Should You Remove Keywords With Low Click Through Rates?

Because the AdWords system rewards keywords with high click-through history (relative to competitors) with better ad positions and lower cost-per-click, click through rate is considered an important performance metric. Along with a keyword’s relevance to ad text and landing page copy, click through rate influences a keyword’s “Quality Score.”

Every PPC campaign is bound to have a few (or few thousand) keywords with low click through rates. You can identify them easily enough with web analytics and campaign reports, but what do you do with them?

You have at least 6 options:

1. Do nothing. You’re always going to have stinkers, why major on the minors?
2. Try to improve your Quality Score, which should improve ad position, which may positively affect click through rates.
3. Add negative keywords if you’re using broad or phrase matching.

4. Create a new Ad Group. Pull poor performers out of your current Ad Group and start over with better ad text and landing page.
5. Create an AdGroup for branding purposes. You don’t expect clicks, but using your company name in the headline is free exposure.
6. Pause or delete them. Either way, you stop bidding.

But before you take action based on click through stats alone, it’s important to dig deeper as to why the click through rate stinks.

Potential Reasons for Low Click Through Rate

If your average ad position is high (1-3), it’s probably not a Quality Score issue. It’s more likely one of the following:

  • Your organic rankings for the keyword are so good, people aren’t clicking on your PPC ads, and the “double listing” of your PPC ad improves your organic click through rate! You pay nothing for the additional branding, and removing the keyword may even slightly hurt you. Do nothing, except maybe “do a little dance.”
  • Your keyword has low commercial intent – meaning people aren’t interested in a purchase, they want information. Are you bidding on “wii news” because it got 22,000 searches in June? Kill the keyword phrase, and consider adding “news” as a negative keyword.
  • Your keyword is broad or phrase matched with insufficient negative keywords in your campaign. Use yesterday’s Google Analytics hack to expose the actual search queries that triggered your ad, and add negative keywords as necessary.

If your average ad position is medium (4-10), you may have any of the above problems, plus:

  • You’re in the Automatic Match beta. You have been automatically included and your ad is showing up for synonyms to your broad matched terms, while your competitors are not. If you are part of the beta, you will see a checkbox to opt out of Automatic Match from your Campaign Settings. Just opt out, don’t be a guinea pig for Automatic Match.
  • Your ad copy stinks compared to your competitors. They have tested and found winning headlines, calls to action and display URLs. They display prices that are lower than yours. They offer guarantees and free shipping in their ad copy. Customers trust their domain names more than yours. Go to the SERPs and see for yourself. And test out different ad versions.

If your average ad position is low (10+)

  • You may be bidding too low vs. your competition or for the Quality Score Google has assigned you. You may have set an initial CPC that was low and performed fine, but competition has entered the picture. Or Google simply decided to raise minimum bids for whatever reason. Increase bid as long as it makes sense to, and within what you can afford.
  • Your Quality Score stinks because your keyword is in the wrong AdGroup. For example, putting “learning toys” in the “educational toys” AdGroup, means your ad might display with “Educational Toys” in the headline, pointing to a landing page that never references “learning toys”. The searcher is more likely to click on results that use “Learning Toys” – it’s more relevant, though it describes the same thing. And, your Quality Score suffers when your ad text is not as relevant to the keyword. Create new Ad Group, but don’t delete similar keywords like “early learning toys” unless they also have poor history. Otherwise, you lose that history.
  • Your keyword is irrelevant to your products. Perhaps you’re a victim of sloppy outsourced keyword research, or a consultant that didn’t fully understand your business. Nix that keyword, and any others that don’t belong.

Can low CTR% be a good thing?

There may be instances you want to lower click through rates. For example, if you sell high end furniture, adding “From $2999″ to your ad for “teak outdoor patio set” will weed out the shoppers looking for Ikea-grade, who are thinking frugal but not expressing it in their search query. Plus, you’ll likely increase click through from luxury buyers. Your conversion rate, cost per conversion and ROI will improve. (It would make sense that Google factor conversion rates into Quality Score, since it is a better indicator of relevance than click through rate. Perhaps it’s one of the “other relevance factors” Google keeps to itself.)

What About Keywords With Low Conversion Rate or Negative ROI?

That’s a bit trickier.

Low conversion rate

Why spend money on keywords that don’t convert, right? The problem is, a keyword may have a 0% conversion rate but still be responsible for many sales. According to a 2005 comScore study, searchers who ultimately purchased online performed an average of 13 searches before converting, resulting in 12 non-converting searches for every sale. If the sales cycle exceeds your cookie expiration dates, some keywords may never get the conversion credit they’re due. (Great article on non-converting keywords by Frederick Marckini at Clickz)

What’s more, online searches can result in telephone orders, or even offline sales – which are even harder to reconcile, since there’s no cookie that tracks those.

Negative ROI

Keywords with negative ROI should be investigated. Are bids too high? Can landing pages be improved? Is broad match burning your budget and could keyword research help? They can even be a blessing. Lessons you learn from attempting to salvage negative ROI keywords may even benefit your campaign as a whole if you can apply “better practices” across the board.

If margin on the products or overall sales are low, you may decide to kill the keyword based on negative ROI to allocate budget for clearly profitable keywords and products.

The takeaway is to never kill a keyword simply because of a low metric. Always investigate the possible reasons for the low metric.


Related Articles

28 Responses to “Should You Remove Keywords With Low Click Through Rates?”

  1. I didn’t even mention click fraud, but that might be another reason for HIGH click through rate – look for unusual spikes in your trend data. If anyone has good articles on click fraud, please comment…

  2. Jeremy says:

    Nice write up Linda.

    I always tell folks to not worry about CTR unless it’s impacting your QS and therefore min bid. CTR, at least in relation to determining QS, is normalized by position. Simply put that means that if you’re in spot 7 with a CTR of 0.20% but are still seeing a “good” or “great” quality score odds are your CTR is still solid in relation to the other ads in that position.

  3. Adrian says:

    Great article.

    I often think people focus to much on increasing CTRs while totally ignoring the fact that it might be hurting their profits. Increasing position or CTR doesn’t always mean higher profits. It all comes down to testing though.. test test test :)

  4. thinks says:

    Reduce costs for running your online business (Small Business Ecommerce Link Digest)…

    Marketing your business online is generally low cost. But here are tips to help you reduce costs even more.
    ……

  5. Nice summary of an important issues. Another key think to consider is how the keyword relates to the search queries it’s attracting, and how the full diversity of these relates to the text-ad and landing page.

    There are a huge range of ‘questions’ and ‘ideas’ that can be matched against one keyword/ad set if you’re not careful – it’s not surprising that many searchers don’t find the ‘answers’ relevant to them. Imagine a salesman who repeats the same pitch not matter what the customer says.

    This is an issue of using the right match type, and more likely a wider range of keywords and match types to ‘trap’ the right queries at the right levels and stear different queries to different ads. I’ve written a number of lengthy posts on this topic recently at http://www.clickequations.com/blog/category/match-types/

    When you have legitimate low CTR keywords, and want to run them, separating them into their own ad-groups has the added benefit of making your reports far more useful, since the averages they report won’t be weighed down by these ‘losers’.

  6. @ Craig – that’s a good point, bounce rate on a keyword by keyword basis would give you a good idea on whether the landing page is what someone is looking for or not. Bounce rate may also be factored into the “other factors” in Quality Score. I wouldn’t be surprised.

  7. [...] are many reasons people don’t click (see this post for a good list). Many could not be translated into paid search metrics without qualitative research. But there [...]

  8. Lars says:

    Not doing anything about low-CTR keywords doesn’t look like a good option to me. Obviously CR is much more important, but a low CTR usually entails a low Quality Score (QS) that could have an effect on the other keywords in the same ad group and thereby negatively influence the whole campaign.
    I would therefore always recommend to at least put these “low performers” into a separate ad groups so that the rest of the campaign won’t suffer.

  9. Hi Linda – great post on a very critical optimization. One other thing that people may want to look at when faced with low CTRs is to review the geographic targeting as well as language settings. In some cases the keyword may have a low intent only for certain regions.

    For example, keywords that may have very similar spelling between languages but very different meanings – especially at risk are campaigns that select multiple countries and multiple languages even though the keywords/adtext are in 1 language rather than splitting campaigns up for better control.

    Same idea goes for a campaign that targets a specific region (say New York State but the product and ad is for New York City – CTR may be much lower for upstate NY than Metro NYC).

  10. Linda, this is a great analysis of potential problems!

  11. Great post overall, though I want to point out that, re: telephone orders, offline sales – “harder to reconcile, since there’s no cookie that tracks those”… there are a slew of options still available to the advertise. Voiceover inbound call tracking e.g. Voicestar, plus you can offer promotional codes, tag documents with special codes, etc. You’re never licked — there’s always a way to narrow it down.

  12. @ Lars

    Just a caution on creating new ad groups for low performers for a couple reasons:

    1. Performance can be improved over time. If you break them into new groups, you lose history, and if you improve them and want to put them back into the original group, or unpause them, well, it doesn’t carry the good CTR history with it back to the group.

    2. Too many similar ad groups can be hard to manage. Your quality score *may* go down if you have too many ad groups pointing to the same landing page. I’ve heard some folks echo the same.

  13. [...] Second, you need to decide your strategy to improve this keyword’s performance. Linda at the Get Elastic blog has posted a very thorough article on how to deal with keywords with poor click-through rates. [...]

  14. JB says:

    “Your quality score *may* go down if you have too many ad groups pointing to the same landing page.”

    I’ve never heard that one before. I’d like to hear more about that, maybe in a future post.

  15. Mike says:

    Low click through ratio keywords will lower you quality score.

    Low quality scores means that your keywords are more expensive.

    Example

    Great quality score = .04 minimum bid

    Poor quality score = 1.00 minimum bid

  16. Just to be clear on this, quality score applies on a keyword by keyword basis, at the time the keyword search is performed. So leaving low quality score keywords in your campaign doesn’t hurt the quality score of the other keywords :)

  17. Dan says:

    You mention that an ad with a low conversion rate may be worth hanging on to. The counter argument is that the impressions without clicks are damaging you ad group quality score and your campaign quality.

    In my opinion, delete the poor performers. You don’t want the dead weight. Plus, you can always come back later a try to revive them if you need.

  18. [...] But before you pull the trigger on a PPC keyword, you might want to give this a read. [...]

  19. Govindaya says:

    A SUPPORTED BY THE DEVELOPER Instruments? It was interesting. You seem very experienced in your study.

  20. John says:

    This is a very good article. I enjoy the section on “Potential Reasons for Low Click Through Rate”

  21. [...] Keyword Ideas for Holiday PPC Campaigns Holiday SEO: Using Amazon Bestsellers for Keyword Research Should You Remove Keywords With Low Click Through Rates? Stop Google Analytics From Stealing Your Valuable AdWords Keyword Data Negative Keyword Research [...]

  22. [...] PPC temptation is to kill keywords that don’t convert. Perhaps your keywords are not performing well because you’re not optimizing the experience [...]

  23. [...] Can Product Images Improve Conversion? Showing Products in Context Stop Google Analytics From Stealing Your Valuable AdWords Keyword Data Should You Remove Keywords With Low Click Through Rates? [...]

  24. Great little summary on how to handle different performance scenarios!

  25. http://n-pinokyo.com/ot/blogs/math-tuition-certainly-makes-the-grade.html says:

    howdy,good submit! to the educational costs expertise in addition to Post wonder perhaps there is almost any alternative for a start up exactly who would not know high of net? are usually propose software just like intended for home tuition ? if you will have several assist in the area is usually significantly recognize the value of seeing that pondered spend handful of an hour exploring incorporate your web site for getting it. i am aware the house college tuition enterprise have got a great prospective client in foreseeable future understanding that the main reason we are shopping good application designer who will supply me precisely what i want that will me growth our enterprise. any person with ideas you should can concept myself back plus want your website have more website visitor.

  26. hi there,fine publish! for the educational costs expertise and also I ponder perhaps there is any alternate for a start in place who won’t understand much of web? will be advise application just like with regard to home tuition agency? if you will have a number of assistance from the community is definitely greatly understand when pondered spend very few hr searching include things like the web site to receive them. i know the home tuition enterprise have a very very good prospective client throughout future which the reason why i’m searching fine software designer who will offer me personally just what exactly i need to help me personally growing my own small business. anybody who’ve thoughts please do concept me personally backside plus hope a person’s website have far more visitor to your site.

  27. Steve says:

    Just to address what you said about conversions being a factor in QS. There was a google blog that explicitly denied this – and went into the reasons why conversions are not a factor (basically, it would be too open to abuse – all a person would have to do would be to set a stupidly easy criteria for a conversion – say, the user clicking on the only page that’s linked to the landing page – and hey presto, QS would duly rise. Compare that kind of conversion with someone completing a checkout process, for instance, and you can see that unless every conversion was analysed on an individual basis (probably more detailed than just the three ‘types’ of conversion that Google offer) then it would be a fairly meaningless factor.

    Also, Linda Bustos (up thread) said “leaving low quality score keywords in your campaign doesn’t hurt the quality score of the other keywords”. Not strictly true. Poor quality keywords have a small impact on what might be called overall campaign QS and ad group QS (both of which are hidden from us), which then feeds back into the QS of all other keywords in the campaign/ad group.

Leave a Reply

© 2014 Get Elastic Ecommerce Blog. All rights reserved. Site Admin · Entries RSS · Comments RSS