The Secret Behind our Inventory Suggestion Algorithm
March 10, 2022
Insights
-
3
 min read
Tane Jillings
Wholesale Automotive Expert

We came from technology backgrounds where there was an abundance of data, however the ability to use that data was severely limited due to PII (personally identifiable information) privacy regulations. One of the attractions of the auto industry was again that abundance of data, however there was no personally identifiable information attached - the products that we all sell are hunks of metal with, that include a unique 17-digit identifier in the form of a VIN. VIN level data is a game changer!

3 years ago, we decided we wanted to better understand the retail market in the US. This was driven by erratic buying behavior from our dealer partners as we saw demand come in ebbs and flows. One week, they had no appetite to buy vehicles. The next week, they wanted everything they could put their hands on. We had one particular encounter when a dealer said, “I am completely out of F150s, I need every one you have available”. That statement got us thinking - shouldn’t a dealer know when he’s about to run out of their best-selling product? And shouldn’t WE know when our customers are overweight or underweight on particular vehicles?

With that question in mind, we set out to find the answer. We took some of our brightest engineers and created a data science team. From there, we found every retail listing going back to 2018. Our data team started building models focused on identifying the best opportunities for dealers to purchase vehicles with the goal of making customers more successful and more efficient with their time. 

The product we built was an inventory suggestion engine called PIE - our Predictive Inventory Engine. PIE looks at any given dealer, analyzes what has happened for that dealer over the past month, past year, understanding seasonality, how do they compare versus their competitors, days supply (the most common metric used today), and AI driven forecasts of what we think is going to happen over the next 30/60/90 days. Those forecasts are looking at macro factors as well as dealer specific data points to produce a score for every vehicle we have in inventory. The concept is simple, the higher the score, the more likely a vehicle is to move quickly for a dealer. Scores are from 0-99, but to make the scores more intuitive, we have adopted baseball terminology.

* Home Run (81+)

* Triple (61-80)

* Double (41-60)

* Single (21-40)

* Strikeout (0-20)

Pretty simple concept. You want more home runs and triples, and less strikeouts! Out data shows that home runs are 5x more likely to sell quickly compared to strikeouts, so dealers that follow our data driven recommendations are turning vehicles over quicker than vehicles they source elsewhere.

We are excited about future opportunities to not only optimize inventory for sales velocity, but also for maximizing profit for our dealer partners. if you would be interested in seeing what data insights we can give you, or if you just like data as much as we do, please reach out! Our team would love to talk.

Ready to move vehicles faster, smarter, and easier?