Presented by ACV Auctions
In the used automotive digital marketplace, transparent and accurate vehicle information is critical. AI-enabled vehicle inspection has become a powerful tool to quickly deliver on-target vehicle pricing information, so dealers and commercial partners can buy, sell and value vehicles reliably and efficiently. And the condition report is central to all those transactions.
“Understanding the condition of the vehicle is the biggest driver or factor in determining profitability of the transaction,” says Vikas Mehta, COO of ACV. “The ability for a dealer to turn a profit depends on their ability to assess reconditioning costs, and therefore the condition of the vehicle they’re considering purchasing.”
Providing condition reports that guarantee that kind of accuracy requires a combination of up-to-date AI-powered tech and the skilled condition inspectors who visit dealerships to ensure they’re keeping their fingers on the pulse of the market. It also requires a continuous learning mindset, based both on empirical feedback and what turns up in the data, and a world-class intelligent inspection platform.
ACV’s answer to the market need is the Single Inspection App (SIA), which enables inspectors to capture vehicle intelligence in one organized place. Every inspection and every sale helps level up the platform, as they learn more about a car’s make and model, leverage an inspector’s knowledge and expertise, and gain feedback from buyers. The SIA platform represents the collaboration of people and innovative technology, a company-wide initiative, from the field to departments across ACV.
The evolution of SIA
SIA is the result of consolidating three inspection applications into one powerful platform. ACV’s in-house platform was focused on targeting the wholesale marketplace. To extend their reach and abilities, the company soon acquired two new platforms, one a retail inspection company, and the second performing off-lease return inspections. Each platform was strong in its own way, Mehta says, but combining those platforms unlocked the learnings and the best practices from, effectively, three extremely powerful use cases.
A single, powerful platform minimizes maintenance, and improves consistency and training. But the biggest goal in building the inspection platform was not to solve 2022 and 2023 needs, but to build for the future: an all-encompassing data platform that could be configurable and templated to inspect a far broader category of moving vehicles through relatively easy configuration changes. Right now, SIA allows them to localize the platform, in order to expand beyond an English-speaking market, and customize the inspection type, to go beyond retail, wholesale, and off-lease use cases.
“And most importantly, we’ve built with a strong data foundation, knowing that the future that we’re headed for is one that leverages insight and data in every part of the inspection,” Mehta says.
How AI and AMP bridges the digital gap
Artificial intelligence has become the single most powerful tool for online vehicle inspection, and data captured by tools like the audio motor profile (AMP), which lets a buyer listen to the sound of a car’s engine, bridges the digital gap between their marketplace and physical auctions. AMP has become a powerful part of the SIA platform.
“We found that customers were interested in hearing how the engine runs, so that not only could they feel a connection with the inventory, but they could do their own assessment of condition on that inventory, which is part and parcel of the physical auction world,” says Mike Pokora, ACV’s senior director of R&D.
Each AMP audio file is unique to the vehicle, and attached to all of the other metadata within the SIA inspection, including the VIN, the size of the engine, drivetrain or transmission information, and other build characteristics, such as the odometer specs that are captured as part of the SIA inspection process, and more.
All of that gets appended to the specific waveform file — and there are two million audio files strong now. ACV uses that data, across many different makes, models, years, powertrain combinations and mileage conditions, to train their machine learning algorithms.
How AI is leveling up damage inspection
One of the things that the company took an early interest in when developing SIA, was not just how to combine all these different inspection platforms, but also how to advance what they’re currently doing with their bread and butter market, which is wholesale. Similar to the AMP project, they collected massive amounts of data through millions of images of vehicle damage.
The next step was to figure out how to analyze all this data. With their acquisition of Monk SAS, an AI solutions company that automates vehicle damage detection, they can take photos of damage and, in near real time, provide feedback on the part that’s damaged, the type of damage and severity of the damage.
“This has massive implications for the field, to make inspectors more efficient and more uniform in terms of damage capture and collection,” says Tom Peters, the director of product management. “It offers quality control, provides real-time feedback and data is collected and goes back into this ML loop to train in the future. With Monk’s acquisition we’re able to feed more data, do more training.”
The implications of SIA for the digital auto industry
The objective for any of these investments in the inspection platform is, of course, to accurately assess the condition of a vehicle, and give wholesale buyers high confidence in the type of car they’re considering.
“It’s about buyer vibrancy and engagement,” Mehta says. “They’re all looking for vehicles. They’re all looking to find what they believe they need for their local market. There’s a number of platforms they can go to. By being consistent and highly accurate, the impact here would be continued buyer preference, buyer growth, and seller satisfaction, seller conversion coming from that. Those are broadly some of the things we measure.”
From a customer satisfaction standpoint, since the roll out of AMP, ACV has seen satisfaction metrics improve, Pokora adds. This includes a marked decrease in vehicles that come back for arbitration claims where there is a discrepancy between the vehicle a buyer received and what they thought they would receive.
“We have seen some pretty positive impacts from the implementation of ML models into our production system to support the auction listing, to support the inspectors, and to help the marketplace,” he says.
In terms of inspector efficiency, they’re also able to now measure how efficient the field is using the SIA platform at all the different points in the inspection life cycle, Peters says. With SIA, they’re collecting 60% more data than they were prior to SIA, even on the same inspection type. And they’re able to maintain that inspector efficiency through all this data, in making data-driven decisions and moving the product forward. It’s also able to provide, in the end, a more accurate value because the data is more structured. It’s had a good impact on training, too. Because the inspection is more guided in terms of consistency for the field, they’re able to train the field in a much faster time frame.
“We’re truly trying to solve some of the problems that have plagued this industry using technology and data,” Mehta says. “We’re building teams that leverage intelligence around what is perceived to be a rather opaque industry. The need for buyer engagement, buyer satisfaction, accuracy of inspection, trust — it’s just such an obvious space to invest in this technology.”
Sponsored articles are content produced by a company that is either paying for the post or has a business relationship with VentureBeat, and they’re always clearly marked. Content produced by our editorial team is never influenced by advertisers or sponsors in any way. For more information, contact [email protected].