A Scientifical Approach to Adwords
This article is useful to explain the relations between the various variables involved in a PPC campaign and its main intent is to focus the crucial factors that will turn your strategy into a success, or not.
Requirements
It is assumed that you know what PPC is and how it can be used to generate a revenue. This model can be used with PPCs more than Adwords, of course, assumed they are using the same pricing logic.
Moreover, a knowledge of elementary algebra and linear systems is required to understand the mathematics I used in the problem.
A last note: I am not a math geek, so I may have incurred in mistakes. If so, let me know it and I will fix the various bugs.
The model
Okay, let's go inside the problem now.
In this model we focalize the attention to the balance between the money spent and the money earned. We will see what are the necessary conditions in order to have a positive net revenue, and we will use the model for particular cases, such as what is the minimum CTR we must reach in order to have a net revenue (and see if that aim is realistic or not).
Further applications of this model can be done to extend its purposes to a specific product selection or PPC campaign, for instance using statistical/stochastycal methods.
The variables are:
CTR
SALE_PRICE
SALE_COMMISSION
CPC
MAX_PER_DAY
The relations between them are:
SALE_EARNING = SALE_PRICE * SALE_COMMISSION
SUCCESSFUL_CLICKS = CLICKS * CTR
GROSS_REVENUE = SUCCESSFUL_CLICKS * SALE_EARNING
NET_REVENUE = GROSS_REVENUE * MAX_PER_DAY
Being CLICKS = MAX_PER_DAY / CPC
In this model we assume that the net revenue is referred to a time period of 1 day, and that the campaign performance is the same on everyday of the week.
Being more realistic, this is not that true because it is seen from the experience of thousands of marketers everyday that there is a substantial variation of the campaign performance in the days of the week. On a larger scale, there's a variation also in the months of the year and using historical records we can identify yearly trends.
For these reasons it would be more correct to define the above variables in function of the time also, but in this starting model we will limit the analysis to variables that are steady in time.
Using shorter variable names and playing with them a bit the result is:
a = b*c
d = e*f
g = b*c*e*f
h = e*(b*c*f - i)
where i = CPC and h is the net revenue.
Since we want the net revenue > 0, this means e*(b*c*f - i)>0 coming to the two conditions:
e > 0 (of course, we assume e i.e. clicks > 0 because we buy them)
and
i < b*c*f
Where i in this case is the [B]maximum CPC[/B] you can afford in order to have a net revenue.
To be more clear, the maximum CPC you can afford is smaller than the product of the sale price, the sale commission and the CTR.
For example, if you sell an article worth $20, where your commission is 5% and your estimated CTR is 5%, the maximum CPC you can bid is $0.05 . In order to be safer you can define a risk factor R greater than 1 of your choice, so the formula would become
i < b*c*f*R^(-1)
I have made a simple XLS file helping you to calculate your maximum CPC that you can find here: http://www.forumboosting.com/maxcpc.xls (will be active in a short time, got temporary problems with my FTP)
Further developments of the model will be published in the following days. Stay tuned :-)
Falco85
