Continual keyword discovery in PPC search marketing is immensely important. There are actually a number of instances where strategically expanding your keyword portfolio offers more ROI than bid manipulation.
The concept of "diminishing returns" and modern portfolio theory is applicable to more than just economics. In a keyword portfolio, not unlike in a stock portfolio, there is a point in the optimization of a vertical when the work you put in and/or increased bidding will cease to yield the same (and often any) sort of return.
Basically, the premise here is that there is a certain amount of “low-hanging fruit” surrounding a given keyword and associated keywords. My site dedicated to providing people with investment information might be seeing a lot of success from bidding on the PPC or AdWords keywords “portfolio investment strategy”.
That’s great, but anyone who's managed a pay-per-click campaign knows that I’m likely to run into diminishing PPC returns. Let's take a dramatically oversimplified example (complete with made up numbers!) to illustrate this point:
If I start by bidding a dollar for the "portfolio investment strategy" keyword, it might convert at 5 dollars a conversion. Conversions are actually worth 10 dollars to me, so this is great, but I might only get 20 conversions a month. When I bid 10 dollars, meanwhile, it might cost me 12 dollars a conversion, but now I'm getting 100 conversions a month (by bidding more my ad is able to be impressed upon more searchers, generate more clicks, and often result in more conversions overall; but there is more competition). The problem here is that as I increase conversions, my margin is shrinking (and in this case, disappearing altogether):
We can also see this principle represented in the following graphic:
The value of continual keyword discovery is that instead of investing time and resources (via AdWords bidding with bid management tools or by attempting to assign monetary value to content and promotional activities) in trying to find your “efficient frontier”, you get to create a brand new graph.
This new graph is complete with a fresh crop of low-hanging fruit. Rather than spending more money for the 100 conversions, you spend less money for two sets of fifty or five sets of twenty.
But the real question is: how do you effectively create this "second graph."