Dealers, lenders say AI decisioning leads to inaccurate digital auto retail quotes

Ninety percent of dealers and lenders feel the use of artificial intelligence to evaluate consumer creditworthiness has led to shoppers receiving inaccurate car payment quotes online, according to a December poll by auto finance software provider eLEND Solutions.

Fifty-four percent of the eLEND survey pool said customers negotiate payment terms online before the actual lender decision. The study polled more than 300 respondents — 76 percent from auto dealerships and 24 percent from auto lenders, eLEND said in a Jan. 31 press release announcing the results, which it called "extraordinarily consistent across both cohorts."

Under a traditional rate sheet, a lender divides customers into different tiers of risk based on credit score alone, with each tier uniformly assigned a certain interest rate. Lenders with AI decisioning consider more variables from a consumer's circumstances and produce a rate more specific to the individual. Eighty-nine percent of dealers and 83 percent of lenders agree lenders are providing fewer rate sheets.

"Lenders aren't even pricing to those tiers any longer," eLEND CEO Pete MacInnis told Automotive News. "Some are, but most of your major ones have this dynamic pricing model" tied to the unique customer and their vehicle choice.

A rate sheet's more simplistic rules can be fed into an online payment calculator for an estimate of a consumer's likely monthly payment. But an AI model's decisioning can't be as easily predicted by third-party payment calculator software; access to the lender's "black box" outputs is necessary for an accurate answer.

But online car payment quotes typically lack lender access, the survey found. In fact, it's not even a given that an online payment calculator would have the right credit score to make an accurate calculation off a rate sheet, the study found.

Sixty-four percent of respondents agree initial payment quotes online most commonly draw from a customer-provided credit score. Only 17 percent said payment quotes are calculated based on checking a consumer's actual FICO score.

"In a self-identification world, rarely are people correct," said Richard Herod III, managing partner of White Bear Mitsubishi in White Bear Lake, Minn. He said the actual credit score can be higher or lower than what the customer says.

"That's the part of the car business that I don't know that will ever be engineered away from us," Herod said.

Fifteen percent of respondents said payment quotes usually draw from "basic calculator tools," eLEND said. Just 4 percent said quotes typically are based on "actual lender decisions."

Perhaps not surprisingly, the majority of initial digital retail quotes are wrong, the study found. Only 4 percent of respondents said such quotes match the final lender decision more than 50 percent of the time. Seventy-six percent said initial online quotes are accurate less than 25 percent of the time.

Eighty-six percent of respondents said inaccurate online payment quotes hurt the car-buying experience, while 94 percent said integrating proprietary lender models with what eLEND calls "pre-desking technology" would produce a better shopping experience.

Yet the majority of those polled also said "penny-perfect payments" could exist without the lender, with 57 percent defining the term as inclusive of government taxes and fees. Only 43 percent said a penny-perfect amount also would need to incorporate customers prequalified to a lender decision.

"The challenges to today's digital finance are solvable, but only when our industry is willing to come together to change processes, increase transparency and embrace tools that enable sales and finance to begin together at the start of the transaction," MacInnis said in a statement. "The last time we saw disruption around lender-dealer communication was twenty years ago when the big players got together to eliminate the 'faxes' between dealers and lenders. The results of this survey clearly demonstrate that the time for collaboration on the next big disruption has come."