Brad Geddes / PPC Geek
Official AdWords Seminar Leader.
Author of Advanced Google AdWords.
Co-Founder, Adalysis.
(312) 884-9017
Brad Geddes's Theories on Marketing 2 Little Known Ways To Increasing CTR And Quality Score

2 Little Known Ways To Increasing CTR And Quality Score


This is a guest post by Chris Thunder, who just launched a platform that helps Adwords advertisers improve their Quality Scores for cheaper Google Adwords traffic. Check it out at Tenscores.com. Follow him on twitter and see if Tenscores.com can be useful to you.
Overwhelmed Advertiser

With the amount of work required for building and managing PPC campaigns, it's easy to forget the importance of optimized targeting.

If I ask you what you need to do in order to increase your ads’ click-through-rates (CTR) for higher quality scores, your immediate answer might be: write a better ad.

And you’d be right.

But it’s not the only way and too often advertisers rely on ad spit-testing when trying to achieve a higher CTR.

If you’ve read a little about quality score (QS), you know that the biggest factor that influences it is CTR. When we plan about increasing CTR, new ads in more focused ad groups come to mind, sometimes we even remember to include and keep expanding a good list of negative keywords (although we know that that has no direct impact on QS).

What we often forget is that ads have varying performances depending on the hour of the day or the geographical region in which they are displayed. Two things that Google takes into account when calculating quality score but are rarely talked about.

If you’ve hit a brick wall and can’t improve your CTR no matter how hard you try, these are the 2 targeting settings that might get you the results you want.

Day-Part Targeting

Day Part Tageting

Ad performance can vary greatly depending on the hour of the day. Although the data above has not reached statistical significance, it may suggest 9pm (21h00) to be bad hour.

You may be getting lots of impressions at 2 in the morning but very little clicks and conversions without even knowing it. If that’s the case, it would be very beneficial to prevent your ads from showing at those times of the day.

To find out how your ads are performing by the hour of the day:

  1. Choose a campaign in your adwords account.
  2. Click on  the “dimensions” tab (or make the dimensions tab viewable in the sub menu).
  3. Click the “view” sub-menu and select “hour of day”.

You should now be able to see at what times of the day your ad group is under performing.

The next thing to do is to exclude your ads from showing at those times  by changing your campaigns settings: Advanced Settings > Ad Scheduling. Make sure your data has reached statistical significance.

When you do that, the average CTR recorded for your keywords will start to increase and your quality scores will slowly rise.

Geo-Targeting

Geo targeting

Ad performance may also vary by specific geographical regions. Notice the difference between Texas and Ohio.

If you’re like most advertisers, your adwords campaigns are probably set to show on the whole territory of your chosen country right now (or even many countries at once). If you run a geographic report in adwords, you will find that your ads perform better in some specific regions than others.

You need to find the regions where your ads perform really poorly and exclude your ads from showing there.

To find out how your ads are performing by specific regions in the country you’re targeting:

  1. Choose a campaign or an ad group in your adwords account.
  2. Go to the “dimensions” tab once again.
  3. Click on the “view” sub-menu and select “Geographic”.

Now you can see the specific regions where you might be receiving lots of impressions but low clicks and conversions compared to others.

Exclude those areas in your campaigns settings: Under Locations and Languages>Locations click edit.

A window will pop up, look for the Exclude areas within selected locations link at the bottom of that window and exclude the areas that are poorly performing for you.

If you don’t want to completely exclude those areas, you might consider creating a separate campaign for them.

When you do that, your average CTR will also slowly rise since it won’t be affected by low performing locations and it will result in better scores.

Those are two quick but very rewarding actions that you can take right now to achieve higher CTR’s and better scores.

In case you’re wondering how effective these two little techniques can be, here’s what Brad confided to me in email conversation (don’t tell him that I told you):

“I can’t agree more with your two points – I do it all the time – in fact I have an account that just by changing the geo settings and splitting up the ads by their geo CTRs, the accounts CTR almost doubled.” ~ Brad Geddes

If you want to track how these actions affect your keyword’s quality scores and your first page minimum bids, I’d like to invite to you to try our Quality Score tracking tool… sign up here and I’ll send you a quick email the next time we open doors.

Warning

  • Before excluding your ads from showing at a certain time of day, make sure you have enough data to make the right decision. Use splittester.com to see if your data has reached statistical significance.
  • DO NOT use CTR as your only measure of performance. Google uses CTR to increase or lower your scores and as a result, increase or lower your CPCs… however, it is not a measure of your profitability. Learn about Profit Per Impression and use it as the final verdict.

Opinions expressed in the article are those of the guest author and not necessarily bgTheory. If you would like to write for Certified Knowledge, please let us know.

10 Comments

  1. rnasty
    June 14, 2011 at 11:21 am · Reply

    Excellent post. I haven’t paid much attention to segmenting by hour before. I examined it today and found out that my ads are having worse CTR few hours in a row (statistically valid data). But during those few hours, they also have lower avg. pos. which brings their CTR inline with the rest of the day.

    In geo report I found big discrepancies in CTR, but avg. pos. column is empty, so CTR data is pretty useless in that report. I checked in several campaigns and several accounts: in geo report avg.pos column is empty. Do you maybe know some other way how to get that data together with the avg. pos.?

  2. Chris
    June 14, 2011 at 4:57 pm · Reply

    It would be interesting to know whether the low positions are causing the low CTRs or whether it’s the low CTRs that have earned you low avg. pos.

    Personally, average position is a metric I rarely look at – if ever – simply because it can be very misleading. I don’t know where you can find it for geo reports.

    If I were you, I would simply test it out: exclude the hours of day and geo locations of low performance if the data suggests it… then let it run for a while and see what happens.

  3. davidzhawk
    June 14, 2011 at 6:48 pm · Reply

    I think these tips will reduce your reported Quality Score and reported CTR, but won’t do anything in terms of actually increase your Revenue Rank, which is how Google determines how much you actually have to pay for a click.

    Google operates AdWords as a real-time auction that considers numerous factors when determining who is going to show up and how much they are going to pay. Among those factors are geography and time-of-day. Thus, if Advertiser A has an expected CTR of 5% in Georgia at 4pm, and Advertiser B has an expected CTR of 4% at the same time in the same place, Advertiser A will pay a lower CPC for a click, and/or will show up higher in the results.

    So if you went and paused your ads in geo’s or times with low CTR, the only thing you are really doing is eliminating an opportunity to show up at a particular time or place. Your QS at a different time in a different location won’t be impacted.

    Seems to me that you should only pause ads based on your KPIs, like CPA or ROAS, not based on QS factors.

  4. rnasty
    June 15, 2011 at 4:58 am · Reply

    Chris, I understand that avg. pos. is average so at low traffic volume it doesn’t mean much, but still, every time we observe CTR we should take into account avg. pos. If at 3.00 PM you have 3% CTR at avg. pos. 4 and at 8.00 PM you have 4% CTR at avg. pos. 3 you can assume your ads are performing equally good (regarding CTR) at both times.

    David, yes, pausing lower performing hours and geos might not be so wise, but taking them into separate campaign, as suggested in the article, will enable you place higher bids in better performing areas and earn more in the end.

  5. Chris
    June 15, 2011 at 2:07 pm · Reply

    Under the assumption that Google’s normalization algorithm (CTR by position) is perfect – then yes, I absolutely agree Rnasty. But I’m quite skeptical about how well CTR for QS is normalized by position. (And I’m not the only one, read the last comments of Craig’s post:). The assumption you make may lead you into dismissing an optimization effort that could impact your ROI. I would say: simply test it out regardless of av. pos.

    David, you are right. It won’t affect the individual ad auctions triggered by search queries in specific locations – for terms that have already built history in your account. But it will impact your overall history at many levels (account, domain/url, keyword) which in turn impacts any new search query auction you’re eligible to participate in. Hope that makes sense. Thanks for pointing it out.

  6. bloomarty
    June 16, 2011 at 7:38 am · Reply

    Chris,
    This is an interesting idea, but I think the advice is potentially harmful. It’s complicated…

    No doubt, CTR is the biggest factor in quality score and increasing the average probably won’t hurt your QS. But there is still a lot of uncertainty if this could potentially help increase QS. First, there is the issue that rnasty brought up: position is probably the biggest factor in CTR. You are right, normalization for position effects isn’t perfect, but that doesn’t mean that it can be ignored.

    Then there’s the issue that David pointed out: Google considers a lot of factors when weighting CTR, including location and maybe time of day. These considerations are far from perfect as well, but if the data we have is sufficient to be statistically significant, Google should have even better data to work with.

    Speaking of statistical significance: You wrote to make sure that data is statistically significant and you linked to an A/B split testing tool. But this tool just addresses the question whether one entity (an ad, a location, a time of day, etc.) is performing better than another one. So let’s say you have a CTR of 3.5% at 1pm and 3.4% at 2pm, and the tool tells you it’s statistically significant. So should you pause the ad at 2pm? And if the answer is yes, how far should you go? There are 24 different hours in the day and you might find that many of them have a significantly worse CTR’s compared to the best one.

    You also mentionend that one could simply test it. I think you can’t do that: there is simply no good measure for success here. You can look at visible quality scores for your keywords, but those values represent the keyword for all times of day and all locations. Therefore visible quality scores are some sort of an average value, just like a keyword’s CTR.

    So let’s say CTR for Texas was bad and you exclude the state from your campaigns. Afterwards your keyword quality scores rise. What does that tell you? If you had a bad quality score for Texas and that bad score doesn’t factor into the average quality score anymore, shouldn’t those quality scores rise anyway? You can’t conclude that your quality score for, say, California benefits from this because you have simply no way to find out.

    To be a little more constructive, here’s what I would do: If you see that some location or some time of day produces different results, try to find out what causes this. For example, could it be that there is a strong competitor in Texas who is usually above your ad? Use the ad preview tool to find out what the ad landscape looks like in Texas. If you find a reason, see if you can do something about it. And if you can’t, either dig deeper or leave it alone, depending on how important the issue is.

    Before you do anything like excluding times of day or locations, you should always think about what it means for your business (or your client’s business). If, for example, advertising in Texas is profitable, you should continue to do so. If your budget is limited and yields higher profits elsewhere, then by all means, exclude the place. But don’t let some unverified theory on some obscure number no one fully understands get in the way of your business goals.

    The bottom line is we don’t know whether the strategy proposed here has merit. I don’t think it does and I fear it might mislead you into polishing averages. You might even be pleased with the results because averages look better even though you’ve hurt your business goals. The decision where and when to advertise should always be made based on hard facts.

    So long

    Martin

    • Chris
      June 18, 2011 at 12:53 am · Reply

      Hi Martin,

      Thank you for joining the conversation. I’m delighted to have a meaningful dialogue on the subject when time permits so please keep on sharing your thoughts.

      Reading your response, I realize that I might have sacrificed some important details in favor of brevity.

      What I’m suggesting in this post is not a theory. It’s not some kind of hypothesis I dreamed up and decided to share so other people could test and see if it works. Brad would have never allowed it to be posted had that been the case. It’s one of the segmenting procedures I use regularly, and one that has worked quite well for some of my clients.

      But I did not stress enough when to use it and when not to use it. It will not make sense for all campaigns in all markets, and it is not meant to be obsessed about with little variations. What you want are unusual variations that can tell you something important about your market.

      Most of us already do this when optimizing conversions (I hope you do too) and I know most people don’t know they can use it to optimize CTR a little farther.

      If you have a campaign that’s already working well in terms of CTR, you’d rather only look at your ROI only. On the other hand, if you’re struggling with low CTR (like lots of people do) this CTR shaping process will certainly help – especially if your competitors are already segmenting by location and times or times of day. But it has to make sense of course, a 3.4% to 3.5% variation is not good enough to make any kind of decision.

      And yes it does make sense to exclude sometimes, other times it’s much better to move to a different campaign with a localized approach. The take away of this article is that there are more ways than one to optimize targeting settings for better performance, some ways are not obvious.

      I think the biggest mistake to make would be to dismiss adding a new element in your toolbox because you think it’s an unverified theory. Only testing can tell you if it works for you or not. It may work for you, it may not. It has worked for me, sometimes it hasn’t. It has worked for Brad and I’m pretty sure not always for him either.

      If we agree that CTR drives QS, then we can agree that this can be verified by observing a CTR lift after making the targeting changes I propose. If QS is so obscure, then we can put it out of the equation and agree that any increase in CTR that doesn’t impact ROI negatively is indeed a good thing for any campaign (regardless of QS).

      Hope that helps. I still have a little bit of time, I’ll try to answer Rnasty as well.

      Thank you for your contribution, it is greatly appreciated.

      Chris

  7. rnasty
    June 16, 2011 at 12:43 pm · Reply

    After reading this post, I’ve spent some time exploring hourly and geo reports. I should have done that before because those reports should be very useful.

    But now, after exploring geo report more in depth, I concluded it’s useless. If someone can prove me wrong, that would be great because then geo report would be actionable.

    Here are my findings:

    On this link you can find Google spreadsheet with geo data from one campaign I’m running:
    https://spreadsheets.google.com/spreadsheet/ccc?key=0AllylNM0V9bcdHBNalRNdTllcG5xNWdySnBOY0dXY3c&hl=en_US&authkey=CMyArsQN

    First sheet is titled “Metro area vs outside metro”
    If you scroll to the row 341 you can see that:
    30.31% CTR is reported for Metro areas
    0.22% CTR is reported for areas not reported under Metro areas

    That huge difference is telling us something is wrong with how Google collects geo data and/or reports it. And if one part of the data is skewed, how reliable the rest of the data can be?

    I decided to test that and split that campaign into two campaigns. One targets only metro areas and another one targets US with all metro areas excluded.
    First results of that test are confirming my thoughts, geo data is not accuarate, both campaigns are receiving same CTR.

    If you take a look at the second sheet, on the link above, titled “Cities vs all other cities” and scroll to the row 465 you’ll see that:
    22.43% CTR is reported for all reported cities
    0.22% CTR is reported for “all other cities”

    Again, the difference is too big. It is obvious data is not accurate. If geo location is not properly reported, how can I believe that in Atlanta I get 10% CTR and in Albany I get 39% CTR? Probably, for Atlanta I just got more accurate reporting.

    If you conduct more detailed analysis of data in those tables, you’ll see that, generally, the bigger the city is, the lower CTR is.
    That is possible, but due to the huge differences and little test with metro areas, I would say that the bigger the city is, the more accurate data is reported for it.

    It would be great if I am wrong, because in that case geo reports are actually actionable.

    • bloomarty
      June 21, 2011 at 9:11 am · Reply

      Very interesting find, rnasty… I’ve analyzed this for an account running in Germany and Austria and I basically found the same phenomenon. First I thought that it might have something to do with the search network, but I found that the problem is bigger on Google search. Here are my CTR numbers:

      Google Search
      Cities: 5.85%
      All other cities: 0.22%
      Total: 2.42%

      Sites on Search Network
      Cities: 1.22%
      All other cities: 0.13%
      Total: 0.49%

      CTR for reported cities was 26 times higher than for all other cities on Google search. On the search network the factor was about 9.5.

      I have also looked at data on state level:

      Google Search
      States: 2.96%
      All other regions: 0.41%
      Total: 2.42%

      Sites on Search Network
      States: 0.55%
      All other regions: 0.32%
      Total: 0.49%

      The factor here is 7.2 and 1.7, respectively. The differences are smaller on this level, which isn’t unexpected: It’s probably easier to match traffic to states than to cities.

    • Chris
      June 25, 2011 at 6:14 pm · Reply

      Rnasty (is that your real name?), I meant to reply earlier but got caught up with other responsibilities and completely forgot about my guest post. Forgive me.

      Hopefully by now you’ve already figured it out. If not, here’s the answer to your question…

      In your first spreadsheet, take a look at all the clicks you got from a region with no specific metro area, you’ll find that 79% of them are in the “All other regions” bucket (line 369).

      This “all other regions” bucket is where adwords throws in all regions that received less than 100 impressions or not more than 1 click as well as clicks/impressions that could not be mapped to a specific region.

      What wasn’t reported under a metro area doesn’t mean it was “outside” a metro area. It just means there’s not enough data to give you specifics about it.

      Hope you understand why your test with the 2 campaigns did not yield the results you thought it would. You approached it with the wrong premise.

      In practice, I simply ignore that bucket since there’s not much I can do about it – unless it makes business sense to create a separate campaign for every US region – and I focus on the rest of the data. (One targeting option you could use is tell Google not to use search intent to determine a user location in the campaign settings, but I wouldn’t recommend it to you)

      Something actionable?

      Example: In your third spreadsheet, line 176, Crowley received 2413 impressions, 2 clicks and a CTR of 0.08%… which is tremendously low compared to all your other cities. That is definitely something you can act on.

      So, yes… the geo reports are actionable.

Leave a Reply