Methodology

The dataset comprises of 1,161,264 vehicle sales transactions, identified by unique combination of VIN and transaction date. The transactions cover the period from APR-2022 to MAR-2025, with an average 32 thousand transactions per month.

The sales transactions are as reported by 147 distinct dealership groups with between 1 and 42 stores, with the median group reporting 3,502 transactions over the 3 years period. As the dealership level variation is a main source of heterogeneity in the data, the data have been imputed using ARIMA models to create  an additional synthetic 50,000 transactions to account for some dealerships not reporting in some months.

The data comes from the Authenticom feed for the dealership groups, with the exception of a large dealership group in the Northeast that has provided a separate file with sales data.   

The dataset covers at least 10,000 transactions for 24 distinct vehicle brands, with the top 5 of FORD, TOYOTA, HONDA, KIA and CHEVROLET covering 42% of all sales.

The sales are nationally representative with 90% coming from 23 geographically diverse states with more than 10,000 sales each.

Top 10 Dealership Groups by Sales Volume includes only High-Polly-Engagement dealerships (at least 50 quotes and quoting at least 2% of transactions) over the past 12 months.

Hard-to-Insure represents the share of car buyers who requested a quote from Polly, and either did not find a single carrier to insure them, or all quotes received where for $800 a month or higher.