• http://www.facebook.com/mr.larry.kim Larry Kim

    Hi David nice article. Have u checked out my free Adwords grader? R u at heroconf?

  • tony

    Hi David – really like the article. In relation to the The Lin-Rodnitzky Ratio do you think this is an overly simplistic view of adwords performance? It doesn’t take into account any assisted conversions or the value of higher funnel generic keywords that drive conversions to brand terms down the line? Turning off and or reducing bids on some of our higher cost keywords which appear to be poor performing in terms of last click conversions, has led to a decrease in the overall ROI of our accounts in the past…

  • http://www.alanmitchell.com.au/ alanmitchell

    Hi David,

    I’m a big fan of techniques which quickly and efficiently highlight areas of strength and weakness in sometime overly-complicated campaigns. One technique I often use is the 10% clicks rule (http://www.calculatemarketing.com/blog/techniques/the-10-percent-clicks-rule/), which allows you to highlight your ad groups which are most likely to benefit from expansion.

    I’m interested in your Lin-Rodnitzky Ratio, as it sounds very good in theory. However, surely it breaks down when you use larger and larger date ranges? For example, if you used a date range of 2 months, you might have an L/R ration of 2.5, but increasing the data range to 1 year, for example, will mean that more and more different searches will start generating conversions (i.e. there will be less skew at the head and more of a long tail), so the L/R ration might drop to 2.0? An even longer date range (i.e. 10 years) will means than more and more of your non-converting searches will start to convert, therefore lowering your L/R ratio closer to 1.0. Is that not right?

    I have tried this with multiple different campaigns, and the longer the date range I select, the more the L/R ratio creeps towards 1.0.


  • http://www.facebook.com/gohawks David Rodnitzky

    Alan, thanks for the comments! The L/R ratio, it turns out, does not break down over long ranges, simply because it is basically a measure of proper account structure, and a bad account structure will continue to be bad over weeks, months, and years!

    For example, let’s say you sell floor mats and you get matched on the word “yoga mats” – which you don’t sell. If you never realize this (which sadly occurs more often than you might think) you could have a high L/R ratio for years and years!

  • http://www.facebook.com/gohawks David Rodnitzky

    Tony, no doubt it is simplistic. It’s just an initial analysis of an account but I agree that it should be the “end all be all” – there is a lot more deep diving that you can do to get better data. Still, I find it a nice litmus test!

  • http://www.facebook.com/gohawks David Rodnitzky

    Yes, and yes (and sorry for the late reply – great seeing you at HeroConf)!

  • http://www.alanmitchell.com.au/ alanmitchell

    That is very true, but over a longer and longer date range, more and more of your unique search queries will start to generate at least 1 conversion, thereby reducing the L/R ratio toward 1.0.

    For example, all it takes is just one rogue conversion from your ‘yoga mats’ search query (which is probably very likely over a large enough time period and tens of thousands of clicks), and the search ‘yoga mats’ will be unfairly added to your Conversions > 0 data set, thereby massively reducing the L/R ratio.

    I think it’s a good rule of thumb for comparing the structure of campaigns within the same account, but due to the conversions > 0 requirement not growing as the amount of data grows, this will naturally create a bias.

    Perhaps a more relative benchmark, such as ordering keywords highest to lowest by the number of conversions, then dividing the spend of keywords which generated the top 20% of conversions, by the spend of keywords which generated the bottom 20% of conversions, would allow the L/R ratio to maintain its relativity with any amount of data and over any date range?

  • Fran Nagaro

    Really good article. But surprised to read about a Rodnitzky ratio and find out at the end of the post that is actually yours. Aren´t that sort of things named by others as a recognition for good work?