Wordy, Your Keyword Potential Pal

As I mentioned in my post on how to improve page 2 rankings, boosting the performance of keywords you already get traffic for can have a great ROI.  But how do you see beyond analytics data overload to figure out which keywords have the most potential?  Here’s one way to do it:

  1. First group your referring keywords by theme.  In Google Analytics use the “containing” filter at the bottom of the Keywords report. (i.e. only show keywords containing the word “kid”).
  2. Use the following formula to determine the potential of a keyword grouping:
    Keyword Potential* = (1-Keyword Volume/Search Engine Referrals)(PV)(Time)(1-Bounce Rate)


  • Keyword Volume = total # of referring words in the keyword group
  • Search Engine Referrals = total # of referrals from search for the keyword group
  • PV = Page Views/visit for the keyword group
  • Time = Time/visit for the keyword group (excel turns x:xx into a numeric value)
  • Bounce Rate = I’ll let you guess this one
  • If your site has the idea of a conversion, you would also multiply the formula by conversion rate.  You might even delete Time and PV as ultimately all you care about is conversion.

The keyword groups with the highest Keyword Potential* score should be the ones you focus on improving rank and expanding variations of the groups’ keyword.

For example:

Site X has two keyword groups: “kid” and “how to” (it’s a mommy type site).  The Keyword Potential* for each keyword group are as follows:

Kid = (1-179/701)(13.12)(6:03)(1-.28) = 1.78

How to = (1-869/1017)(4.21)(3:58)(1-.50) = 0.05

Even though phrases with “how to” in them drive about 50% more referrals (1,017), they drive far fewer referrals per keyword (869/1017) than phrases that contain “kid” (179/701).  The other metrics for “kid” are also much better.  This implies that each additional phrase that contains “kid” should improve the site metrics by a lot compared to phrases that contain “how to”.

This data could be used to argue that Site X should add more content that contains the word “kid” v. “how to” content.   Even better, go for content that targets “how to” do something with your “kid”.  See my post on keyword expansion for more ideas on how to do this.

Update: The data might be skewed if a phrase in a keyword group is overly dominant (e.g. one phrase drives 90% of all of the referrals in a keyword group) so you’ll want to look at the underlying data for each group and not just rely on the formula.

*Keyword Potential* is a trademark of Local SEO Guide as soon as I figure out how to add the “TM” superscript to this article.

Share This Story!

About Author

21 Response Comments

  • David Mihm  June 30, 2008 at 11:29 am

    What an awesome formula, Andrew. This might be the best keyword resarch post I’ve ever read. Maybe a little advanced for small business owners who aren’t SEOs, but for those of us who serve SMB’s, this is perfect.

    Also, for the TM symbol, you can use “™” in the HTML editor of WordPress.

  • David Mihm  June 30, 2008 at 11:30 am

    shoot, that didn’t work. it is


  • Andrew Shotland  June 30, 2008 at 11:33 am

    Thanks for the tip™ David. btw it’s “&-#-0-1-5-3-;”

  • Mark Barrera  June 30, 2008 at 3:09 pm

    Great stuff Andrew! I have never thought about creating a formula like this even though we always do something similar to mine Analytics data for additional terms to target.


  • Brian Carter  June 30, 2008 at 7:23 pm

    Holy Crap Andrew, I love freakin’ equations, and that was cool! Especially equations that use (1-x)

  • Andrew Shotland  June 30, 2008 at 7:27 pm

    I’ve got to say Brian I too am a sucker for 1-x. We should start a group on Facebook or something.

  • Stephan Miller  July 1, 2008 at 9:16 am

    Does your current rank for the keyword in Google factor into this formula? I.E. it is much easier to just from #20 to #12 but the move from #12 to #4 is way more valuable.

  • Andrew Shotland  July 1, 2008 at 9:25 am

    Stephan, you could definitely add the current rank into this formula but it’s pretty much keyword dependent as to how valuable the move upwards is. My rule of thumb is to use an arbitrary cutoff in terms of how much traffic you are getting from the keyword group (e.g. ignore any keyword group that accounts for less than 2% of your traffic) and then look at the ranking potential of different terms within that group.

  • Aidan @ inkode  July 1, 2008 at 7:54 pm

    This is a most excellent formula Andrew, I reckon it would be a sweet idea for a keyword tool. Is that a road you have considered travelling?

  • Andrew Shotland  July 1, 2008 at 8:06 pm

    It has crossed my mind but there are plenty of guys who can whip this out in their sleep. I wouldn’t be surprised if this shows up in Enquisite or another similar service.

    It would be cool if Google Analytics worked with plugins so you could add something like this to it.

  • Adam Green  July 2, 2008 at 6:01 am

    Very useful formula. I have a few clients that would benefit from this. Many thanks for sharing.

  • Jeff R. James  July 2, 2008 at 6:21 am

    If I told someone I was going to spend their money optimizing their site using keyword performance data that doesn’t account for the absolute # of conversions per keyword…they would quickly dispense with me.

    If your clients find this fascinating you must be working with the local pet shop.



  • Andrew Shotland  July 2, 2008 at 7:29 am

    Jeff, thanks for that illuminating comment. Really top notch thinking there.

    Perhaps you missed the update on using conversion rates?

    I guess you must be working with the local ignorant SEO shop.

  • Jeff  July 2, 2008 at 10:19 am

    In all seriousness, Andrew – retracting the ad hominem remark (my apologies), it is top notch thinking.

    “The data might be skewed if a phrase in a keyword group is overly dominant (e.g. one phrase drives 90% of all of the referrals in a keyword group) so you’ll want to look at the underlying data for each group and not just rely on the formula.”

    Regardless of any skew – the fundamental driver of any optimization decision is conversion analysis based on volume (the correlate of which is position), not a formula relying on the product of multiple parameters which may not be good predictors of the future success of a keyword.

    That was my central point.

  • Andrew Shotland  July 2, 2008 at 11:13 am

    Apology accepted Jeff (and not necessary – I don’t mind a little mixing it up here, although seeing that first thing in the morning did get me bit cranky).

    While I agree that conversion is the be all and end all, it’s not the all.

    For example, I work with a lot of media sites and for them conversion = page views and time on site. I also work with e-commerce sites and part of their focus is purchase and part of their focus is new customer acquisition and current customer loyalty. They want you browsing their products even if you are not in buying mode yet as they know that it often takes multiple visits to a site to generate a purchase.

  • Jeff  July 2, 2008 at 4:35 pm

    I can see where it might be useful for a site focused on publishing since success is less clearly defined – and you definitely need some statistic for optimization purposes for both appearances and utility.

    So, with some qualifications it has applications.

    I guess we’d agree that the higher the visits/conversion ratio is, the more useful this might be, as it’s never clear when the originally arrived, through what keyword and on what page. For direct response, hit or miss type sites though, not so much.



  • Andrew Shotland  July 2, 2008 at 7:51 pm

    This is where I see a lot of sites, particularly ecommerce sites, mess up. Sites should be cookieing the user on first touch with a permanent cookie. The cookie should record time/date of visit, referring source and referring keyword if applicable. When the cookied user converts then that data gets written to the CRM database and the site now starts to understand how SEO, or any other marketing tactic, contributed to the conversion. Too often I see 30-day cookies that don’t collect relevant data and disappear or are easily overwritten.

  • Paul Burani, Clicksharp Marketing  July 7, 2008 at 5:42 am

    Thanks Andrew, I’m looking forward to taking your formula for a spin. FYI, I’ve found that one of my most powerful metrics is a more rudimentary stat which is (unique visits from all keyword phrases containing the one core term)/(all unique visits from organic search). This has been very helpful in illustrating that, even though the traffic level from organic search has stayed the same, the efficacy has evolved over time.

  • Shubham  February 11, 2010 at 6:09 am

    thanks man.! You are my guru.. 😉

  • Dee  April 22, 2010 at 1:09 am

    Not sure how to use the formula. Do you paste it into Excel? If so, what’s the exact syntax for that?