The Life and Times of AdMob

Today we’ve updated the AdMob engineering blog with two new posts.  The first of these two posts continues an earlier discussion about ad serving mechanics for the mobile web.  The earlier post in the series talked about how mobile ads are retrieved, whereas today’s post motivates why ad retrieval for mobile sometimes needs to be different than it is for the web.

Our second post today discusses AdMob’s new ad creation user interface and describes some of the techniques that went into building it.  We also offer up a hint at the first of several upcoming contributions back to the open source community, an extension to the EXT JS AJAX library’s tree widget.

–Kevin Scott, VP Engineering

one web?

July 13th, 2008

The mainstream media seems to be taking notice of the middle web trend we pointed out awhile back

http://www.nytimes.com/2008/07/13/technology/13stream.html?partner=rssnyt&emc=rss

again with the iPhone? :)

July 8th, 2008

There has been some interesting discussion making its way around the blogosphere on our present and our future, so I figured I’d take the opportunity to comment.

It started with a venturebeat article with a hypothesis on how we are doing and how we will be doing soon. (that is not the part I wanted to comment on)

This led to a number of postings which more or less call out the iPhone as particularly dangerous to our business, and I’ve included links to a few just for reference:

http://yardley.ca/2008/07/07/there-is-no-mobile/

http://www.seriouslytech.com/2008/07/07/is-the-mobile-ad-market-poised-to-break-out-or-to-disappear/

The question around the iPhone is certainly not new to us… actually a search on our blog for iPhone turns up a number of interesting things:

Our thoughts on what the first iPhone meant for AdMob

Our discussion of our iPhone ad units

Some of the neat things we learned from the iPhone

A prediction that iPhone (and similar class) devices will have a dramatic impact on mobile browsing

As a last point, I figured I’d post a (really long) comment I just left on Silicon Alley Insider in response to this article.

See you next iPhone :)

Dan, very interesting post. You asked for some AdMob folks to chime in, so I figured I’d oblige you. We’ve been spending a great deal of time thinking about and working with the iPhone. We share your view that it is only a matter of time (and not a long matter of time) until the majority of mobile browsing is done on devices that have full featured browsers. To that end we have been working with iPhones since they launched and have been allowing advertisers to reach iPhone users for quite some time (you can see a few of our campaigns here - http://www.admob.com/s/solutions/iphone_ad). To put some numbers around this, we served 52 million impressions on the iPhone and iPod Touch in June up from 39m in May (32% month on month increase).

Although your analysis makes sense on the surface, I would take issue with a number of the underlying points you made. The first is that our sole differentiator is based on the installation and serving of advertising. Although it is a different process than online ad installation, the fact is that this does not differentiate us from any of our competitors in the mobile advertising space (Google, Yahoo!, AOL, etc…). All of them have similar installation procedures. Also, considering the fact that the install code is publicly available it would be very difficult to build a defensible business around installation. The reality is that our differentiator comes from scale of mobile specific advertisers and some fairly sophisticated engineering and optimization built around which ad to serve and when.

Now to another central contention you make: You essentially assume that simply because “regular” ads can be served, they will inherently be optimal. The reality is that the value of an ad is based on three parties: the publisher, the advertiser, and the user. In the case of the publisher, they would like to see advertising that is relevant, plays well with their content both visually and thematically, and obviously, maximizes monetization for them. In the case of the advertiser, they are interested in making sure they have reached the most interested audience they can, in the right place at the right time (and now on the right device), and that they are paying a fair rate. In the case of the user, if you can show them a relevant ad that is respectful of both their experience and privacy, you will have a win.

The reality is that in all three cases, ads designed for the web as viewed from a PC are not and will not be optimal. Let’s think about advertisers for a moment. Let’s say you run an ecommerce site and are willing to pay 50 cents a click to get someone to come to your site and buy a product. The fact is that they are much less likely to go through your shopping cart on a phone, regardless of the browser. The same holds true for a lead gen advertiser who is expecting users to fill out a long form. In these (and many other) cases, the advertiser will actually be upset if a publisher or ad network recklessly shows their ad regardless of device. If they bought assuming it would show on a PC, it’s simply not fair to show it on a phone. On the flip side, there are plenty of advertisers who would pay significantly more to show up in the context of a smart phone with all of its unique advantages (multimedia, phone calls, etc…), as long as they know that’s where the ad will appear. As a publisher, you will be getting much more relevant (and higher paying) ads if you work with someone to sell specifically for the device.

Although the idea of a unified web where everything is exactly the same (including advertising) is intriguing, the reality is that all the recent advancements in digital advertising (search, behavioral, demographic, and psychographic targeting, etc…) have been about differentiation of audiences rather than amalgamation. In the context of advertising, there is significant value locked up in even subtle differences and we know from experience that the device you are browsing on is no subtle difference.

In short, we are well aware of the impact that the iPhone and smartphones in general are going to have on mobile browsing. The capabilities they make available both in browser and in app will further accelerate mobile data usage, and mobile advertising along with it. We view these developments as the future of our company, rather than some impending doom, and if we could magically wish their adoption to happen faster, we would.

AdMob in South Africa

July 3rd, 2008

Those of you who read our Mobile Metrics Report know that South Africa is one of AdMob’s top markets by impressions served, which means that South Africa is home to a large group of active consumers of the mobile web.  South African mobile users generated 159.2 million mobile ad impressions in the month of May in our network making the country our fifth largest market after the United States, India, Indonesia and the United Kingdom.  Why is browsing the mobile web so popular in South Africa?  Low landline penetration, high mobile phone adoption and low data fees have created an environment where mobile web usage is thriving.

Since South Africa is an important market for us, we’re growing our presence in South Africa by putting a team on the ground to work with local advertisers and publishers.  Most of the traffic from South African users in our network is on mobile websites hosted in other countries.  We see the fact that users are mostly visiting international sites as a tremendous opportunity for local sites to create compelling content for the South African market.  As more South Africans start visiting local content-focused mobile websites, local advertisers will likewise adopt mobile advertising as a great way to engage their target audience.  If you’re as excited about opportunities to grow the mobile web in South Africa as we are, drop us a line: SouthAfrica@admob.com
- Clay
Director, Business Development

Over the past year AdMob engineering has been growing rapidly. Our engineers have been building increasingly interesting pieces of technology to serve the needs of our customers & the mobile ecosystem. Some of the work that we’re doing has a very visible, user-facing component. For instance, we’ve recently discussed AdMob Analytics and improvements to our ad creation process.

We’re really proud of these products and their user interfaces. However, we’re equally proud of all of the machinery that powers AdMob that isn’t always visible to our users. In order to share some of the details of this behind-the-scenes machinery, we’ve created a new AdMob Engineering Blog. In addition to simply sharing our engineer’s enthusiasm for things that we’re building, we also hope that this blog allows us to contribute back to the engineering community some useful best practices we’ve discovered as we’ve employed a mix of proprietary and open source technology to build a highly-available, 100M+ request per day Internet service.

The types of things that we hope to share with you over the coming weeks are how we’re using Hadoop to process large volumes of data, the tools and techniques that we use for reliable message delivery between our internal services, how we’re using a combination of third-party and custom AJAX component libraries to build rich, in-browser user interfaces, the tools and techniques we use for service monitoring, and how we’re managing application deployment across hundreds of machines and multiple data centers.

Posts on the AdMob Engineering Blog will be written by engineers and targeted at a technical audience. That said, we invite everyone who is interested to subscribe to the blog’s feed or to place this blog in your regular reading list.

We’re kicking off our engineering blog with a post about the mechanics of mobile ad serving. We’re often asked how mobile ad serving differs from traditional, online ad serving. This post provides part of the answer to this question.

–Kevin Scott

Copyright © The Life and Times of AdMob. All rights reserved.