• http://searchengineland.com Bradd Libby

    “For example, if you could just get your click-thru-rate from 2% to 2.1% on a keyword that drives sales for you, that would mean one thousand more revenue-driving customers for every million impressions you run.”


    Unfortunately, differences in clickthrough rates (or conversion rates) that are this small are extremely difficult to see. Jen Syverud of the Rimm-Kaufman Group has a great blog post on this topic Recognizing Signal & Noise in Paid Search.

    Download her spreadsheet and have a look at the variance in conversion rates for 1000 clicks. (The same principle holds for your example with impressions and CTR, though the specific numbers would be different.)

    If I have 2 ads, one of which has a 2.0% CTR and the other a 2.1% CTR and want to be 99% confident which is which, believe it or not, it will take more than 1 million impressions (evenly divided between them) to tell them apart.

    You have a great list of tips here, but I’d say “Above all else, get to know math better than anyone else you know (or team up with someone who does).” Doing otherwise could be expensive and painful.

  • Josh Dreller

    That’s a good point, Bradd. The key takeaway for that tip was that over time, results sometimes plateau after deep optimization. I want to let beginners know this may happen and to not get discouraged because even tiny lifts will work together to continue towards positive gains. Winning at paid search means CONTINUAL testing and optimization even if you’re not getting 20% lifts anymore. It’s just a pain point I’ve come across with junior SEMers and I wanted to address it before this column really starts cranking up with the PPC creation and management.

    That’s a good article by Jen. I just downloaded the spreadsheet and I’m playing with the numbers now.