Since it was founded in 2006, CarGurus has knocked down milestone after milestone.

The online automotive marketplace pioneered transparency into car shopping to help consumers identify great deals and overpriced vehicles alike. The company went public in 2017 at a $1.7 billion valuation, made its first acquisition in 2018, and hosts more visitors and listings than any other major online car market in the U.S. Now, COVID-19 has spurred new growth in automobile sales, meaning CarGurus is now busier than ever.

But the company’s solid position within its market poses a new challenge — once you’re on top, how do you stay there?

“CarGurus is going through a period that a lot of companies experience when they grow very fast,” said Jasper Rosenberg, who has been with the company for all of its 14 years. “You have this monolith-type approach because it lets you move very fast and at a certain point, to scale, you have to start splitting things out.”

Rosenberg leads engineering efforts for CarGurus Labs as its SVP of engineering. Launched at the beginning of this year, the team is charged with designing, building and testing new products for the company.

“It was really a recognition by the company that certain kinds of problems need to be tackled more like a startup,” said Lisa Iannucci, who is the VP of operations for Labs. “These projects are a little bit more experimental in nature, they have longer time horizons and less of a straight line to a solution.”

In an interview with Built In Boston, Rosenberg and Iannucci used one recent Labs project — a reimagined peer-to-peer selling experience — to highlight the team’s culture, outlook and processes.

Tell us more about why the company launched CarGurus Labs.

Rosenberg: This was very much a brainchild of our CEO Langley Steinert. He wanted to apply the model where folks can focus and iterate really quickly and make decisions really fast as well. 

Iannucci: CarGurus has a really impressive core business that we need to keep growing. So one of the things that Labs is trying to do is have those dedicated resources so that we can work on those big opportunities that might be big revenue opportunities two, three or four years down the line — but may not bring in new revenue for the business today. There’s always a trade-off between a dollar in today’s revenue versus a dollar of tomorrow’s revenue, and how you think about resource allocation there.

What’s an example of a project you’ve worked on with Labs?

Rosenberg: One area that Labs has spent a lot of time in is looking at the private sale space. Something like 13 million people sell their car privately every year — and nobody is doing a great job facilitating that. It’s very Wild West. We spent a lot of time trying to understand that space and figuring out how to help people do this online in a way that’s a lot more secure and less intimidating.

Every state is different. Every DMV operates differently and there are even different rules about what’s considered a bill of sale. Figuring out how to streamline that process was a really interesting challenge for the team, and we now have this product that looks very seamless to any user.

Iannucci: A good portion of those 13 million private sales are still done at the end of the driveway — it’s a market that has not seen a lot of innovation or modernization. So part of it is educating consumers that through CarGurus they can access the largest car shopping audience in the United States, just like dealers. They get real leads from real customers who are interested in buying their cars, just like dealers would.

Because of the way we coordinate our search results based on great deals first, the peer-to-peer sellers on our site actually have a chance to show up really well in the search rankings alongside dealers. CarGurus’ business model is that we have a few spots at the top for dealers on some of our paid packages — but otherwise everything is ranked organically based on the best deal first. So if you’re a P2P seller and you’re selling your car for a great deal, that car is going to get a lot of eyeballs, in the same way that a dealer would. It’s a really innovative take on the P2P selling experience versus some of the other options that are out there in the market.

Rosenberg: It does require a little user education because I think most P2P sellers are used to a Craigslist or a Facebook, where you list your car and it’s at the top for five minutes and then it gets pushed down and you have to try to force it back to the top. Here, it’s not a time-based thing — it’s a value thing.

What’s the engineering culture like on this team?

Rosenberg: CarGurus in general has a pretty flexible process — different teams do different things. At Labs, we probably operate closest to a Kanban approach where we’re pulling off what we think is the highest lever at the moment and people are contributing to that queue from all different levels, be it the individual engineers or product or design.

We push very frequently. And all across CarGurus but particularly in Labs, we do a ton of A/B testing. It’s one of those lessons from Langley from long ago — if you can fail fast and get that learning on the cheap and not invest a bunch of engineering just to learn something, you should definitely do that. That’s a very important part of the culture.
 

We’re breaking new ground, trying new things, and there’s definitely room to come in, experiment and shape it.”

P2P was built mostly on top of the core CarGurus stack. But we are in the process of decomposing what we view as the deployable website, and P2P was the first consumer-facing piece that was split out so that we can actually deploy it independently in a Docker container. It’s a React-based app sitting on top of a Tomcat Java base with Spring.

I think it’s worth saying that at Labs we are still part of CarGurus engineering as a whole. We’re not like a tech R&D group. We’re all collaborating across engineering and able to leverage all the same build infrastructure as the rest of the organization.

What else are you working on?

Iannucci: One of the products that’s really in the alpha stage right now is a product that’s helping dealers source inventory directly from consumers. There’s a lot involved in that process: the coordination between the dealer and the consumer, the logistics of making sure that the consumer gets to the right place at the right time, ensuring the transaction goes off seamlessly, and so on.

There are a lot of real-world operational elements there. We are talking to both dealers and consumers about what we want that flow and that experience to look like. We’ll be looking at how we can build that product and make it come to life in a way that’s efficient, scalable and a good experience for both the consumer and the dealer.

I think the cool thing here is there’s so much that’s left to be defined. We’re breaking new ground, trying new things, and there’s definitely room to come in, experiment and shape it.