Optimizing Your AdWords Campaigns

SMS Text

Pay per click advertising is a form of marketing with major benefits. It differs from other methods of marketing in its ability to allow advertisers and marketers to measure and analyze the results of their techniques, to refine and optimize them to increase ROI beyond almost anything possible in the offline marketing world. Google AdWords is currently the prominent pay per click resource available today. Learning some of the ways to optimize your AdWords campaigns can cut costs while keeping conversions up in several ways.

Here are four ways to refine your AdWords campaign to see where money spent is producing the best return and to adjust settings to maximize return on investment.

1. The Segment option. This is one of the newest additions to AdWords. The Segment option is available within your Campaigns, Ad Groups and Keywords tabs. Within each, the options are different, but allow for viewing of your results in specific ways of measurement. This can include breaking into specific units of time, such as weeks or days of the week.

By viewing your data this way you can discover what weeks are most active for a seasonal market, as well as information such as which days of the week are most prone to getting better conversion rates. By knowing this data, you can reduce costs where your conversions are more expensive, increase bids where you are seeing better conversion rates, and in all these ways increase ROI.

2. A-B Split Testing for ads. This can be one of the most effective means of increasing CTR for your ad groups. Have two or three ads active for each ad group. After a number of clicks (ideally at least 20 clicks on each ad), choose the better performing ad. Replace the more poorly performing ad with a different ad copy. Repeat the whole operation to find which advertisement will produce the best CTR or the best conversion rate.

Doing this can often improve your CTR by a factor of 2 or more, and a higher CTR improves Quality Score as well as increasing lead generation. To get an accurate reading for proper A-B split testing, make sure your campaign setting for Ad Delivery->Rotation is set to “Rotate: Show ads more evenly”.

3. Remarketing with the Audiences tab. This is an element of AdWords that works only for the Content Network. When the Audiences tab is enabled, you can create remarketing lists to target specific visitors to your site. By placing code on specific pages within your site, a cookie is put on all users who visit these pages.

Then when they visit other sites with similar targeted themes, a very relevant ad is shown to these users to give them motivation to return to your site to make a purchase or opt-in. This works by reaching out to visitors who had visited your site previously and displayed some kind of interest in that market. Often this remarketing approach can increase conversions by contacting people who have an established interest in your products and reconnecting with them.

4. Changing advertising frequency/rate settings. Advertising can be adjusted in AdWords to not just advertise for certain days and hours of each day, but the amount for CPC can be adjusted to raise or lower bidding rates for certain days and hours. The setting is within each Campaign settings area, under Advanced settings->Schedule and choosing Ad Scheduling, the selecting “Bid Adjustment” inside the scheduling option.

After doing research for performance for specific days within your campaign you can make changes to cut costs and increase ROI nicely. For example, after viewing that you are getting little or no conversions on certain days of the week, you can pause these days with this setting. In the same way, any days that have better conversion rates can have an increased bid amount to maximize CTR for those days. This can all help improve business results for your AdWords campaign.

By knowing about and using these different options within AdWords you can greatly improve your pay per click results. AdWords optimization is a definite way to increase ROI and open additional opportunities to expand your marketing methods. Properly using these techniques can cut costs, increase profits, and help you make the most out of this powerful tool.

Eric Gesinski
Eric Gesinski runs Tulsa Marketing Online doing professional web design and SEO work in addition to doing business pay per click and internet marketing projects.
Eric Gesinski

Latest posts by Eric Gesinski (see all)

Get the latest news from Search Engine Journal!
We value your privacy! See our policy here.
  • Mark Thompson

    Good points, I really like some of the new features that Google Adwords has been releasing. It seems like everyday there is a new feature that I never knew about.

    One thing I did want to mention that is possibly overlooked, it having a sound account structure. One thing I will do is stand in front of a whiteboard and map out the entire account (Campaigns, Ad Groups), this really helps to organize your thoughts and keep each “segment” or “theme” targeted to very specific keywords.

    I think Adwords will be putting a lot of Bid Management Tools out of business, because of all of the new features it keeps adding. I would imagine that will start to add automated features like the BMT do. For example if a keyword does not perform at a certain CTR or Conversion Rate, it will be paused.

    One thing that I wonder about is how Google will come up with new ways to make money on advertising. At some point PPC will get too expensive that only companies will large pockets can offered to keep running campaigns (which low CPC was the reason new advertisers were coming on board). We are already starting to see minimum CPC through the roof.

    Great post Eric.

  • andykuiper

    Very helpful article πŸ™‚
    With regard to # 2. “A-B Split Testing for ads.” I would add that when creating a new ad (B) one creates a clone of the existing ad (A). So you would now have the existing ad (A), a clone of the existing ad (A1), and your new ad (B). The reason for the clone is to truly compare A and B one would really need to compare A1 and B… because A has earned 'history' with Google AdWords and thus automatically receives preferential treatment. So the real comparison should be between A1 (no history) and B.
    Andy πŸ™‚

    • Eric Gesinski

      Good point to clear that up. If I'm correct, the “preferential treatment” you're talking about it more about showing the ad more frequently. Changing the Ad Delivery setting to the Rotate option will help curb that a bit, but I've seen that even in that case it may still “prefer” the existing ad. However, Quality Score is only applied to keywords and not ads, so I don't think actual position for the ad is adjusted, even with “preferential treatment”. And basing your ad quality purely on CTR (and/or conversions) for split testing this can still function with an established ad with history.

      I do think I may have to test your tip, though, to see what kind of differences show up when leaving ad A alone or cloning it to do a more “raw” A-B split test. Thanks for the feedback!

  • andykuiper

    Hi Eric πŸ™‚
    Her's an elaboration of what I was getting at. If there are flaws in my assumptions or logic, or whatever; I'll be the first to apologize and learn — I'm surely no expert in all of this πŸ˜‰

    I was assuming one followed your good advice and changed the default away from 'rotate' (it's amazing how this step is missed by so many, usually to their detriment).

    If one doesn't change the default to 'rotate' then yes, part of the ad's preferential treatment will be more display frequency (assuming the QS relationship between the keyword and ad is higher than other ads).

    My belief is you're right Eric, in that preferential treatment will still exist for the established rotated ad (your testing should confirm this).

    My understanding is that a KW's QS is derived by a number of relationships, including the relationship between an ad group's ads CTR 'as an average', as well as a 'particular' ad's CTR, and a KW. In the case of A/B testing the KW/Ad QS relationship is reflected differently in two important areas:

    1) The KW minimum bid, which is based on:
    an 'average' of all the active ads historical click through rate in an ad group including the factors (–) outlined below.

    2) The Ad's Ranking (which of course effects it's future CTR) which is based on: The relevance of the keyword and the “matched ad” (a quote from Google AdWords Quality Score Formula) to the search query… so the “matched ad” in my example would be ad (A). Ad (A) is the ad that has a historical CTR. The new ad (B) has no historical CTR between it and a keyword.

    So the introduction of a new ad (B) compared to an existing ad (A) would, if (A) has a good CTR history, not be fair… the existing ad (A) has the edge (assuming equal bids) in that its QS relationship to the KW would allow it to rank higher and thus 'more likely' receive a higher CTR than ad (B). Thus my suggestion to include a clone of ad (A) which would have no QS relationship with a KW and thus would be competing on even terms with the new ad (B).

    According to Google a Keywords QS is based on:

    — account history; measured by the CTR of all the ads “and” keywords in an account
    — the historical CTR of the display URL's in the ad group
    — the quality of the landing page
    — the relevance of the keyword to the ads in its ad group
    — the relevance of the keyword and the matched ad to the search query
    — the account's performance in the geographical region where the ad is shown
    — and some other 'undisclosed' factors
    Andy πŸ™‚

    • Eric Gesinski

      Awesome reply! From my understanding (and I'm always open to correction, if I find established proof otherwise), the Quality Score is given to keywords individually. Every element you listed is part of the QS, and as you also listed, that does not include everything. Since it's not publicly disclosed, there's no way to know for sure the ideal way to optimize QS.

      That being said, I'm interested in to find out how ads are affected. I have as of yet not seen any solid indication that ads are given the same or even a similar Quality Score ranking as the keywords are, but are independently “ranked” based on the relevance to the keywords they are shown for and (possibly) the CTR for each ad.

      This ad ranking would be for each ad individually. My question is then on whether or not the ranking is grown or is instantaneously given to an ad for each search query it is shown for purely based on keyword relevance. If it's grown, then your suggestion would be far more accurate to do true split testing, so that the CTR can be compared simultaneously for both ads from an equal ground. If it's instant, then any ad that shows better keyword relevance will immediately get a boost in position, but CTR will not make any difference in that.

      At any rate, I do think for the purpose of split testing, we're splitting hairs here. Even IF the ad ranking is a grown one and someone were to split test with an established ad versus a brand new one, a solidly better ad will still consistently show better performance on average, even if the ads are not always shown in the exact same position. And conversion rates are very easy to judge even ignoring CTR differences.

      As it is, for people who really want a TRUE split test, I'd agree completely with your recommendation. My only concern would be the pause in the entire ad group while Google sets both ads to “pending” for approval, which can be quick but can sometimes last a while. For people that have a large amount of traffic, even a day's break in advertising can be a huge difference. For that reason alone some advertisers may NOT want to clone the “A” ad, unless they do it without removing the original “A” ad.

      It's good to be able to talk about this all to someone who's put as much thought into it as I have – I know there aren't too many people who enjoy analyzing minute details like this to the utmost degree. I'd love to hear more about whether or not there is a true ad ranking that can be grown, and if so, what elements are part of that.

  • andykuiper

    As I see it… πŸ™‚
    “This ad ranking would be for each ad individually.”
    — Exactly, each ad has CTR history tied in with individual keywords. —
    “My question is then on whether or not the ranking is grown or is instantaneously given to an ad for each search query it is shown for purely based on keyword relevance.”
    — The ranking grows (acquires history) in relation to each KW, per impressions and CTR. When a new ad (B) is introduced, there is no history (no impressions or CTR) that's why I suggest the introduction of the clone (A1) at the same time as the introduction of the new ad (B) as neither has acquired any history —
    “…a solidly better ad will still consistently show better performance on average, even if the ads are not always shown in the exact same position”
    — One would hope so, but if ad (A) has a significantly positive history (which is often the case) it has a distinct advantage; making a fair comparison difficult (initially) unless (B) is a remarkably better performer. The larger the number of impressions for (B) the easier it is to start teasing apart if it's a better or worse performer. Introducing ad (A1) at the same time as ad (B) leaves no doubt as to which is the better performer.
    “And conversion rates are very easy to judge even ignoring CTR differences.”
    — I agree with you 100% on this one, it's all about conversions; the ad with the best CTR may not be one's best “performing” ad by a long shot (even though Google would often have you believe otherwise) —
    “My only concern would be the pause in the entire ad group while Google sets both ads to “pending” for approval”
    — I don't pause ad (A). I just introduce (A1) and (B) at the same time. Ad (A) keeps running throughout. Ads (A1) and (B) get approved and start rotating along with ad (A). IF ad (B) is a dud (compared to ad A1), I pause ad (B) and create a new ad (B) – I pause it rather than change it as a change will eliminate its history, as ANY change to an ad clears it's history. I pause ad (1A) to retain it's limited history for the benefit of the ad group and the account; I don't delete ads as deleting the ad deletes ALL its history. I then start a new clone ad (1A) for the new (B) ad and the process starts all over again.

    When I finally get a new ad (B) performing better than ad (A1) I pause (A1) as it isn't needed anymore. Now I have the new (good performing) ad (B) and the original (always running) ad (A) going. The thing I then assess is whether to pause Ad (A) (retaining it's history for the ad group and account) and introduce a new ad (B) to compete with the new ad (which was B but is now A).

    * A new ad (B) may outperform the clone (A1) so it's a keeper… but it may take a while (sometimes a long while) for it to outperform the sometimes positive history laden ad (A). So I usually leave them both run until (B) is clearly outperforming (A) than I pause (A) and the (B) ad becomes the new (A) ad… and the A/B split testing can start all over again.

    This kind of thing may not be the kind of thing some folks want to spend time on (I don't anymore – I focus on SEO now), but I spent close to 500k on AdWords over a couple of years and had to get pretty good at tweaking things.

    Andy πŸ™‚

  • Laurent

    Hi all,

    Hi Eric,
    Good points you mention in this post, however I do not agree with your 2nd point (A-B Split Testing for ads) when you say that 20 clicks are enough to take a decision about which ads to keep and remove. I don't think this is enough to be statistically significant. Personally, I wait for 200 clicks at least to make sure I've made the right decision.


    • Eric Gesinski


      I said that 20 clicks are a *minimum*. For a lot of people they prefer more to get a more accurate test. However, some campaigns, while effective, don't get a lot of traffic, so waiting for 200 clicks before deciding could take several months. I feel that 20 is the least amount you can get any real concept of which performs better, although more is better. I think this is really just a personal taste – it's market testing, and there's no objectively definitive number to it. But you make a good point – more clicks will generally provide more accurate results. It just comes down to where you want to draw that line.


      • Laurent

        I get your point Eric, but I'm just saying that personnally, I would not take any decision with only 20 clicks. I've made a lot of tests yet, and I can assure you that results are pretty “unsteady” (depending on competition, keywords position, weather, landing page, site performance…).
        But I agree with your argument about the time to get 200 clicks. So the question is: do you prefer to test quickly but with a poor significancy or do you accept to take time to make sure you do the right choice?

        A solution for the rich ones ;-): increase CPCs to be higher in position and get more traffic SPEND MORE for the launch to SPEND LESS later.


  • jostumpner

    Initially, I was surprised by the suggestion of only 20 clicks on an ad to determine performance. However, I would have to agree in a few instances. What if you received 5,000 impressions and only 20 clicks? This is certainly enough info for me to know this ad is performing poorly. I think the reference to the number of clicks should be shifted to impressions. The threshold for what is acceptable will vary depending upon your strategy and goals.

  • electronic cigarette

    Really cool!

  • electronic cigarette

    Really cool!