Data specialists help rental start-ups spot demand

Young companies such as RelayRides, which lets owners rent out their vehicles for hours at a time, and Airbnb, an online rental-lodging service, rely on a delicate balance of supply and demand to be successful — not an easy challenge for firms with far-flung national aspirations.

As a result, a new crop of specialists and businesses is helping firms manage what some are calling the “sharing economy.”

One such company is Domo, an Internet-based “cloud” business intelligence service. It was hired by RelayRides to analyze how and where owners and renters request service.

RelayRides operates in cities across the United States, including Boston, Atlanta, Los Angeles and Washington. Owners upload their car’s profile and a fee, and renters request the car, subject to the owner’s approval.

“We’re trying to build an online marketplace,” said Andrew Mok, RelayRides’ director of business intelligence. “There’s always a bit of the chicken-and-egg problem — do you add cars first or renters?”

By using Domo’s analytics to track where renters search for cars and where cars are available, RelayRides tailors its marketing to the side of the equation that needs help, he said.

“If we’re seeing there’s a lot of search volume in Phoenix, in combination with the fact that the approval rate is low, it means that area is supply-constrained,” he said.

Before its contract with the Boston-based Domo, RelayRides had a full-time employee manually gathering data, giving the company less time to react to such imbalances, Mok said.

RelayRides also learned that processes matter. For instance, Mok noticed owners abandoning the online registration process when asked to upload a photo of their car, because they did not have one on hand. The drop-off came after they had already typed in other important details, such as license plate number, make and features of the car. So RelayRides implemented frequent reminders, prompting car owners to finish their profiles. That change doubled the site’s conversion rate of registrants to actual users, Mok said.

San Francisco-based Airbnb, letting hosts worldwide rent out their homes, rooms, and other forms of lodging for a few nights at a time, collects similar data internally. Several years ago, for instance, Airbnb’s analytics head Riley Newman noticed a spike in searches for Lyon, France, in December, corresponding with a French festival of lights.

“None of us had heard of the ‘Fete des Lumieres’ but the surge of demand signaled a growth opportunity,” Newman wrote in an email. “So the next year we knew to prepare for this and were able to forecast the volume of supply needed to meet demand relative to general growth of the business. This is one example of many; because we’re a travel business, seasonality is a very strong pattern that enables us to stay ahead of the growth curve.”

During the past year, Airbnb’s data team has tripled in size, reaching about 15 people, hiring from LinkedIn, Google and President Barack Obama’s campaign data science team, according to the company. The team is now building a price recommendation algorithm, notifying hosts when their listing is priced significantly above or below similar properties in the market.

Lyft, a San Francisco-based mobile app that lets users request rides from drivers willing to chauffeur them for a fee, also collects data internally, but not yet in real-time, founder John Zimmer said.

Lyft operates in Washington, Chicago, Seattle and Boston, among other cities. Drivers are asked to place pink, fluffy mustaches on the car to identify themselves to potential riders.

“On the supply side, we see where drivers are available, and can look at their locations. We can match that against demand, which can be seen from app-opens to ride requests and ride completion. We’re constantly looking at that, predicting the future of both those [supply and demand] curves,” Zimmer said. “Because we’re constrained by our team’s time, the time we spend as a team will be focused on one side or the other.”

Lyft determines whether to run marketing campaigns targeting drivers or riders based on data from the previous week, but will soon transition to real-time data collection.

“Without insight into the right data, we could be spending money [on] the wrong side of the equation, putting in marketing dollars into supply when you actually have a demand problem. The amount this data saves is in the hundreds of thousands of dollars, and soon probably millions of dollars that will be more efficient,” Zimmer said.

Once it starts collecting real-time data, Lyft will be able to assess factors such as weather, which could immediately affect demand.

“If it starts raining, there’s this surge in demand. We could find a way to let the driving community — the supply out there — there’s an increase in demand. If they happen to be out, there’s a good opportunity.”