WePredict Ready to Rock Component-Quality Analytics

WePredict's Deepview study promises the capability to more-accurately project automotive component-repair frequency and trends—with clear benefits throughout the automotive value chain. (WePredict)

Somewhere near the zip code of Big Data is the emerging neighborhood of Big Analytics, where computer scientists, research analysts, mathematicians and statisticians intake the ever-deepening information that Big Data offers and do something previously undoable with it.

WePredict executives don’t carelessly toss around the overused term “artificial intelligence,” but they do assert machine learning is a part of their new-age data alchemy of “predictive methodology” they promise will open new vistas of understanding about repair frequency for all manner of automotive components. The company’s first wide-ranging study of predictive repair frequency is called Deepview and launches this fall.

“Big Data’s been there for a while—but nobody’s really known what to do with it,” Renee Stephens, WePredict’s vice-president, North America, told Automotive Engineering in a late-summer interview at an industry conference. Stephens has been around quality-related automotive data for just a bit: she was the quality data director for GM, where she spent nearly three decades, and before joining WePredict in 2017, she was VP of U.S. Auto Quality at J.D. Power.

WePredict VP Renee Stephens: "Ultimately, consumers will be the beneficiaries" of the company's sophisticated predictive analytics for repair of automotive components. (WePredict)

She said WePredict’s data researchers—the company has headquarters in the United Kingdom and in Ann Arbor, Michigan—create modeling information, then further apply their varied disciplines to “provide the analytics behind it. It started out being an art—now it’s a science.”

Information is king

The promise of WePredict’s science seems nothing if not immense. The company’s ability to extrapolate how an automotive component should perform might arm engineers, product developers, planners and warranty specialists with previously unimaginable insight into specifying components and predicting how they will perform once in production vehicles.

“It helps an OEM see what they couldn’t see before,” merely studying patchwork warranty and customer-response data, Stephens said. And automaker product-developers won’t be the only beneficiaries. Of equal importance, she added, WePredict’s data will fix “a big blind spot” for suppliers, who to now have had to rely mainly on inconsistent and limited warranty data from automakers regarding the quality and reliability of components.

And because of the historically secretive nature of warranty data, a tool such as Deepview, the company said, will provide automakers and suppliers with reliable information about how their components compare to those of competitors. Not just in the past, but in the future, too.

“Now what we’re adding is all the competitive data,” Stephens said, adding that all the data the company secures and disseminates is anonymized.

James Davies, WePredict’s founder and CEO with actuarial roots at Lloyds of London, said the company currently has access to more than 10 million vehicle identification numbers (VIN) in the U.S., which serves as the basis for the first Deepview predictive automotive-repair study. Data from Cox Automotive indicates the current U.S. automotive parc is roughly 270 million light vehicles, so WePredict seems to have a solid starting sample that will expand along with the Deepview study, which the company intends to generate bi-annually.

Beyond traditional reporting

The data WePredict is collecting also extends beyond the usual methods. Information from the service industry comes not just from warranty repair, but also so-called “customer-pay” service work as well—an aspect of the repair network that has a blotchy reputation for inconsistency and inaccuracy. Stephens said one important aspect of the company’s initiatives for Deepview also encompasses a start at standardizing repair-order forms and communizing the language.

“This is an industry first,” Stephens asserts, confidently adding that WePredict’s methods are so proprietary that it has no competitors.

If the merging of Big Data with Big Analytics goes the way WePredict ah, predicts, the $2-trillion auto parts market is in for a virtual reformation. And so, too, the engineering and development process:

“I don’t want to wait three years to see what’s going to happen with my 2018 product,” Stephens said. And WePredict reckons after the launch of Deepview—which she said can cover components ranging from entire transmissions to window switches—the industry might wonder how it ever did business without it.