on Mar 2 2016
We’ve all wanted technology to tell us exactly where to park for a long time now. But the challenge has always been obtaining real-time data on available parking.
I’m a data scientist. I received my PhD from Stanford in artificial intelligence and then became a professor at the University of Illinois working on both AI and machine learning. While in Illinois I realized that while others have tried and failed to solve this problem for 15 years, they only ever approached the issue from one angle. For Parknav we chose to use data from as many sources as possible. We source data from our own users, other drivers, navigation companies, and telecom companies then analyze it via our custom algorithm. What we do is essentially connect many more dots than have ever been connected and spit out a highly accurate probability of parking availability.
Some friends often ask me: “Why don’t you solve cancer? Why parking?” The truth is that I really love parking. I do it everyday, and I wanted to solve a problem that touches everyone all the time. But it’s merely a jumping off point to other problems that can be solved with the technology we’re building.
When we have enough data we are accurate 85% to 95% of the time. In the worst data locations we can only get 70% accuracy. On average, we can predict whether or not there is parking with 80% accuracy anywhere in the city. The difference is really our ability to scale; we can easily cover every street in a new city 24 hours a day 7 days a week.
First, we’re an app that consumers, or regular drivers, use to find parking spots around their homes. But the real revenue drivers are automotive manufacturers, navigation companies and service businesses like the telephone and cable companies. Companies such as navigation and automobile manufacturers take our data and embed it into their systems for consumers. While service companies (known as fleets) like Comcast and AT&T have thousands of cars searching for parking all day that can save hours for employees with Parknav.
The supply side of the business consists of navigation and telecom companies that provide the data we need for our algorithms. These industries both track all sorts of moving dots around the globe like trucks, cars, and people. Often the data is just a moving pin with little more information than location, but many of the dots correspond to parked vehicles (both illegally and legally parked). We aggregate as many sources as possible for an accurate reading on where all the cars and trucks are at any given moment.
We started with some luck – a child company of Deutsche Telecom invested in Parknav and connected us to some of the world’s most trusted data suppliers. Intros to the right people at the right time gave us the perfect start and enabled us to deals without being taken advantage of as the new guy. We’ll continue to acquire data suppliers but already have enough numbers to serve all of the US and Germany with 85% accuracy.
The consumer app is a great way for us to test the product and hone our data analysis. We have an app for both Android and iPhone that are still pretty clunky but already have 100,000 users in just two cities with another 18 cities just launched. Our focus is almost exclusively on the B2B business because of the larger revenue, data access, and distribution opportunity.
Our initial customers took issue with us giving away Parknav free via the app, so we were slow to launch. They weren’t sure anyone would pay extra for a service they could find free on their phones. They’ve since realized that consumers are happy to pay for the extra convenience of parking navigation built into their cars.
Our ability to stay ahead of the competition depends entirely on the quality of our data. We collect data from more sources than anyone else in the business and have a team of excellent data scientists which are exceedingly difficult to find these days and even more expensive.
Companies like Google could allocate their data science teams to problems like this, and they might, but there are only a couple of companies in the world with talent like we have. To date everyone who has tried to solve parking has failed, while we’ve succeeded. That is our biggest advantage.
All of our competitors are also potential customers. Many of them, like navigation companies, have the right data access but not the right people to build the algorithms necessary to make sense of it all. Large companies: Inrix, HERE, TomTom, Google – these companies show desire to duplicate our product, and have access to data that could presumably be used to create our product. So far they've tried and failed. Medium companies: Paybyphone, IPS – have meter-payment data that could be used to create a limited version of our product. What they have so far is a significantly inferior product.
We have an amazing group of individuals working on Parknav. I was a Stanford PhD and have 20 years of experience in data science including ten years as a professor at the University of Illinois. My cofounder runs all development for Parknav. I’ll never understand how he can work so quickly. Gerhard has many years of business development experience with Siemens and his own companies, while Jeremy is an MBA from UI who does all our product and marketing. We also have two other data scientists that I hired from my classrooms at the University.
Luckily both Sergei and I are experienced data scientists which has helped attract great talent, but we are definitely in the market for another senior scientist to round out the team.
Unfortunately, I can’t mention any official partnerships yet, but we have a number of excellent clients in both the OEM and real estate space. When we got started three years ago the sales cycle was years long, but as the industry wakes up to what we're doing, they’re ready to buy. Now once we send out a quote, it’s a five month cycle and we expect to close many deals in late 2015 and early 2016.
First, we had to prove our technology with proof of concepts in smaller markets with smaller customers. We started in Chicago focused on consumers before realizing the automotive OEM space was a much more lucrative opportunity. Then it was a matter of proving our technology with smaller fleets and consumers; we measured adoption and monetization and then shifted focus to work with the largest manufacturers who need coverage for the entire country. It took us some time to build the data coverage for both the U.S. and Germany but now we can show the largest OEMs proof of concepts that actually cover entire nations and get them really excited.
We have a number of large deals with automobile manufacturers, fleet companies, software companies, and real estate listing sites either closed or almost closed. As soon as our partners are ready for a press release we can make these more public, but by February we expect to have our first two customers completely on board with another five set to launch a few months later. Our average deal is $1M to $5M on three to five year contracts for each OEM.
There was a time when GPS was a nice-to-have in cars and now most people can’t drive without it. In five to ten years Parknav will be just as ubiquitous. It seems obvious, why not have your car just steer you to the spot nearest your front door, or the restaurant, or the baseball game? We just haven’t had the technology until now, and it will take several years to integrate with huge automobile companies.
Parking is just the entry point to the market and an initial revenue stream. We need to lock in all the data we possibly can to get ahead of the competition which we’ll soon mine for additional intelligence. With our data set, not only will we know where to park, we’ll know how people are moving, where crowds are forming, where Starbucks should open their next location, etc. We’ll be an unprecedented source of data on everything the populous does and there are numerous ways to monetize that.
There are many types of customers that could integrate smart parking in the future including real estate and other consumer apps such as Yelp and Groupon. Real estate companies can quickly show their potential buyers all the parking around a given property in real-time. Yelp would work well with Parknav: imagine if your dining recommendations came with live parking data?
Also proximity deals are a possibility for us. Our app will know where you’re parking and be able to feed nearby deals from the Groupons of the world. When parking does run out in a given area, the app will be able to surface deals from local parking garages.
Real estate companies use Parknav to showcase how much parking is available nearby a given property listing. Agents pay us to use the software and then sign up all their potential buyers as Parknav users. They basically pay us to find customers for Parknav.
We face all sorts of problems: data problems, traditional startup problems, employee engagement and interest, etc. We’re a small fish in a huge industry and getting the attention of the right clients has been one of the hardest challenges. Figuring out how to sell to larger corporations, how enterprise sales relationships work and evolve, and how to leverage what little advantage we have has been tough.
The two biggest challenges that will remain are building the best dataset and technology while simultaneously convincing large OEMs that what we’re doing is inevitable. We’ve done both well so far, but the industry is ever-evolving and nobody really knows how cars will navigate in the future. We need to remain in the lead, continue to innovate the technology and find industry leading partners to work with us.
Already have an account? Login
Don't have an account? Signup