codes), economic parameters, asynchronous timelines and sometimes asset usage information. The data comes from dozens of sources but typically is analyzed in specific niches rather than as a whole. Equipment finance big data is a rocky agate beach that must be reviewed, analyzed, sorted and prepared with purpose if users are going to find any agates for the credit officer’s or CEO’s desk.
The good news is that the same trend creating these rocky beaches, AI, is also making them easier to search. AI makes it increasingly practical to explore new types of data and data relationships, test hypotheses and discover useful patterns across data sets that weren’t designed to work together. EMR is foundational to insurers and project leaders, so why not financiers? Other agates might be found among operational data such as equipment utilization, maintenance discipline, workforce turnover and on-time performance, which may have predictive value that traditional financial statements and credit scores miss.
Finding agates like EMR on a Big Data beach will come from directed searches to answer difficult questions such as “Which of my construction contractors is most likely to grow in the age of AI data centers?” The promise of a Big Data beach is not comfort or relaxation. The promise is “with the right analysis and expertise, we can measure almost anything.” As I told my grandson, the agates are out there. You just have to get to work and find them.
Scott Nelson is chief technology and product officer at Tamarack Technology.
