Deniz had been saving for years. He finally pulled the trigger on his dream car—a white 2020 BMW 520. Deniz kept his beloved BMW clean, was a safe driver, and bought an insurance policy from Allianz. Unfortunately, that didn’t stop a careless driver from rear-ending him one Monday morning on his way to work..
Now, Deniz needs to find the right repair shop—and this is where things get interesting. Where he takes his prized 520 matters. The shops might be:
Whichever shop he chooses, Allianz agents will look at similarly related damage data—the brand, the model, the part(s), the work (paint, bumper, or dent repair)—and compare it with similar types of services and the commensurate charges before making an estimate.
Cost optimization is a data-driven technique that helps companies determine the optimal price point for their products or services.
In the insurance industry, this involves analyzing various factors such as a customer’s risk profile, historical claims data, and market trends to determine the appropriate premium to charge for a particular policy—while maintaining adequate coverage for the insured individual.
With auto insurance—cost optimization often focuses on estimating damage repair and replacement costs—especially anomaly detection.
To get the most accurate estimates and comparisons—Allianz knew they needed a machine learning solution. The insurance giant tapped Aigoritma to deliver more accurate estimates with the power of machine learning.
For example, the historical data for the cost of repairs similar to Deniz’s bumper might show consistent pricing: $1,500, $1,550, $1,500, $1,450, $1.600, $1.500, etc. If the estimate for Deniz comes in at $1,800 or as high as $1.900—then Aigoritma will alert an Allianz expert that the estimate is too high and that they should bargain down to a lower price—thus avoiding claims leakage.
Claims leakage refers to the difference between the actual cost of claims and the amount that an insurer expected to pay for those claims. High variance in unit labor rates and other factors can contribute to higher claims leakage because it can lead to inconsistencies in the amount paid out for particular types of claims.
An Insurance thought leadership study notes that the industry benchmark for claims leakage is 2-4%, but stated the de facto number is actually much higher–and more likely to be hovering around 20-30%.
Service stations can ask for any price they want—but if Allianz has accurate data—it’s much easier to get a fair, optimized price. The result? Allianz has been able to save 1–1.5%—huge savings when you consider the volume of claims the insurance company handles per year.
Allianz usually estimates nine different invoice items and uses these features to make accurate estimates:
These costs comprise both parts and labor costs. Labor costs can be especially tricky for two reasons. The cost of auto repair labor has been dramatically increasing and the continued high variance in unit labor rates—which can lead to higher claims leakage. These problems are not specific to Turkey.
In the U.S., motor vehicle repair prices have jumped a staggering 23% from 2022 to 2023, according to U.S. government data. In addition, labor costs have grown so high in the EU, that residents of Germany, Austria, Belgium, and France are now taking their vehicles to get repaired in Türkiye.
Within Türkiye, however, the cost of labor and spare parts can be higher in big cities—such as Ankara, Istanbul, and Izmir—than in rural areas. For example, the average auto mechanic salary is 17% higher in Istanbul than the average auto mechanic salary in the rest of Türkiye.
By helping Allianz better pinpoint the estimated labor costs and necessary amounts of the invoiced items that field agents require—Allianz is able to save 2–3% per annum.
Would you like to see how Aigoritma can help your organization solve its own real-world problems with a robust and ready-to-use End-to-End AI platform that doesn’t go off budget? Our platform, ML Studio, provides ready-to-use templates for various business problems, including cost optimization, fraud detection, predictive maintenance, sales prediction, and much more.
Get in touch for a personal demo.
Copyright © 2023 All Rights Reserved
