If we can make computer programmes that can predict what the weather
will be like in fifty years time and how we can change it, surely we can make a
computer programme that will help us to succeed at Paid Search Marketing (PSM).
The simple answer is yes, we probably can, but it is far from easy. Indeed a
number of approaches to this challenge are currently in progress.
The simplest model is the square root law. This states that the increase in sales volume is the square root of the increase in advertising cost. It is a silly model that has no bearing to reality. You are advised to ignore it, it is incorrect, though some people believe it.
PSM research can provide excellent data. For each keyword combination we can count the searches carried out in which the ad appeared, count the clicks and measure the sales. This data can be incorporated into predictive marketing models that can be used to optimise the PSM and pay per click campaign. For simple PSM campaigns using only a few keywords this can be carried out relatively easily using spreadsheet calculations. A problem is the complexity of the problem increases exponentially as the number of keywords increase. For instance, with say 50 keywords, trillions of data points need to be evaluated.
Key to this approach is the ability to describe the problem in such a way that it is solvable. One of the most sophisticated methodologies applied to this has been the development of special keyword models that use large quantities of historic data. Optimal solution methods using optimisation algorithms have also been developed.
Put into practice this approach provides automatic compensation for the ever changing marketplace, sophisticated keyword modeling, and solutions that give the exact keyword bid price to maximise revenues.
Of course this is not available to the small retailer as the costs of implementation are substantial. This approach is really targeted at campaigns in the hundreds-of-thousands of dollars region. However, these things have a tendency of simplifying and filtering down. Eventually you might be able to hook up to the internet and let your PC manage your PSM campaign for you.
The simplest model is the square root law. This states that the increase in sales volume is the square root of the increase in advertising cost. It is a silly model that has no bearing to reality. You are advised to ignore it, it is incorrect, though some people believe it.
PSM research can provide excellent data. For each keyword combination we can count the searches carried out in which the ad appeared, count the clicks and measure the sales. This data can be incorporated into predictive marketing models that can be used to optimise the PSM and pay per click campaign. For simple PSM campaigns using only a few keywords this can be carried out relatively easily using spreadsheet calculations. A problem is the complexity of the problem increases exponentially as the number of keywords increase. For instance, with say 50 keywords, trillions of data points need to be evaluated.
Key to this approach is the ability to describe the problem in such a way that it is solvable. One of the most sophisticated methodologies applied to this has been the development of special keyword models that use large quantities of historic data. Optimal solution methods using optimisation algorithms have also been developed.
Put into practice this approach provides automatic compensation for the ever changing marketplace, sophisticated keyword modeling, and solutions that give the exact keyword bid price to maximise revenues.
Of course this is not available to the small retailer as the costs of implementation are substantial. This approach is really targeted at campaigns in the hundreds-of-thousands of dollars region. However, these things have a tendency of simplifying and filtering down. Eventually you might be able to hook up to the internet and let your PC manage your PSM campaign for you.
No comments:
Post a Comment