First we go in and look at your Yahoo! logs and pull out your very most important keywords, as well as a bulk list of everything ever searched for to reach your site for the past 6 months. This helps you optimize your Yahoo! Shopping listing as well as give you a leg up on the OVERTURE.COM keyword buys.

From this we compile two lists:

(1) A list of keywords and phrases that actually produced revenue, and (2) a VERY LONG list of unique phrases that produced traffic. >From these two lists you can decide what words to buy on OVERTURE.

The last client we did this for got a list of 3,000 unique words and phrases used to search for his store. His sales have really jumped in the past 60 days, and so much of his traffic is from OVERTURE and their affiliated partners Yahoo, MSN, ALTA VISTA, NETSCAPE SEARCH, etc.

Example of datamining results

Just did some "keyword mining" on one of our Yahoo! Stores. There were some very interesting results.

From roughly 215,000 visitors you can see exactly what 60,000+ folks typed in various search engines and directories to find our Yahoo! Store. This data came from 50 different sources with roughly 30,000 different search terms. The rest of the traffic came from directly typing in the URL, "unknown search terms," banner ads, and links to the site.

Filter the data into two groups: (1) Revenue Producing Keyword Phrases - (keywords that produced online sales) and (2) Traffic Generating Keyword Phrases (keywords that generated traffic, but no "trackable" sales.)

Only about 15% of my online sales are traceable to the exact keyword phrase, so we just have to weight the known data to reflect this.

REVENUE KEYS -- The first group of keywords are now prioritized by sales dollars to go after the most profitable keywords, and the goal is to try to stay in the OVERTURE top 4 and still make money. There were 561 unique keyword phrases that directly resulted in online sales.

The great thing is that 54% of this store's directly trackable keyword sales came from the TOP 40 keywords and phrases, so that's where we focus our dollars and attention.

Looking at the competition in OVERTURE, about half of the words don't even have bidders. And most of the other half consists of bottom feeders paying 2 or 3 cents for the keywords. Holy COW!

There were also 3 phrases where a national manufacturer who sells direct is paying close to a dollar per click through. That tells us how much THOSE phrases are worth to that guy! Probably need to open another Yahoo! Store just to go after THOSE phrases!

TRAFFIC KEYS -- After consolidating the same phrases that were on multiple search engines into a single unique list, the second list boils down to about 15,000 unique words and phrases. For example, a search for "blue baby bed" on Yahoo, Alta Vista, and Excite would count as 1 unique phrase with 3 searches.

Since only 15% or so of our sales come from directly trackable links with the specific search phrases, we know that some of these 60,000 shoppers with 15,000 keyword phrases really ARE buyers. How do you get this list down to a manageable size?

I take this list of 15,000 keywords phrases and start chopping to remove the words that don't really make sense. In your Top 50 Sources of Traffic, there may be as many as 25-30 search engines, meta search engines, and directories. Any keyword phrase that was only requested 2 or 3 times from these sites over a period of many months probably isn't worth spending the time to write a title or description.

I also use the Pareto Principle, also know as the 80/20 rule (80% of your results comes from 20% of your efforts) to see where the break is. I can get 70% of my "bulk keyword traffic" with about 2000 keywords. OVERTURE is going to cut this list pretty hard anyway. My goal is 750-1250 nickle keyword phrases when I get finished.