It’s no secret that the mobile app market has exploded. Roughly 224 million people use mobile apps on a monthly basis, compared to 221 million desktop users, according to mobile analytics firm Flurry. Advertisers are eagerly reaching out to this growing audience. Enter Mobilewalla. The three-year-old startup is betting that the demand for targeting ads across mobile apps is only getting started.
AdExchanger spoke with Mobilewalla founder and CEO Anindya Datta.
AdExchanger: When Mobilewalla started out you were in the app discovery business, but you’ve moved away from that. What led you to pivot, and what does Mobilewalla do now?
ANINDYA DATTA: At its inception, Mobilewalla quickly became one of the most used independent app search and discovery engines in the marketplace. But when we looked at our business strategy for MW, we felt that the massive amount of data collected from the native app stores (which were being analyzed to support semantic discovery) could yield analytics that had greater business value than the discovery service itself. Hence the pivot.
What we have now is an analytics platform. We collect vast amounts of data regarding apps. When I say data, we don’t have an API or handle that developers put inside their apps to collect data. We actually collect from the market. So for instance, we get every piece of information that you could get about apps from the app stores like iTunes and Google Play, as well as from Twitter, YouTube and Facebook.
How do you come up with the analytics for measuring app audiences?
First, I should mention that for a standard audience measurement in most media segments like print, radio, TV, etc., that typically involves extracting information from a small group that you can apply to larger populations. For instance, if the show The X-Factor is popular today, you expect it to be popular 90 days from now. Basically popularity persists, and this is what makes panels work. We have a ton of data that shows if you were to take a snapshot of the top 20 TV shows today and again in 30 days, the churn is very minimal.
In the case of apps, if you take the top 20 apps, and if you look at the top 20 apps a month from now, the churn is nearly 50%, and in 90 days the churn is nearly 90%. Sustained popularity is what’s missing in the case of most apps. We recognize that, and so we built this big data method of doing audience measurements. What we do is we compute lookalikes. We break the huge universe of apps, which is about 1.3 million right now, into hundreds of clusters of apps that have the same audience.
From each of these lookalike clusters, we select one or two reference apps, and these are apps whose demographics we know because they are part of our network and it’s disclosed to us, or we can find it. And on top of that we overlay a bunch of signals to create audiences.
How large would you say the mobile ad market is?
The conservative view of the mobile ad market in 2012 is that it’s about $2 billion. In general, about 75% to 80% of that was in-app ads, $1.5 billion or so. That’s small compared to display, but it’s growing really fast. Some of the clients we work with – these are large ad network clients – in the last year have gone from 20 billion impressions a month to 100 billion impressions a month.
At least on the impression side, on pure volume, it’s growing very fast. But as you know, the market side increase is not just a function of volume but the price of the CPMs as well. CPMs have simply not gone up very much. When you have the ad CPMs start to go up, you’ll see this super-linear increase, but we’re not quite there yet. There are various reasons why ad CPMs are low. For one thing, mobile ad creatives are not very good, or people haven’t come up with great ways to show apps yet. In time, those problems should be worked out.
Who do you see as your competitors?
If you think of us as an app analytics company, there are competitors, but if you think of us as an app audience company, we really have no competitors. We do not know of anyone who has a realistic basis of providing an app audience. The only company we’ve run into is called Onavo, which claims to have supplied audiences. Other than that, I don’t know anyone.
What can you tell us about your revenue and growth?
Mobilewalla has raised over $7 million in equity funding to date, including a recent $4.2 million Series A round led by Madrona. We have 30 employees, and Mobilewalla is based in Seattle with research operations in both Singapore and India. Our clients include well-known mobile ad networks [like] InMobli and NativeX.
What are you working on next?
From a publisher product perspective, we used to have these MW analytics, where over 7,000 apps from over 1,000 publishers had signed up for tracking capabilities. We recently took it down because we are overhauling it completely. Over the next 180 days, we are going to release in batches our revamped portal, which will let you do detailed competitive tracking, and a whole bunch of features that we’re going to roll out.
At some point in time, we are also thinking of building technology that will give publishers the complete lowdown on the ads that are being shown on their apps. They’ll be able to see which ads are effective at which location. For example, we’d be able to say coupons on movie tickets in LA between 4pm and 7pm on a Thursday are very popular. Technology like this exists in display, and we’re thinking of doing the same for mobile, using the publishers that are hooked up to us.
On the ad tech side, we are trying to increase the efficiency with which we serve our data. For example, if an ad exchange is about to serve an impression, it would come to us and say, “Here’s the impression I’m going to serve, give me the demog [demographic] info,” and the whole thing will happen in 20 to 30 milliseconds. We’re creating a large infrastructure where such queries could happen more efficiently. We are continuously adding new data items. For example, all our clients send us data impressions, and we’re creating maps of ad effectiveness based on location, type of device, etc. We plan to significantly augment our data feeds using this kind of data as well.
