I just got an email from an old associate.  He is a PPC guy and is doing an SEO project.  He asked for basic insight to “double check” on the keywords that his client provided to him, wanting to make sure that he wasn’t going to spend his time optimizing for keywords that were seldom searched.

Good call.  In fact, when Ephricon takes on a new client we ask for lots of input from the client, but we reserve the right to have final say on the keyword priority list.  I feel as though that’s part of what clients pay us for.  We hold ourselves responsible not for rankings, but for traffic and leads.  As such, we need to have significant say as to which keywords we target.  This goes for both SEO and PPC.

My reply to him is below.  I intentionally tried to keep it fairly high-level.  The guy wants recommendations, not thorough discussion of every tool ever made.

Hi <<Name>>,

Good hearing from you.  There are a number of options, of course, but the fastest/easiest one is the good ole Google Keyword Tool.

Search volume is pretty much the same for PPC as with SEO – b/c at the end of the day its still just counting the number of people who are searching.  Your approach to target them (PPC ads vs. organic listings) is different, but there is still the same “supply” of people searching.

We also use KeywordSpy, SpyFu, Trellian’s Keyword Discovery and other programs too, but I feel like Google’s free tool gets you 80% of what you need quickly and easily.

That’s all volume.  For relevance, I apply a more subjective measure.  If you have conversion data than you can use that, but if you are starting from scratch you have to go with your gut, and when in doubt lean more towards longer tail keywords and keywords with “buying” intent.

For competition, we again use the Google Keyword Tool and also SEOmoz’s Keyword Difficulty Tool.  Neither is perfect, so we also do a lot of “custom” analysis to determine difficulty levels.  The SEOChat tool is actually pretty good too… it doesn’t look as cool as the others but I tend to agree with its scoring a bit more.

Your thoughts?

I’m chatting with a marketing consultant who is looking to bring me in for the organic SEO piece for a couple of clients of his.  He is going to handle the PPC campaign as well as a few other items more on the branding, messaging side too.  Smart guy, understands lead generation, market segmentation, etc.  He wants to do some keyword research for his client this weekend and has not done much of this before, so he asks me to basically tell him what he needs to do and what tools he should use.

My reply:


It’s tough to teach “how to do keyword research” via email in 5 minutes…  But here are some resources and a rough recommendation of how to go about it:

Step 1: Brainstorm
– your client’s site
– your intuitive list of keywords you are thinking of (no tools, just your brain and imagination)
– your client’s intuitive list
– list all the “services” your client provides
– list the various sub-topics or features related to each service
– talk to sales guys and learn what the customer is looking for
– review your existing Google Analytics data for keywords currently referring traffic

Step 2: Competition
– your client’s competitors sites
– do a few searches for your initial list… check out the sites that rank well… what else are they targeting?
– use tools like www.spyfu.com and www.keycompete.com

Step 3: Organization
By now you should have a decent length list of lots of random keywords with very little structure or organization.  Look at the keywords.  Again, use your brain and no tools right now.  How can you break this list down into 4-5 different categories?  What do the keyword phrases have in common?  How are they different?  What are the common themes? What are the different “parts” of the keyword phrases?

Step 4: Keyword Combination Tool

Identify the various parts of the keyword phrase and break it down…  For example, with kitchen remodeling contractors you could break it down into two groups:

a) kitchen remodeling, kitchen renovations, kitchen remodels
b) contractors, companies, firms, remodelers

Then use the tool to create all the different combinations.  Additionally, you might want to use this to combine your root keywords “kitchen remodeling contractors” and “kitchen remodeling companies” with all of your regions like “DC” and “Maryland”.  Make two lists, and combine them.

Step 5: Google Keyword Tool
Use this to gather search volume data once you have your lists.  This tool will also make recommendations for new keywords, but don’t rely on that.  Do it yourself first via the above steps, and any new words you find here are a bonus.  Otherwise you’ll go down a narrow path and miss lots of stuff.  You need to use the human brain first, then the automated tools.  You can also use other data sources like WordTracker and KeywordDiscovery but for you purposes just use this tool for now.  Use the “average month” number for your keywords and use the “phrase” value rather than the broad or exact (its a dropdown you must select).

Step 6: Sort & Prioritize
Sort the keywords now according to search volume.  Then, starting at the top (highest volume) go through them and assign a number from 1 to 10 that indicates how likely it is that someone who searched on that keyword is a perfect match for your client.  10 means its almost a sure sale.  1 means its not very relevant at all.  I would probably give “kitchen remodeling” a 2 or 3, whereas I’d give “kitchen remodeling contractors” a 4 or 5 and maybe “kitchen remodeling contractors in Washington, DC” an 8 or 9.  Pardon the shameless plugs there, but hey they are totally relevant!

Step 7: Pick Your Targets
For PPC you will probably use most of these and then refine them out based on cost and performance (ROI) over time.  For SEO you can’t practically target a ton of keywords, so you have to pick the ones that are the most worthwhile.  You’ll also take into account 1) competition and 2) your current situation.  If its a new website you’ll need to target “long tail” keywords.  If its already pretty strong with a bunch of inbound links you can go a bit less long-tail.

Also see:  My Updated Opinion on Various Keyword Research Tools

I would typically estimate 4-5 hours per client here.  Its easy though to go much higher or lower.  Also, I would recommend that you probably do the data collection part on your keywords BEFORE adding in the various regional words.  Otherwise you’ll get a bunch of low numbers that are either “no data” which isn’t helpful to you, or they will have actual numbers but be highly prone to skew since its such a small data set.  Thus, run your data collection to identify whether “kitchen remodeling” gets more search volume than “kitchen renovation”, as well as what the magnitude is there, and then make the (relatively safe) assumption that if “kitchen remodeling” gets 3x the search volume of “kitchen renovation” than its probably pretty likely that “kitchen remodeling DC” will get somewhere in the range of 3x the search volume that “kitchen renovation DC” gets.

Okay so that was 15 minutes… helpful???

So over the past 1-2 years most of the keyword research I’ve been doing has been supplemental in nature.  That is to say I did a “major” analysis maybe 2-3 years ago and have just been building on top of that, updating things, etc. rather than starting from scratch.  This is primarily b/c I’ve not taken on a ton of new clients since most of our clients partner with us for ongoing, long-term campaigns.  The main tools I’ve been using lately are the free Google Keyword Tool and the not-free Keyword Discovery by Trellian, along with a heavy dose of data analysis via Google Analytics and any other sources showing actual keyword referrals.

But a couple of days ago I went back to starting from scratch.  We have a client who started with us about a month ago and first tackled some structural and tracking-related issues with the site before really hitting the keyword research formally.  We had some intuitive ideas about keywords, and the client has a good bit of historical data… plus its a segment where the diversity of keywords just doesn’t seem to be that great.  Anyhow, I decided to try out a number of tools that I’ve had accounts with for a while… probably 2-3 years or more.  In some cases we’ve had free trial accounts, and in other cases I’ve been paying for membership even though we haven’t used it much.

I was looking for keyword suggestions and (hopefully) some sort of quantitative pieces, and was sort of doing this with my “competitive analysis” hat on.  That is, I was interested in finding out what is driving traffic to the competitors’ websites, and then see if we are already targeting those keywords or not.

I gave between 5 and 30 minutes to each of the following:

Keyword Discovery – still a pretty good tool, mainly b/c of the size of their data source.  Its still clunky and slow to use and is missing lots of intuitive features that would be very helpful, but the data is pretty good considering most other sites.  Its also ugly.  Hire a designer, guys!

SpyFu – a lot of people bring up this site, but I must be missing it.  They had only a few suggestions for the keywords and competitors I gave them.  Not impressed.

WordZe – nice looking icons on the login screen, and the promise of lots of tools… but the core keyword research function is lacking.  The suggestions and data I get look like junk compared to the better services out there.  This copyblogger review gives it a thumbs up, but to me all they are saying is “its pretty and easier to use”.  Okay, but what about the fact that their data sucks and does not provide a good number of quality keyword suggestions?

KeyCompete – Again, not impressed.  I was using the free version, but that’s only b/c they only seemed to have a few keywords anyway.  Had they shown me something enticing I would have gladly paid for the subscription, but why should I when they only had 2 keywords for each site I searched on?  I got to see up to 5 for free.  The next site had a ton more data.

KeywordSpy – Very impressed.  Lots of great data, and easy to use.  Fairly fast too.  Even had ROI estimates, though I don’t know how much I’d trust those numbers.  Anyhow, good stuff.  I signed up for a monthly subscription.

Google Keyword Tool – You have to use it given the data source, even though everyone believes the numbers are vastly inflated and obviously rounded quite dramatically.  They used to say “high”, “medium” and “low” for volume, so having actual numbers is an improvement even with how rounded they are.  I believe there are like 20 or so different volume levels and they just round to the nearest.  That said, the data is all relative so it still has value.  And given that their data accounts for like 80% of all search data, its hard to ignore this.  I still don’t like the interface though, but this has to be a piece of your keyword research process.

Another tool that helps with keyword research is a keyword combination tool like the one at Ranks.nl.  All it really does is a concatenate function that you can easily do yourself using Microsoft Excel, but it is a little more user friendly and will save you 3 minutes of creating Excel formulas and copying/pasting.

Summary and Key Learnings

KeywordSpy really impressed me.  I signed up for a subscription for like $89 per month or something.  Feeling bad about increasing my monthly expenses, I logged into WordZe.com and gave them a few minutes of my time.  Didn’t find any real value there, so I canceled my $35 per month subscription, thus saving a bit on the monthly commitments.

Question: Should I give Wordtracker another try?  It has to have been like 5 years since I used them.  Also what about Aaron Wall’s tool?  Last time I used that I remember it being almost too much data and not enough “good data”.  Thoughts?