• Paul Bartlett

    Be VERY careful of two things here:

    1. “greatest potential to drive the most traffic and thereby conversion”

    Traffic numbers often have little to do with actual conversions. Run a free contest, and you’ll get loads of traffic, most of it non-converting. A really great way to burn PPC dollars.

    2. Some of the sub 1000 branded keywords can convert magnificently, so don’t necessarily bury them with other poorly performing keywords just based on traffic numbers.

    An effective strategy we’ve been testing lately is to put ALL keywords in to their own ad group, and using fewer keywords overall, especially for smaller businesses. To raise conversions and/or ROI, optimize your landing pages to subscription based services, products, and communications. Ongoing revenue is the key.

  • Mark Kennedy

    Great post. A good portion of SEM work comes with the setup. A good setup will not only help performance, but will save an SEM professional a lot of time in the future.

    Also, I agree with Paul. It really pays to take the tail approach with small campaigns. Yes, it may seem like overkill to be really granular, but for a small budget, it really helps. And in some rare cases a small campaign is better off with just the low volume tails.

    However, for medium to large clients, I like your approach. One thing I like to do is if I see a tail keyword with a poor Q-score in an adgroup like you described, I’ll pull it out and give it it’s own adgroup with a very specific ad. That alone sometimes increases the q-score. So you don’t have to break the adgroup down into 50 little ones, but after some data you can tell which keywords need a little extra attention.

  • Bonnie Schwartz

    I don’t think this post is saying you should NEVER get granular, and I agree with Paul and Mark100% that being Granular is essential for a well managed campaign. The SEER PPC team adheres to that approach 100%.I dont actually think this post is saying that is not true.

    Like Mark said, for big accounts with 1000′s of keywords it just not feasible to target every word in its own group. The 1K impression rule just helps you focus at the start on the terms that will drive volume. Tail terms often get VERY little maybe no traffic so focusing on them in the beginning is not a good way to maximize your time sometimes. If you have 1000 terms that get 1 impression a month each, does putting each of these terms in its own group and writing copy for 1000 terms directly really worth the effort?

    We also use the approach sometimes of putting very competitive terms in their own ad group alone so that we can focus on that term directly to increase the QS

    Once you launch the account, if you notice some tail terms getting volume you may want to break your groups down farther.

    I love Mark’s idea of taking low QS terms and putting them in their own group and targeting them directly. If that does not work you may even want to delete the low QS terms from your account since QS is also held at the account level. However, I sometimes leave the in because I find competitor terms often have a low QS and I want to show up for those terms despite the QS.

    The 1K impression rule also helps when you have an existing large account and you want to look to break your ad groups down. Any term that Google’s Keyword tool says has more than 1K impressions, (which actually is not that many) and actually pretty granular definitely should be targeted directly.

    This advice comes right from the SEER team’s Google’s rep, an optimization specialist at Google and I know personally and for the SEER team as a whole has really helped us optimize our accounts.

  • davinam

    @Paul Bartlett: I agree with you that traffic estimate is not always right in predicting conversions but it gives us information on terms that have potential and received a lot of queries. This information is a good starting point to structure our ad group.

    This method has saved us time and again helped us creating a better ad group structure, especially for clients with high volume search (which something that I should have included in my post). If for some reasons, we find that these terms are not performing well (after launching), we could try a different approach.

    Lastly, thank you for sharing your strategy for the small business. How long have you been implementing this strategy?

    @Mark Kennedy: I like your strategy for medium to large clients! To follow up on Bonnie’s point, are you also implementing this strategy for the competitors’ terms? These terms usually have low quality scores and I’m curious to see the impact of your strategy on their QS.

  • Scott

    This may be a dumb question, but when you say “If a keyword does not have more than 1K impressions determined by Google’s Keyword tool then the keyword does not need its own ad group.“ — do you mean 1k impressions a month? a day? forever since the beginning of your campaign?

  • davinam


    I looked at the local search volume column and this column collects data from the last available month.

    If you click on question mark button next to the column, this is the explanation
    “Local Search Volume: [Last Available Month]
    This column shows the approximate number of search queries matching each keyword result. This statistic applies to searches on Google and the search network in the most recent month that we have data for. It’s specific to your targeted country and language as well as your selection from the Match Type drop-down menu. If we don’t have sufficient data for a particular keyword, you’ll see not enough data. “

  • Mark Kennedy

    @Bonnie and @davinam

    Hi Guys! Looks like it’s been a while since I checked back, but better late than never :)

    When bidding on competitor terms (and some other low q-score but relevant terms), it’s a tough call and case by case. I agree that bidding on a competitor name can really have a strong ROI (although not always a lot of volume) and deliver results despite a low q-score. If that is the case you have to leave it in. And you can’t really re-optimize since you’re not going to put a competitor name in the title of your ad. In this situation it’s a case of if it ain’t broke…

    Then there are those borderline terms that Bonnie is talking about. You give them their own ad, landing page, and love, yet you just can’t get any q-score improvement. In that case you weigh the results (and potential results) vs. the pull it has on your campaign q-score. Then it’s just a case of delete or not delete.

    Again, this was a good post and the comments as well. Keep blogging!

  • davinam


    Thanks for sharing your experience!

    and Yes, we will keep on blogging :)

  • chrisco

    Where does one find an AdWords pro to setup and run Google AdWords, analytics, and website optimizer (or at least AdWords)? Thanks.