Highlights of SES “Advanced Paid Search Techniques” Session

I went to SES but I missed the “Advanced Paid Search Techniques” session. Fortunately, Keri Morgret of Strike Models was kind enough to live blog the session – she took great notes. Some of the points I thought deserved to be highlighted.  If you’d like to refer back to Keri’s original notes, you can find them here.  The numbers represent the time that the comments were made (remember – it was live blogged), which may make it easier for you to refer back to the original notes.

10:48  The AdWords Search Query report is much, much improved. It shows a ton of search queries for which there may have been only 1 impression and 1 click. Although it may be true that other analytics data will be more granular, I think search query report data will give most folks 90-95% of what they need.
10:53  When using the AdWords search query report, focus on cost to figure out which search queries you may want to add as negative keywords. Queries with a lot of click spend and no conversions are likely candidates. Search queries that received conversions at a target CPA or better should be added as keywords if the search query type is not ‘exact.’ Bid your new keywords using cost/conv and avg. CPC data from each specific search query.
10:58  Set ad-serving to Rotate! This is very important and it’s often over-looked. CTR very often has an inverse relationship to conversion rates, so you need to make sure that all ads are rotated equally regardless of CTR so that you can figure out which wins in terms of conversion data (cost/conv or conversions per thousand impressions or whatever).
11:02  When doing ad optimization it’s very important to make sure that all ads you are evaluating got served up the same number of impressions. Otherwise you may jump to the wrong conclusions when deciding on champion ads.
11:06  I love that someone talked about how to make your long tail shorter. Yes, you can have too many keywords in your account, and this can have an adverse effect on your account performance in terms of 1) it can drag down overall account quality scores; and 2) it can add extra, unneeded “weight” to your account, which can make daily management less efficient (e.g. latency with AE and web interface). Focus on the long tail *that users actually search on* and the long tail keywords *that actually get conversions* and you’ll stay ahead of the game.
11:06  Start simple, and go where the data takes you. This will save you a lot of time, which means you’ll get to better-performing PPC accounts more quickly.
11:38  Yes yes, LP testing is definitely for business reasons – don’t do it in order to raise quality scores.

10:48  The AdWords Search Query report is much, much improved. It shows a ton of search queries for which there may have been only 1 impression and 1 click. Although it may be true that other analytics data will be more granular, I think search query report data will give most folks 90-95% of what they need.

10:53  When using the AdWords search query report, focus on cost to figure out which search queries you may want to add as negative keywords. Queries with a lot of click spend and no conversions are likely candidates. Search queries that received conversions at a target CPA or better should be added as keywords if the search query type is not ‘exact.’ Bid your new keywords using cost/conv and avg. CPC data from each specific search query.

10:58  Set ad-serving to Rotate! This is very important and it’s often over-looked. CTR very often has an inverse relationship to conversion rates, so you need to make sure that all ads are rotated equally regardless of CTR so that you can figure out which wins in terms of conversion data (cost/conv or conversions per thousand impressions or whatever).

11:02  When doing ad optimization it’s very important to make sure that all ads you are evaluating got served up the same number of impressions. Otherwise you may jump to the wrong conclusions when deciding on champion ads.

11:06  I love that someone talked about how to make your long tail shorter. Yes, you can have too many keywords in your account, and this can have an adverse effect on your account performance in terms of 1) it can drag down overall account quality scores; and 2) it can add extra, unneeded “weight” to your account, which can make daily management less efficient (e.g. latency with AE and web interface). Focus on the long tail *that users actually search on* and the long tail keywords *that actually get conversions* and you’ll stay ahead of the game.  As an anecdote, we once took over a paid search program from a large agency and within one month we increased the number of unique keywords (excluding match types) that were contributing 1 conversion or more from 351 to over 735, an increase of 100%+. We did this while simultaneously decreasing overall keyword count by 85%, from over 200,000 keywords to about 15,000. It’s not the number of keywords in the account – it’s the number of converting keywords you have in your account that really matters.

11:06  Start simple, and go where the data takes you. This will save you a lot of time, which means you’ll get to better-performing PPC accounts more quickly.

11:38  Yes yes, LP testing is definitely for business reasons – don’t do it in order to raise quality scores.

  • Great post, thanks for highlighting some of the important points from the paid search session.

    I completely agree with the 11:06 point – that an account can often have too many long-tails.Search marketers definitely need to find a balance between having enough long-tail keywords and being able to manage them effectively, but in my opinion this generally comes down to experience, knowing what is likely to be searches and what is a waste of time.

    Also agree with the ‘start simple’ point. Starting with only a hundred or so highly relevant ad groups then expanding once search query data has been collected, is a much better approach that trying to start with thousands of keywords for every possible search under the sun.

Shares