Web Science/Part2: Emerging Web Properties/Web Search Ecosystem/Metrics for (online) advertisement/script

In online advertisement there are several metrics. They are called CPC, CTR, CR, BR, CPM and CPA and are often confused. We know want to understand how these metrics are defined and which of our stakeholders should pay particular attention to which metric.

The set of metrics can be split into two subsets:
 * The first set consists of CPM, CPC and CPA. They all start with a CP which stands for "Cost per". Technically one could say that these are not metrics but just payment formats or price tags for advertisements. We will see that that in combination with the second set of Metrics they can also be used as metrics.


 * The second set consists of CTR, CR and BR. As we see they all end with an R which stands for rate. So these metrics are usually given in percent and roughly state the relative amount of people who did a certain task while interacting with the advertisement.

Let us have a deeper look at the first set again:
 * The CPM states how much an advertiser has to pay to the ad agency in order to have the advertisement delivered 1000 times to the customers of the website of the affiliated publishers. As in other cases the revenue is usually shared between the ad agency and the respective publishers. This metric is highly motivated from the offline world where it was also used for magazine and television ads. It favors the publisher a lot since he exactly knows how much money he will earn for attracting 1000 page views. On the other side the advertiser has a very high risk. The ad could be unappealing to the customer and not generate any traffic to the advertiser's website.
 * On the other hand we have CPA. It states the amount of money the advertiser pays (to the ad agency) if a consumer interacts with an advertisement in a way such that a certain goal on the advertisers website is met. An example could be shopping of an item. In this case the amount of money could be a percentage of the price of the sold item. Drawbacks are the fact that affiliate marketing is complex to realize as it requires tracking technology from the ad agency on the advertisers website. Another one is that there is a high disadvantage for the publisher. Now he carries all the risk. If no one of his users becomes a customer of the advertiser he will not have any revenue.
 * The best mix of both worlds seems to be the CPC. In this setting The advertiser pays a certain fee for every click on the advertisement. The publisher still has the risk that no one interacts with the ad by clicking on it but on the other side he gets paid already even if the customer is not subscribing to a service or buying a product from the advertiser. CPC has one big drawback which is click fraud. It is actually even a risk to the ad agency and has to be detected by them. We will discuss this problem in a later unit of this lesson.

So let us turn to the other set of metrics.
 * The CTR is defined as $$ \frac{\mathrm{Clicks}}{\mathrm{Views}}$$ or also $$ {\mathrm{Clicks} \over \mathrm{Impressions}}$$. This means if an ad is displayed 100 times and receives 5 clicks the click through rate is 5%. Depending on the payment model this rate is rather interesting for the publisher or the advertiser. Since CPC is the dominant payment model we have a closer look at this scenario. The click through rate has no influence to the advertiser. He buys a certain amount of clicks and pays per click. A higher click through rate is of no value for him. This is different for the publisher. Let us stick with the 5% click through rate. If he sold 50 clicks to an advertiser this means hat he needs 1000 views to fulfill the goal. Let us assume he has a constant 1000 page views per day. If he could double the click through rate he would have fulfilled the goal after only half the views (which means 500). In this case he has another 500 views left which are opportunities to sell more clicks to another advertiser. A really important fact to note is that the click through rate tells us what the user thinks about the relevance of the ad to him. So if the user think the ad is relevant to them the click trough rate will rise. Obviously the click through rate becomes more important to the advertiser if he pays on a CPM basis. Just pause the video for a moment and think about the importance of the CTR for the publisher if the ad is delivered on a CPM basis! BREAK You are right in the CPM setting the publisher does not need to care about the CTR.


 * Since CPC is the dominant model and CTR was not important for the advertiser we look at another metric which should be his main focus. It is called the BR which is given by $$R_b := \frac{\mathrm{T_v}}{\mathrm{T_e}}$$ where 1.) Tv = Total number of visitors viewing one page only and 2.)Te = Total entries to page. The bounce rate is more complicated to measure than the click through rate. It requires session handling on the site of the advertiser. The bounce rate tells us how relevant the ad really was to the user. If the ad was not interesting the bounce rate will rise.


 * The CR is defined as $$ \frac{\mathrm{Number\ of\ Goal \ Achievements}}{\mathrm{Visits}}$$. An achievement can be anything but has to be defined by the advertiser. For example an achievement could be the completed shopping process of some products, registration to the site, clicking of another link, submitting a web form, subscribing to a newsletter or some service, becoming a follower in a social network, leaving a comment,... Obviously the Conversion rate depends heavily on the type of achievement. Usually conversion rates are pretty low (less than a percent). The conversion rate should be the single most important metric to any advertiser. From the conversion rate one can estimate the acquisition cost of a customer. This can be done by the following formula: Acquisition cost = $$\frac{\mathrm{CPC}}{\mathrm{CR}}$$. As we can see lowering the CPC or increasing the Conversion rate will lead to a lower Acquisition cost which should be the goal of any advertiser.