We started using AdMob to deliver advertising inside our iPhone applications shortly after the launch of the app store. The mobile advertising network deserves praise for being one of the front runners in the race to deliver effective mobile advertising but the system is far from perfect. There are a lot of things I really like about AdMob but frankly until some things improve and change the network will not be profitable for a large portion of mobile publishers and advertisers. To be clear I have much respect for companies that build solutions capitalizing on cutting edge technologies, I also understand how hard it can be to try to bring those solutions to “blue ocean markets”.
Case #1: Where’s the Clicks?
In testing we setup several external landing pages and some basic ads to looks at the analytics. What we found was pretty scary in this world where numbers mean everything. Of the clicks we paid for we only saw about 60% of them actually ever hit the page at all. After a few variations of all settings we came to the conclusion that the vast majority of the missing clicks were “misclicks” on the device and the user closed the mobile safari window before it actually loaded the page, thus we pay for the clicks but the analytics pixel on the page never fires because the page does not load.
This is obviously a huge problem and many factors contribute to this case. Developers are far from mastering the mobile user interface, bad interfaces and bad ad placements will result in misclicks on a system with so little real estate available in the UI.
Another contributing factor is that AdMob advertisers cannot target specific mobile sites or applications for ad delivery, this means that you cannot filter out the sites or apps sending “misclicks” or bad traffic in general.
In summary AdMob must allow advertisers to target and filter the applications and sites that display their ads. In addition AdMob must develop better controls to protect against click fraud, misclicks, and other types of generally bad traffic.
Case #2: Demographic Targeting
Before I start in on this topic I want to give AdMob some credit for their SDK. Software developer kits are a rarity in the Internet advertising and marketing scene as a whole and we were particularly happy to work with it, great code, great documentation, great functionality.
Unfortunately the functionality that the SDK provides us as a developer seems to have little point in the end. The main advantage provided application publishers in the SDK is the ability to provide AdMob demographic and location data for the individual user under the auspicious that AdMob will return highly targeted advertisements for that individual prospect.
Through testing we have seen absolutely no evidence that this data is being used to deliver more relevant advertising in any regard. As we are both a publisher and an advertiser in AdMob I know that there are very little targeting options at all in the advertiser interfaces. There are some great device and location options but nothing for demographics. I suspect the missing ad targeting is the result of a lack of advertisers able to benefit from the targeting and deficiencies in the AdMod software that returns targeted ads dynamically.
Case #3: The Impression Exchange
I love the idea and thinking behind the impression exchange feature that AdMob recently launched. The impression exchange allows publishers to exchange advertising impressions in their applications. I find it peculiar that they enabled it by default in our portal for one of our apps but I understand why they did. Without any impressions in the exchange there is nothing to exchange, the default enable looked like a vehicle to help the exchange reach critical mass and have value sooner.
As I said I really like the idea of being able to back fill my inventory with exchange impressions and even trade out the lowest paying impressions for exchange impressions but in this case the bottom line is that currently it just flat out didn’t get us any results. We saw no noticeable download increases after switching our entire inventory over to the exchange for several days to test. On top of that there is no reporting to facilitate the exchange, it just lumps your exchange clicks in with your overall clicks. You cannot tell your exchange clicks from any other type of click and in this case the total number of clicks went down dramatically even after optimizing the ad text line and graphic.
AdMob is is leading the way in a very young industry and they deserve credit for overcoming the obstacles that they have overcome to get to this point. The ideas and philosophies behind the platform are right on in my opinion but the immaturity of the industry and lack of attention to critical details hinder the system in it’s current form.
We will continue to use AdMob and continue to test new and existing features but we are making plans to keep the doors open for other solutions. We are currently working on new internal mobile adserver technologies that will give us a layer sitting above AdMob and other providers so that we can dynamically insert and rotate other networks and advertisers with the AdMob inventory.
AdMob is a great way to get a little extra cash out of a free application and if you need downloads to jump start new applications it is one of the few ways to get them. With $350 million in new funding from Sequoia I expect to see improvements in the areas of concern I covered in this article, how fast that might happens is anyone’s guess.