May 1, 2013
Q&A, Di-Ann Eisnor, VP of platform and partnerships, Waze
Waze’s Di-Ann Eisnor was among the experts we spoke with for our recent trend report “13 Mobile Trends for 2013 and Beyond,” which is based around insights gleaned at the GSMA’s Mobile World Congress in February. Waze, a “community-based traffic and navigation app,” won Best Overall Mobile App at the Global Mobile Awards held at the Congress. The company claims 40 million-plus users around the world and, after launching an ad platform in late 2012, counts Target, Taco Bell, Walmart, Ramada Hotels and Dunkin’ Donuts among its advertising partners. As Eisnor explained, the company is using its data on “how people move through the world and what their patterns are” to help brands fine-tune messaging to consumers on the go.
Can you give us a quick overview of Waze?
We are a social GPS and real-time traffic—and that means that we have this group of 40 million drivers, or more now, and they are anonymously giving us GPS traces and timestamps just by turning their phone on to get their free voice-guided turn-by-turn navigation and their traffic information. And then we use that to try to make sure that everybody saves as much time as possible every day. Our users are unique in that they’re not just using navigation for when they don’t want to get lost; they’re commuters for the most part. And so they’re using it with great frequency—on average I think it’s now seven and a half hours per month. It’s a fairly high frequency use.
We’re trying to do something fairly new that takes advantage of everything we’ve all learned about location and about navigation. We’re kind of this companion that gets people to where they want to go at different parts of their day—we know if they’re going from home to work and what time of day that is. We know if they’re going from work to the airport or from work to their kids’ school or to the daycare or to the soccer field or to the gym, because they’re doing those searches as soon as they get into the car.
For the advertisers, we call this a location-guided advertising platform. Often when you hear about location advertising and mobile advertising, a lot of people really understand the “what’s near you right now”—like, if there’s a special nearby where I happen to be. But it doesn’t take into account my context or what’s going on with me. What we’re trying to do is weave in, “Where was I? Where am I now? Where am I going?” And to use that information not necessarily as a person-by-person target, but to understand these patterns of insights about how people move through their day and at what time of day different offers or advertisers are relevant.
Can you give an example of that location-guided advertising in action?
We work frequently with Dunkin’ Donuts, and obviously people are doing searches for coffee at a certain time of day and it’s usually when they’re on their way to work. On their way to work is also when they’re open to stopping off at a Dunkin’ Donuts. If you’re doing a search for coffee, Dunkin’ Donuts will come up as the sponsored option. So there is a search component. Then there are pins on the map at the locations along the route or near you. We give you an opportunity from there to reroute to the destination.
If you look at the whole Waze drive experience, we have different advertisers working across different parts of it. When you open the application, if we have an offer that is incredibly good for Wazers, you’ll see something at the beginning of the experience, before you do the search. Then there’s, while you’re doing the search, what results are coming up and how do we highlight advertisers that are part of the community. Then there are the pins on the map. But then there’s, each time you stop—and again, this has to be a relevant offer for users—we have what’s called a “takeover,” and it only happens at zero speed. We’ll put your offer in front of people and give them an opportunity to navigate.
So we’re tracking that whole funnel from impression to click to navigate, and then all kinds of other secondary offers too. Say I don’t want to go to Taco Bell this instant, but I want to save the offer for later or save the information they’re giving me, then that’s automatically saved into your inbox, and you’ll get an email reminding you that you have an offer saved there.
So it’s that whole part of the process, all the way to actually capturing who’s arrived at your location, and that’s pretty unique. As an industry, we’ve wanted to be there for a long time. And I’m excited about the different bits of context we are starting to bring in.
Any other examples of how you’re working with brands?
We have a really fun example with Hollywood Records, when they gave away a free song to Wazers this past Valentine’s Day. That was that takeover scenario: If you’re going at zero speed, then you have an opportunity to download the free song.
Nintendo started a program for the holiday season, and they were promoting a bundle of a game system and some games. They have different bundles that they do with the different retailers, so it was a combined effort between Nintendo and Walmart and then Nintendo and Target, where we could promote the specific bundle and then watch people actually navigate there.
There’s one quick-serve restaurant who considers themselves both a coffee company and restaurant. And we can do an analysis of search behavior at different times of day by different cities, even by the different locations. So let’s just say in San Francisco people are searching for coffee from, let’s say, 8 a.m. until 9:30, but in Chicago, it could be from 7:30 to 9. We have a unique ability to target by that time of day and by that destination to make sure you’re putting your correct offer in front of people at the right time.
That is something we’re developing for this one company right now, and then we’ll be able to roll it out to everybody. The dayparting already exists, but to be able to understand what time of day we serve it for every single location across the country is the part we’re weaving in right now for that company. We call it the temporal interest graph. By May you’ll see that out.
Is there any other context you are taking into account?
The other context that’s related to mobile and specifically to location-guided in this way is weather. So that’s something you’re going to see from us. It’s incredibly important when you’re hitting the road because not only does it change your time to your destination, but there are different things that are going to be relevant to you when it’s raining or when it’s icy than when not. So extending that to advertisers, whether it’s allergy medicine or tires or even take a retailer like Home Depot, whose offers will need to change based on what the weather is outside.
At MWC you talked about how the mindset among marketers has been changing quickly. How do you see it evolving right now?
Most of the companies we’ve worked with so far, we’ll get a call from the VP of marketing and they’ll say, “I’m a Wazer. How can I put my brand on here?” So it’s not that we’re at a massive scale yet. But when we were having these conversations when we launched our ad platform in November, there were very small test budgets. And I think those early customers saw the value so fast that they just kept allocating more and more. We launched with somebody in December, and they’ve already allocated a budget for us that’s their No. 2 spend in all of digital for the year. So that was a big surprise.
An experimental budget was literally $5,000, and then, December or January, it was $50,000, and now many of the conversations we’re having around experimental budgets are more like $500,000.
Have you had any pushback from users who are now seeing more ads?
We’ve had some pushback. I think anytime you’re bringing in ads for the first time, you’re going to get people saying, “Oh, I don’t want ads.” But when it comes down to it, I think people are glad they have this service for free, and most of the sentiment when we reach out to people is fairly positive to neutral. I was surprised at how much positive sentiment. On almost every program we’ve launched, I find great things on Twitter about how happy people were or how they bought something or how it’s part of the map.
Is there any personalization based around individual users or is messaging based more on general context?
Well, it’s personal to individual users in that it will be [tailored] to the end destination. So if something called “work” is the end destination and something called “home” is the beginning destination, we take that into account. If something called “airport” is the end destination, we’ll take that into account, or something called “Walmart,” for example. The destination and the origin are really important to advertisers. So we’re focused on that. And that would be by the individual drive but not the individual user.
One thing we want to do is to have categories of drivers based on the behavior. For example, many people have asked us to identify if there’s a category of driving behavior called “frequent business traveler” based on number of trips to an airport or number of cities where you’ve opened the application. And you can tell maybe a specific kind of professional if the commute time is, say, before 7:30 a.m. There are different driving behaviors based on the time of day, based on frequency of driving, based on the number of different places, the consistency of the route. We’re working through that now.
But right now we’re not doing too much that’s related to a specific user. We want to walk that line very carefully and together with our users and with our advertisers. But what we’re doing so far is just gathering insights about the way our users are using the system and where they’re stopping, and it’s pretty incredible. For example, the number of people that still stop at a bank on the way to or from work is much higher than we would have imagined.
We’re trying to figure out, what does the work commute look like, because it’s not just a here to there. It’s a there to drop off the kids, to pick up your coffee. It’s all these different pit stops. And on the way home, it’s much more flexible. So depending on the day, it’s either, I’m stopping for groceries or I’m going to a restaurant to meet friends or I’m going to a friend’s house or I’m picking up the kids or I’m going to the gym—there’s all these different things that constitute this notion of “my drive back home” at the end of the day. We’re learning a lot about that and sharing that with our advertisers and using that to help make sure that we’re connecting brands and users in the right way.
How do you envision Waze evolving as you get a more sophisticated sense of your consumer base and more advertisers?
I’ve touched on a lot that has to do with the combination of time, of origin and of destination, and then weather is one layer on top of that. We’re not very far away from being able to serve up those messages according to the optimal time frame in each location for each group of drivers. One of the things I think is exciting is that mobile in general is going to take a much, much bigger share of how advertisers understand how to communicate with users. Right now it’s all so new. I think we’ll be broken out of the category of “experiment” as a group in mobile.
My vision here is that the way Google was for purchase intent for the Web, we like to imagine that we are becoming purchase intent for the real world. You need all these different pieces of context, and each drive is any number of given searches. That’s kind of where we’re headed.
We have platforms out there who can actually tell you your impact on sales. We’re working with some brands right now on, if there’s an impact on turn and velocity. We’ve always said we wanted this information, but now we might be buying against it as a combination of brands and performance. And performance is no longer possibly a proxy for just a search or a click; it might actually be dollars spent. And so this is a new and kind of scary area that we don’t know how to charge for—we don’t know what it all means, but it’s where we’ve wanted to be.
A lot of advertising has taken us away from what we wanted, which was to impact sales and build our relationship with the customer. And now these platforms allow you to actually get back to, how are you both going to build that relationship and that engagement, but actually know how it impacts sales?
So you know exactly whether that person went to the store, how much they spent, what they bought?
It’s going to be at least until next year to really tie in that “how much they spent” piece of it, but the arrival piece we’re already tying together now. So yes, if you click on the “navigate” button [after seeing a takeover ad pointing to a retail location], we can actually tell if you made it to the destination.
How do you see this kind of targeting evolving?
We talked a lot about the trend of context and how it really needs to include the origin and destination—not just the now, but the before and the after. The technology and the algorithms that are needed to pull that off are going to advance significantly; targeting is going to look very different. We’re going to be able to target based on context, but it will also automated, the way we’re used to now as we’re targeting 18 to 24. So I think it will be all about executing very, very well on those systems.
We currently have these different systems where you couldn’t think of customer acquisition in the same way that you can think of loyalty in the same way that you could think of basket size, for example. And some of these new platforms are allowing you to have a relationship across all of those, because when you talk about location and you talk about patterns, then you’re able to get information all across. You’ll be able to weave together better a customer from acquisition to basket size or to lifetime value even.
Do you have a hypothetical example of how that would play out?
It can play out in many different ways, but if you look at the data, even on an anonymized level, you can tell when, say, this device ID used to be seen at Starbucks, now it’s seen at Dunkin’ Donuts. It’s been seen there 12 times this month. The dwell time was seven and a half minutes. And most people have a good sense of, if someone stays there X minutes, how much was that sale. And then through offers and things like that, we can track specific redemption values and dollars.
We’re building a picture of how people move through the world and what their patterns are. And pulling all of that information together is going to be very, very exciting, all the way from to the user to the advertiser, and put us all on the same team.