Advanced Machine Learning, Data Mining, and Online Advertising Services
In this article, we review two well known marketing channels: RTB (Real Time Bidding) for display advertising and Google search engine advertising product (AdWords). We compare the performance of two advertising campaigns we have been running on these two channels and compare some characteristics of two marketing channels.
When web browsers use Google search engine to search for a product, item, or service, that provides an opportunity for Google to show relevant ads to people and make money. Google has productized this advertising service with their AdWords product. Using AdWords, advertisers and marketers can create advertising campaigns to advertise and sell their products. In AdWords campaign advertisers bid for relevant keywords that people might use in order to find the company's service/product.
Behind the scene, Google runs a customized version of a Second-Price auction to detemine who wins the ad spot everytime a user searches on Google website for a product. Wining the auction, the marketer will be allowed to show his ad to a web user. If a user clicks on the ad, then they'll be directed to the landing page where they can review and hopefully take the action (e.g. purchase, signup, etc). Google only makes revenue when Click event occurs. So, they call this type of advertising CPC (cost per click) search advertising.
The second type of advertising is called the display/banner advertising. There is a large number of publishers who own web and mobile app inventories with a large user base. There is also a large number of advertisers/brands who want to get in front of relevant audiences and show their ads to them. This creates a great opportunity to open an advertising market. One big problem is that how we can determine the optimal price for each inventory that publishers want to sell. Real Time Bidding is a micro mechanism that allows such transactions happen in real time at impression level.
So, as explained we have a market where publishers (inventory provides such as CNN) act as sellers and the advertisers/brands such as Adidas/Niki play buyers role. In game theory one way to determine the price of inventories in a market is to run an auction where a third party called auctioneer (e.g. Google Ad X, OpenX, Rubicon etc.) run the auction and make sure that the transactions happen properly. RTB is a mechanism that is run by all players including sellers (i.e. publishers like NBC), buyers (i.e. advertisers and DSPs), and auctioneers (i.e. Ad Exchanges).
Everytime a user opens an app or website like Facebook app on her cellphone, Facebook will send a request to their corresponding Ad Exchange (e.g. FBX) saying that they have an inventory that want to sell. FBX put that inventory on an auction and sends a bid requests to all advertisers who are participating. Next, advertisers have around 120ms to figure out how much they want to bid for that impression. The advertisers send their bid values and FBX determines the winner and make sure the winner's ad is shown to the corresponding Facebook user.
Ad Exchanges run the second price auction. Thus, the marketer with the highest bid value wins the winner has to pay as much as the second bid plus 1 cent.
For one of our customers we had to run an AdWords campaign where our goal was to bring the visitors to the customer's website to make the purchase. For this campaign we created a list of relevant keywords tieing them together in the same ad group and bid on them to generate traffics. Below we show some high level stats of running the AdWords campaign for a few days:
Another customer of us wanted to advertise some type of Dumbbells on mobile RTB channels. See below for some stats from running the campaign for a few hours:
It's interesting to compare some stats of these two types of campaigns such as Traffic scale, CTR, Cost etc. The first metric to compare is the scale of impressions. In the case of Adwords we have received 8490 impressions while for RTB we got 29k impressions for a few hours. Part of this is because the AdWords campaign was very targeted while RTB one is more in exploration phase.
The second metric is CTR. We see that AdWords CTR is 10 times higher than RTB's. This is partly because Google AdWords algorithm is programmed to maximized your clicks; however, this is not the case for our RTB campaign. Finallyi, it's interesting to see that average CPC for AdWords is $1.08 while average CPC for RTB is only 8 cents.
Even if above campaigns are not really comparable from statistical point of view, but still above numbers tell us something about the nature of those marketing channels. The last thing that we want to highlight is that the pricing model for buying traffic in RTB is CPM (i.e. cost per mile). Thus, if the price of an inventory (e.g. mobile app or website) is $1.0, we'll pay $1.00 for showing our ads to 1000 users from that app.
For instance, for above RTB campaign average CPM across all bought inventories is $0.177. This means that if we show our ads to 100k users, we will be charged around $17.0 which is not that expensive compared to AdWords. A big reason is that when people use search engine to find an item, they are very likely to be at the end of the purchase funnel and that justifies AdWords higher price compared to RTB.
In the next post, we will digg deeper into the technical side of RTB. So, please stay tuned by then!
If you have a business and want to advertise/market your app or product to acquire new customers or for branding, contact us at info@AIOptify.com