Admob Marketplace – Burning cash and fighting click-fraud

[Amethon]

As a leading provider of analytics solutions for the mobile industry, Amethon Solutions has focused its market positioning on the accuracy of our Mobile Analytics solution, especially compared with traditional page tagging methods. In order to verify our accuracy, we use a number of testing methodologies including known traffic sets as well as manual analysis of real mobile web traffic.

In order to generate mobile web traffic we have set up a small mobile site (amethon.mobi) and pay ad networks including Admob Marketplace and Google Adwords to drive traffic to the site. The traffic is then manually analysed from the server logs in order to verify the various metrics generated by our Mobile Analytics solution are correct.

We have good experience using Google Adwords to drive traffic to our main website and utilise many of the features such as geo-targeting, publisher targeting, key word selection and daily budgets. While there are a lot of similarities between Google Adwords and Admob, there are also a number of important differences.

First is the payment method. Google is effectively post-paid where your credit card is charged monthly in arrears based on the cost of your ads. Admob is based on a pre-paid wallet where you add funds in advance of spending them. The minimum you can add to your wallet is US$50 as shown in the image below.





While US$50 is not much compared with what many people would be spending on Google Adwords per month, it is a lot for someone looking to run a small test of mobile advertising with Admob.

Of more concern is the daily budget limits. With Google Adwords you can budget as little as US$1 per day but with Admob, the minimum daily spend is US$100 as shown in the picture below.







$US3000 per month is definitely more than most people would be spending on their web campaigns so to have this as the effective lower limit for an ongoing mobile campaign (short of constantly recharging your wallet with only $50) seems very restrictive. Having said that, we worked with a client who was analysing the performance of a $1,500 per day Admob campaign!

The next step is to create the ads and set the CPC bid price. Admob supports both standard WAP ads (1c minimum bid) and iPhone specific ads (3c minimum bid) as shown below.







We created three different ads with standard WAP and iPhone variants of each as shown below. Since text ads are limited to 35 characters it is difficult to have compelling creative but it is also possible to upload image ads with four different sizes (120x20, 168x28, 216x36 and 300x50) which are served depending on the screen resolution of the handset.

Funny Videos

Mobile Analytics

Content Fingerprinting

Having already committed $50 to our wallet, we chose to enter the minimum bids assuming that this would reduce the total impressions and slow down the click through. To our surprise, US$50 was consumed in a little over 1 hour! As can be seen by the screenshot below, our campaign generated more than 1 million impressions resulting in 5,000 clicks with an average CTR of 0.48%.







One would assume that the ‘Free funny videos from Amethon!’ ad would significantly outperform the product specific ads but as shown from the Mobile Analytics screenshot below, this was not the case.







The product ads combined to deliver more click throughs than the Funny Video ad while in terms of click through rate, the Funny Video ad was only marginally better (0.51% CTR) compared with the two product ads (both 0.47% CTR).

This raises questions about the quality of the inventory and the effectiveness of Admob as a credible source of traffic. While it is pleasing to think that 1518 people were interested in our Mobile Analytics product, a more likely scenario is that mobile users struggle to differentiate the advertisement from the rest of the site and click on the link anyway. Of the 5000 visitors to our site, only about 7% went to the Mobile Analytics product page despite the fact that 1518 clicked on the Mobile Analytics advertisement.

The final issue that we found with our Admob traffic was the high proportion of what we believe to be click fraud. This is where publishers repeatedly click ads on their own site in order to generate revenue.

When analysing both the click path reports from Mobile Analytics and the raw traffic logs, we noticed that a number of visitors were exhibiting strange behaviour. Rather than arriving at the landing page and then going further into the site or simply bouncing away, these visitors would view the landing page repeatedly without going any further into the site. When we looked at the referring page (the publishing site where our ad was clicked), the domain would be the same but the individual page would be different which suggests they clicked our ad, when back to their site, move to another page and then clicked our ad again. From this we can assume that Admob will block obvious click fraud i.e. repeatedly clicking the same ad on the same page.

The biggest offender clicked through 458 times in about 30 minutes without going to any other page and we estimate that the total number of fraudulent clicks was around 13% of the total click throughs. This suggests that it is a serious problem for Admob and potentially other ad networks, and is something that advertisers need to consider when investing in these types of campaigns.

So in summary, Admob is very good at generating traffic for our Mobile Analytics testing but I would hesitate to recommend it as a useful marketing tool given the restrictive budgeting capabilities, the quality of the subscribers and the high rate of click fraud.




























Get More from the original blog...