We just launched Metaresolver this past week. I’ve held off on talking publicly about a lot of the decisions that went into the process, cause I was never sure when we would find some new chunk of info that would make me completely reverse my thinking. But when sitting down recently to talk through the process with my trusted friend and advisor Saad (thanks for everything Saad!) he pushed me to start blogging again about some of the things we’ve learned. Then I found myself writing a bunch of emails and discussing the same issues over lots of coffee with friends and colleagues. So it seems the time is right.
When talking with Seamus about the ideas that would become Metaresolver our hypothesis was that mobile advertising isn’t really delivering the value it should be able to given how much time folks spend on their phones. And we had a feeling that getting clean data into the hands of advertisers was the way to solve that. The setup around the opportunity in mobile advertising is well documented, so no reason to really spend much time there. It’s worth spending a bit of time on why that gap persists and how the mobile advertising world compares to online.
First off it’s important to understand how the online advertising components have evolved. In online ad serving I hear that a majority of display ads are now served through a set of services that provide realtime bidding (RTB). I don’t have any hard stats to point at for that one, but I’ve heard it from a few folks active in the environment and it seems to match up with logic and personal experience.
The nice thing for the publishers about moving inventory through RTB is that the advertisers can pay a price inline with their value for the individual impression. For impressions appropriate for retargeting the value can be much higher than the average value. The setup for RTB also allows for moving around third party sources of information called data management platforms (DMP) which raise the value of the inventory on the whole.
Sometimes an advertiser will have a source of info they directly want to hook up for marketing, but a lot of times the advertiser will want to tie into a third party source of information like BlueKai. The information can be anything from demographic info or credit score to purchase history – data which drives a huge amount of value for a marketer. Moving the data so it can be used with RTB requires a pretty complex set of machinery to populate and sync cookies.
Over the last few years the infrastructure in mobile, the operational systems, have quickly moved in the same direction as online. Thanks to some companies with fantastic vision like MoPub and Nexage the mobile side has actually moved a lot faster than the online world did. Probably because there was already a model system architecture to emulate and some standardized APIs like OpenRTB to build on. However, the business models don’t quite match up and there are some gaps in the mobile version.
The fact that folks spend most of their time in apps instead of in the browser is at the root of many of the issues. Since each app is generally sandboxed off from others, and there isn’t a consistent way to apply techniques like third-party cookies to deliver information, it makes it very difficult to hook into a data management platform or apply techniques like retargeting when working in mobile. Those techniques drive a lot of the value that comes out of using realtime bidding. So some folks have tried to use workarounds like using the UDID on iOS, which is a constant source of a lot of noise and discussion. And has led to Apple introducing an explicit identifier for advertising, which it seems for most people has just confused the issue even more even though it seems like a great step in the right direction. These techniques at least allow for the use of retargeting in mobile, even if there’s still a lot of work to make it consistent without being invasive for the user.
Another major difference that crops up in a bunch of different areas is the lack of visibility into application content. This was something we spent a lot of time poking at with Chomp, cause we were trying to figure out how to classify apps and understand their functionality. On the web this is much easier. You do what Google does and you crawl the pages and use the links between pages to make assumptions about the meaning and role of what the user is looking at. In applications you can’t crawl the content, and there are no links to use. The whole AdSense advertising model is based around that index of keywords and being able to match up an advertiser with what the user is looking at. There’s no equivalent to this in the mobile world as well. Not to mention the lack of ability to crawl the app content making the advertisers uncomfortable about putting their ads next to “unknown” content.
That’s all the downside stuff, things that make the current system of advertising in mobile end up less profitable than it otherwise could be. Fortunately they mostly stem from trying to jam the online model directly into mobile. So while we’ve been doing some bits to try to help solve those problems it’s all been fit into the overall framework of trying to detail the context of the user. And when I say context of the user I don’t mean just where are they, but also the device capabilities and constraints needed to paint the full picture of who this user is and what should match their current interest and intent.
This is all stuff that has long been held up as the promise of mobile advertising, but which we’ve never quite achieved as an industry. It was stuff we were talking about at AdMob in 2006, so it’s been a really long time coming. But this is all necessary for getting real commerce flowing through mobile, something we need to focus on to keep mobile healthy and on-track. We’re still at the narrow edge of the wedge when it comes to mobile services. Mobile has already touched a huge number of areas, but it hasn’t brought about the transformation I know is possible. There are even greater changes coming, and it’s time to start building for them.