So – if you're doing keyword research, where should you begin? Unless you're an affiliate marketer you already have a product and since you're the target audience of this article – I'm going to assume that's the case. For the purpose of this article I'm going to pick a hobby of mine and also an area where I don't have a client and imagine I'm doing keyword research for the imaginary online downhill mountain biking store DH Mountain Bikes.

So Where To Begin ...

The first thing one needs to do is try to think up all the possible phrases that might apply. I call this my seed list … it's the list of phrases that my research starts with and is generally based on brainstorming. In this case the list would be:

• downhill mountain bike
• dh mountain bike
• mountain bike

The keyword tool I generally use first is Google's keyword suggestion tool. There are other great tools but I've found Google's tool to be as accurate as any other, the price is definitely right (free), and they're very good about providing the information required to know just how wrong the data is if you know where to look. So let's do just that.

Before we begin you'll need to head over to Google's keyword tool at https://adwords.google.com/select/KeywordToolExternal. In the top left (for now) you'll see a link to a beta version of the tool. Click on the link and you'll be at the new version of the tool which will provide you easy access to much more information – as long as you know what to look for. So let's begin with our three seed phrases.

When you see the list you'll first have to know what the numbers are. This tool is a tool designed for AdWords and the default number is the Broad match which means it includes every phrase with the term. For example, the term “mountain bike” has a broad match total of 2,740,000 which will include “downhill mountain bike”, “mountain bike parts”, “kona mountain bike”, etc. etc. What we want to know is how many searches are for “mountain bike”. Down the left-hand side you'll see a set of check boxes. Deselect “Broad” and select “Exact” and you'll get the Exact match numbers – the number of searches for the exact phrase. You'll quickly see that 2,740,000 drop to 450,000. This is how many people searched the GOOGLE SEARCH NETWORK for “mountain bike”. Why is this in caps – because it's so commonly misunderstood that I definitely want your attention brought to it. This isn't the number of searches on Google.com – it's the number of searches on all sites who's search is powered by Google. From YouTube to Beanstalk's blog search – it's all in there so the data starts to get skewed from the start. Then let's add in all the automated queries from rank-checking tools and just manual searches from you and your competitors and the data gets further skewed. This skewing will exist in all data – the thing I like about using Google is that at least we know more about what's adjusting the data.

OK – so from there we need to organize the data into a more useful set of information. To do this one needs to understand the columns of data. The first column is the keyword, the second you'll see is a link to the term on Google Insights. We'll get into this later. The next is Global Monthly Searches – this is the average number of searches/mth worldwide. This can be helpful in some industries but in ours – I'm only concerned with the US market which is where my imaginary store ships to so I'm more interested in the next column Local Monthly Searches which is the number of searches in the US (or whatever region I've specified when entering my keyword phrases). This is the data I'm interested in. The last column is the search trend. This is extremely important but often overlooked. It is a column that wasn't visible by default in the old/current version.

OK – let's organize our data by search volume. Click on the “Local Monthly Searches” and you'll see the keywords order by descending search volume. With this data in front of me I then typically look over to the Trend data to see what I can find there. In our case we're going to see an increase in search volume in the spring and summer. This make sense of course. Think of your industry and see if the trends reflect what makes sense.

I'm also looking for anomalies. Often I'll see phrases that jump for a single month. One has to know that unless there was a news story or other event that would spark interest in a single term or brand – a tool or some other such incident is likely falsifying the data. You need to look at these trends and see if they make sense. If not – you need to either test the phrases with PPC or just skip over them and select different phrases. There's little worse as an SEO than focusing energies on a phrase only to find that the search volume is not what was expected based on the estimates delivered.

So now what?

So what do you do once you've filtered your data down to just what you're interested in looking into competition levels on. Well – the first thing I do is to look to the trends to see if there are any phrases that obviously need to be filtered out. In this case there really aren't any high in the search volume column. So the only thing left is to look at the competition levels to see what makes sense. For our purposes we'll be dividing the list and research into two categories:

Major phrases – We need to decide what the long-term goals are going to be and the targets for the main pages. These will be the totally generic phrases such as “mountain bike” and “downhill mountain bike” as well as brand or type specific phrases such as “specialized mountain bike” and “full suspension mountain bike”.

Longtail phrases – We also need to look into the types of longtail phrases we're going to want to target. In this case I know I'll want to target specific parts which will require new research. I will spare you the details there but I'll end up with specific models of components such as “hayes mx2”. You don't need to know what that is – you need to know the makes and models in your industry (or other longatil opportunities such as “new york hotel with jacuzzi”, etc.)

I generally would gather together a list of 15 or 20 major phrases and 50 or 60 longtail phrases and would then head into the competition analysis to determine which phrases to move forward with.