Long tail keywords are widely considered to be fundamental to search marketing success due to the low-cost traffic that they generate, however should this still be the case in 2017?

This article is going to analyze data from case studies conducted by Clicteq and PPC Hero that suggests that long tail keywords are now almost obsolete when it comes to AdWords.

For the sake of this article we are going to classify long tail keywords as being four-words-long or more, though there is some debate with people arguing that keywords that are three or more words long should be classified as long tail.

Search marketers have generally accepted that the key phrase curve looks like the one below.

It shows that a small percentage of traffic comes from 1-3 word phrases and the vast majority of traffic (around 70%) comes from the long tail (keywords that are four or more words in length).

Image Source

However analysis from more recent studies shows that as much as 52% of traffic is generated by single phrase keywords, which is in stark contrast to the 10% depicted by the diagram above.

The two long-tail case studies

This article will make three key arguments about the effectiveness of the long tail:

  1. From a time perspective you are 33% more effective working on the top 20% of keywords as you are working on keywords that contain four or more words.
  2. 90% of your impressions are generated by search terms that are four words long or fewer, therefore the long tail only accounts for 10% of all paid search traffic and not the 70% that the widely accepted key phrase curve depicts.
  3. For keywords that contain four or more words you would need on average to add 200 keywords to generate one click per month. Adding these keywords would be a poor use of your time and make your account hard to manage.

Argument 1. Your time is 33x more effective spent working on the top 20% of keywords as opposed to keywords that are four or more words in length.

Here is the data from PPC Hero’s case study that showed that keywords that are 5+ words in length generated 15 out of 608 conversions, resulting in a tiny 2.4% of the total number of conversions.

These 15 conversions were generated by 138,638 keywords, which is an awful lot to manage considering that they provide such a small return. 

Clicteq's case study delivered similar results when it conducted the same analysis. Here you will see that only 15 conversions came from keywords that were five or more words in length. This accounted for 2.7% of the total number of conversions.

However in this case there were only 533 keywords that were five words in length or longer. The graph below shows the percentage of the total number of conversions versus the length of each search term.

If you were to spend your time optimizing the top 20% of keywords that generate 85% of your conversions, instead of the keywords that are 4+ words in length, your time would be spent much more effectively.

For example if you were to increase your Quality Score from five to seven you would reduce your CPC by 26% according to studies by Wordstream

If you did this for all of the keywords within your account that are 4+ words in length this would result in a 0.62% decrease in CPC across your account. However if you were instead to do this to your top 20% of keywords you would decrease your CPC across your account by 20.80% which is 33X more effective.

Argument 2: If you swap all of your short tail keywords for long tail keywords you would lose 90% of your impressions and 80% of your conversions.

The table below shows the number of impressions and conversions for keywords depending on the number of words within the search term.

From the search terms report you will see that 434 conversions out of the total 532 were generated by keywords between 1-4 words long, accounting for 81.5% of the total number of conversions.

You should also see that 581 conversions were generated by search terms that were 4+ words long, accounting for 18.5% of the total conversions.

Furthermore 94.9% of all impressions came from search terms that were four words in length or fewer. (Some impressions appeared to be omitted within the SQR as the CTR was not around 20%).

There were similar results when looking at PPC Hero’s data when Sam Owen did the same analysis.

PPC Hero found that 93% of impressions came from search terms that were 1–4 words in length. So if you were to change out your short tail keywords for long tail ones you would loose 93% of your impression share.

PPC Hero’s data also clearly shows that the existing consensus that 30% of impressions come from keywords that are 1-3 words long is inaccurate in 2017. From their data you can see that 74.5% of impressions and 51% of conversions came from search terms that are between 1-3 words in length.

PPC Hero’s and Clicteq’s data would indicate that changing out your short tail keywords for long tail variants will result in a loss in conversions of around 80%. By combining the data from both case studies in 2017 the long tail curve looks like this:

Argument 3. You would need 200 keywords containing 4+ words to generate one click per month, which would be inefficient to manage.

One of the biggest issues with catching long tail search queries with exact keywords is that you will require a lot of keywords, which becomes inefficient to manage.

When Sam Owen from PPC Hero analyzed the number of impressions per keyword he found that search terms that contained 4+ keywords saw a significant drop off in conversions.

For example, to generate one impression you would need an average of 10 keywords that are 4+ words long. Furthermore, to generate one click per month at a 5% CTR you would need 200 keywords on average.

Even if you are able to capture all of these search terms with exact keywords you are unlikely to actually increase the performance of your account.

Once you start using keywords that are 4+ words in length you start to find that you can’t make your ads any more specific due to AdWords character constraints. Furthermore, you can’t really set specific bids for each keyword as you won’t have enough data and therefore will have to use aggregate data to determine the bids.

If your long tail keywords are generating 0.1 impressions per keyword per month then you would have to wait about 100 years to get enough data to make a statistically accurate bid judgment even if the keyword had a 10% CTR.

As you have to use aggregate bids and can't make your ads more specific, you may as well use shorter tail keywords that utilize the phrase or broad match types to catch these search queries.

If you are interested in running this analysis on your own account follow these instructions.

How long should your keywords be?

Based on the analysis from both Clicteq and PPC Hero the best length for keywords is between 2-4 words long. But it should be noted that in some industries keywords that are one word in length will perform well, as PPC Hero’s analysis showed.

When keywords start to exceed four words in length they generate very few impressions and conversions, and it comes to a point where time spent optimizing them would be much better spent working on your top 20% of keywords.

With regards to distribution, based on the two studies the bulk of your keywords should be between 3-4 words in length as these generally provide the best ROI when considering the amount of time that you are spending optimizing.

Three effective tactics for finding mid-tail keywords

1. Use keyword multiplier tools

If you are a retailer, keyword multipliers are a really smart way to generate a large list of mid tail keywords. Fashion retailers might use keyword multipliers to create keywords for each different size/colour of items.

For example, here is a page on ASOS’s website that has men’s polo shirts.

We want to use the different filters down the side to create keywords for each of the different types of polo shirt.

STEP 1. Open Google's keyword multiplier tool within AdWords Keyword Planner.

STEP 2. There should be three different fields where you can add lists to be multiplied together. Here you will want to start with the root word “polo shirts” in the first box and then add one set of filter values to the second box, and then another set of filter values to the third box.

For example, I have added the sizes to one box and then the colours in another as shown below.

STEP 3. Once you have done this, click “get forecasts” and then download the keywords that have at least 10 searchers per month to use in your campaigns.

2. Use Ubersuggest keyword tool

Ubersuggest is a very effective tool for finding mid to long tail keywords by scraping results for Google's autocomplete suggestions.

STEP 1. Go to Ubersuggest.io

STEP 2. Enter your root keyword, in this case “polo shirts”.

STEP 3. You will then see a large number of suggestions. At this point you will want to select relevant keywords and add them to a list, which you can download by checking the radio box next to the keyword.

 

STEP 4. To find even more relevant keywords, click on the little blue arrow next to relevant keywords and then click to expand this keyword. This will then show you additional relevant suggestions.

STEP 5. Once you are happy and have selected all of the relevant keywords, download them by going to the 'keywords selected' tab as shown below.

3. Reviewing your search terms report for long tail variants

Your search terms report can be a holy grail for finding new long tail keyword suggestions.

It shows you all of the different terms that users have typed in to find your ads. Here is a diagram that shows how the process works.

Image source 

STEP 1. Navigate to your search terms report as shown below.

STEP 2. You can now go down the list of search terms and look for relevant mid tail keywords to add to your campaign. Once you have found a relevant search term that you want to add to your campaign check the radio box next to it.

STEP 3. Once you have selected all of the mid to long tail search queries that you want to add as keywords, go to the top of the page and select "Add as keyword". Once you have done this you will then be prompted to click 'save' on the next window.

In conclusion...

When analyzing the data from the Clicteq and PPC Hero case studies, it is apparent that the accepted long tail curve no longer holds true in 2017.

In 2017 around 94% of impressions will come from search terms that are four keywords or fewer, compared to the 30% that was previously accepted, making it virtually impossible to double your revenue by changing out your short tail keywords for longer, more specific ones.

When considering the amount of time that it takes to optimise an account, the long tail is not as effective as it was first thought. The PPC Hero study found that 2.4% of conversions came from keywords that were 4+ words in length, while in the Clicteq study the figure was 2.7%.

When comparing the effectiveness of the time spent optimizing long tail keywords to that of optimizing the top 20% of keywords that generate 80%+ of your conversions, your time is spent 33x more effectively on the latter.

Finally, we found that on average you would need to add 200 keywords that were 4+ words in length to generate one click per month, which would be highly time consuming and a poor use of your time.

By all means these stats will vary slightly from account to account and industry to industry, but we have not yet found an account where the main arguments have not held true.

With this being a controversial topic I am interested in hearing other people's options on the study and welcome debate.

To learn more on this topic, check out these resources:

Wesley Parker

Published 9 February, 2017 by Wesley Parker

Wesley is the Managing Director of Clicteq and a contributor to Econsultancy. You can connect with him via LinkedIn.

2 more posts from this author

You might be interested in

Comments (18)

Comment
No-profile-pic
Save or Cancel
Kristina Angelova

Kristina Angelova, Digital Strategist at RIKA Digital

It's apparent that in 2017 the important metric is impressions rather than real engagement and saving time for the agency, instead of delivering meaningful results for the client...

I am by no means an expert on the topic of PPC but all of the above seems at odds with the rise with longer, contextual voice searches. Last Summer, Google announced that 20% of all searches are now voice searches and the key characteristic of those is the query lengths of 3+ words. These are also the searches that demonstrate intent and, from what I have seen with clients, are the ones that convert better. The younger generation are also much more comfortable using natural language when searching for products or services.

Given that, I would be reluctant to change strategy and announce long-tail is irrelevant, unless all you are after is traffic or cost/time reduction.

10 months ago

Wesley Parker

Wesley Parker, Managing Director at Clicteq Limited

Hello Kristina, by far the most important part of any PPC campaign is always going to be ROAS with other vanity metrics such as impressions and CTR fading into the background.

I'm interested in your point about voice search, I would agree with you that voice search is on the rise and that query lengths will increase as such. The question really is finding the most effective way to capture these searchers.

As you are unable to make your adverts any more relevant to longer terms over 4 words in lenght and there is going to be such as vast range of different search terms that data will be sparse for each meaning that you must aggregate data. You are likely to be best using shorter term 3 - 4 word keywords is most likely to be effective based on this data.

In house or agency side time is money and as such you must be doing the tasks that move the needle most, as with the point that I made the 80 / 20 rule.

Following this strategy you will maximise the performance for the amount of time you have to manage your campaigns. considering in house most digital marketing managers are juggling several different channels and agency side account managers work on several Adwords accounts this I feel is an important consideration.

10 months ago

Kristina Angelova

Kristina Angelova, Digital Strategist at RIKA Digital

Yes, completely agree with the points raised in the article in terms of operational efficiency, I was just wondering how this correlates with the emerging behaviours. It all comes down to the objectives, I guess of a given campaign and I am looking at this from a slightly different angle.

10 months ago

Wesley Parker

Wesley Parker, Managing Director at Clicteq Limited

It will be interesting to see how this analysis will pan out in say 2 - 3 years when voice search has been more heavily adopted, though personally I think it should hold true.

It make sense on certain accounts that are after say very cheap leads that can only be gained though aggressive long tail expansion then it will still be an effective strategy!

10 months ago

Adam Pritchard

Adam Pritchard, Director at Project Octo

We recently undertook some programmatic content creation ( exposing a valuable database of information to users) and raised site traffic by 400% - from 4000/mth to 19500/mth. This almost doubled site conversions (financial transactions) and we are continually working on improving upon this brand awareness.
So yes, the long tail does matter. The SEO work took us 6 weeks and was incredibly worth it

10 months ago

Avatar-blank-50x50

Roy A, Director Business Development at http://www.webecommercepros.com

I agree with the arguments in the post. When the length of the phrase gets longer than four words, the product or service becomes more clear for the visitor and the user is more able to make decision by reading it. Long tail keywords make him realize whether this is exactly the thing he is looking for or not. If it is what he is looking for he will click otherwise not.

But, four or less word keyword creates probability for the visitor to get his desired product or service by clicking on it. The hope and curiosity drives him to do so.

Medium length, like four words or less, leaves curiosity for the visitor.

By making lengthy keywords, we enable the user to take decision, and that's why very few click on it ( just the ones who are exactly looking for that particular product/service ).

Thanks for the very insightful post and comments..

10 months ago

Wesley Parker

Wesley Parker, Managing Director at Clicteq Limited

Hello Adam, I would agree with you about long tail searchers being important and very profitable as you have demonstrated. The article is more about the best way to capture them in PPC using different keyword lengths and match types which works differently from SEO.

10 months ago

Wesley Parker

Wesley Parker, Managing Director at Clicteq Limited

Hello Roy, I would totally agree with you here there is a strong correlation between the length of the keyword and the stage in the buying cycle. Thanks for your support!

10 months ago

Avatar-blank-50x50

Oren Netzer, CEO at cClearly, Inc.

Thank you for sharing Wesly and bringing to light such an important topic. I probably couldn’t disagree more with conclusions in this article. First, It is unclear which vertical this analysis was based on. The PPC Hero case study (from 3 years ago) seems to be based on analyzing one small AdWords account - their own adwords account, i.e. businesses searching for a ppc agency. I couldn’t find an indication of the vertical used or how many different accounts were examined in the Clicteq case study, which seems to have been based on one account as well. It is also unclear whether these were branded keywords or non-branded keywords (or a mix of those). Different verticals obviously behave very differently in terms of their “long tail” curve. Branded and non-branded keywords also have a very different “long tail” curves. Accounts of different sizes also behave very differently and the keywords currently in the account can affect the results significantly as well. I would certainly caution from drawing broad conclusions from analyzing one or two accounts in one specific vertical.

There are many other factors that I find have been overlooked here as well. One major factor is that long tail keywords cannot be properly evaluated when they are captured by phrase and broad match. If you capture them with a broad of phrase match you will be serving up ad copy and a landing page that is probably not very relevant, therefore you don’t get to reap the performance benefits of the long tail keyword. We have seen that if you take that same search term that was previously captured by phrase or broad match and add it as an exact match keyword in the right campaign and ad group, you could be getting 10 times more volume from the keyword and significantly higher engagement metrics!

I also disagree with your statement about search queries getting shorter which seems to have been made based on search query report data. Search query reports contain only search queries that received at least 1 click and omit many search queries with impression but no clicks, so I don’t think they are a good source to determine search query curve. In addition to the anecdotal evidence around voice search mentioned by one of the commentators, we have done our own analysis of search query length using panel data which showed an average search query length of 4.75 words per search query for non-brand search queries in the US and 4.35 in the UK, and even longer search query length in verticals with geo-modified search queries.

I do agree with your statement that managing a large number of keywords requires software/automation and is hard to scale otherwise (full disclosure: my company, cClearly, built a product for adding and managing long tail keywords at scale). Depending on the client and vertical, you don’t need as many long tail keywords as you would think to make a major impact on the account either. Marketers using our products have been able to increase conversion volumes by 30%-40% in relatively short order just by adding a few thousand new long tail keywords to the account and making an even bigger impact with a larger keyword volumes. If you want to read up more about some of our results you can find links to whitepapers http://www.cclearly.com/whitepapers/ and relevant articles http://www.cclearly.com/news-and-press-releases/ we wrote about long tail keywords in the past year.

10 months ago

Wesley Parker

Wesley Parker, Managing Director at Clicteq Limited

Hello Oren, thank you for providing a very detailed reply, I would agree with you that there is a very small sample size and there will be variations, being an agency not a PPC software company it is hard to collect together a large amount of data to put together a conclusive study, so I have included a link so that you can analyse accounts yourself. I would be interested to see a large scale study of this across many verticals and different size Adwords accounts . Although I would be surprised if the conclusions do not hold true, We have done it with a fair few Adwords accounts and got similar results though as mentioned above it is still not representative. I would also agree with you on the point that branded v non branded do have different curves. More well known brands are more likely to have a larger proportion of shorter tail keywords as they will have a large amount of branded traffic. The accounts that we ran this analysis on focused heavily on non branded traffic with branded traffic making up less than 10%.

Your point with the landing page and ad not being highly relevant is a good one, but I would at this point disagree with your conclusion. The problem with capturing search queries that are 4 or more words in length with relevant adverts is that character length constraints restrict how relevant you can make the advert. If you are using the headlines and the paths to make your adverts relevant then you are going to struggle to create highly relevant adverts for any search term that is 4 or more words in length. So for search queries that are 4 or more words in length you will be serving up the same adverts and not making them any more relevant that the ads you are showing for search queries that are 4 words in length if that makes sense. Granted you can make the landing page more relevant but this is a very time consuming process considering that you might have hundreds of thousands of long tail keywords within an account.

In response to "We have seen that if you take that same search term that was previously captured by phrase or broad match and add it as an exact match keyword in the right campaign and ad group, you could be getting 10 times more volume from the keyword and significantly higher engagement metrics!" I would fail to see how you would get 10x more traffic if you were unable to make the advert more specific to the search query due to the character constraints of an Adwords advert.

I see your point with regards to the search query report not necessarily being 100% representative of all of the search queries that have triggered adverts. However I would argue that it is likely to be strongly representative. Search Queries that are 4 words or more in length will all be displayed adverts that are not highly specific because of the advert character limits, so this should not skew the data towards a shorter search query length in the search query report if that makes sense. I would be interested to learn more about your panel research especially if there is anything on voice search!

I will certainly have a look at your white papers as this is a topic that I'm very interested in at the moment. Thank you again for your response to my article it's good to see both sides of the argument!

10 months ago

Avatar-blank-50x50

Oren Netzer, CEO at cClearly, Inc.

Hi Wesly - thank you for your response. In regards to how we get 10x more volume and better performance from the search queries when we add them as keywords, and how we deal with the constraints of creating very specific ad copy and landing page at scale, the answer is that in most cases you don't need to create new ad copy and landing page, you simply place the new keywords in properly matching pre-existing ad groups. Placing them in an existing ad group that has more relevant ad copy and landing page will increase all metrics. How we get to 10x? Not all clients get a 10x but the range is usually between 4x to 10x. Here is a simple example:

Suppose that by adding the search query as an Exact match you get 2x higher impression share, 2x higher click through rate and 2x higher conversion rate (relative to what the search query was getting previously when it was matched by a phrase or broad), you are now getting 2 x 2 x 2 = 8 times more conversions from that keyword relative to what it was generating as a search query!

This is not just a theoretical exercise - you can compare the click and conversion volume of the keyword to the click and conversions volume of the search term when it was matched by a phrase or broad match previously, you will see a similar increase. If you sample one keyword this can vary significantly but when you try this exercise on a large number of keywords you will find that the empirical results align very closely with the calculation we outlined above and that by adding the keyword as Exact match you will almost always increase impression share, click through rates and conversion rates dramatically.

I hope this clarifies my statements.

10 months ago

Wesley Parker

Wesley Parker, Managing Director at Clicteq Limited

Hello Oren, Thank you for your reply, I'm still confused by " you simply place the new keywords in properly matching pre-existing ad groups. Placing them in an existing ad group that has more relevant ad copy and landing page will increase all metrics." I don't see how this would improve the metrics as you are not making the advert any more relevant and you are not making the landing page more relevant which is where your previous line of argument was focused. If you are placing the advert in a more relevant ad group the increase here is not due to adding long tail keywords its due to using effective ad grouping and cross pollination keywords to ensure that the head term keywords do not cannibalise your long tail keywords budget.

Do you have a case study that compares the click and conversion volume of keywords before and after adding long tail keywords overall? I would be interested in having a look at this if you do! I might run a test myself over the next week and then write an article on that.

10 months ago

Avatar-blank-50x50

Oren Netzer, CEO at cClearly, Inc.

Wesley - here is an example: suppose you have an ad group for Exact matches for ‘takeaway Aberdeen’ that include Exact match keyword combinations of takeaway in Aberdeen. Now suppose you have another broad match catch all ad group +takeaway that captured the search query ‘takeaway in aberdeen open now’. The search query originally got matched to the general +takeaway ad group that has more of a generic ad copy and landing page (not geo specific). If you added [takeaway in aberdeen open now] to your ‘takeaway aberdeen’ ad group and the search query now gets matched to that ad group, it will now show a more geo-specific ad and landing page for takeaways in Aberdeen. Of course you can further improve performance if you create a special ad copy and landing page for this new keyword but I wanted to demonstrate that you can improve performance significantly by fitting into existing ones. In this particular case, the group of keywords added by this client increased conversion volumes up by 6 to 9 times.

9 months ago

Avatar-blank-50x50

Phil Wade, Search Consultant at Organic

Is there a reason why "For the sake of this article we are going to classify long tail keywords as being four-words-long or more" but then in the very first example the 101 conversions from 4 word keywords are ignored as if not long tail?

Other than brand v non-brand, the really key factor overlooked in this analysis is the ROI at either end of the keyword spectrum. It is quite often the case that generic short tail keywords are just too expensive to provide a profitable return.

Having a higher number of conversions does not equal more profit if the CPA is too high. In all the examples above it was assumed the target CPA was below that of the (usually higher CPA) short tail keywords.

For example, bidding on "Life Insurance Quotes" will cost £25-£50 for a top 4 placement, compared to the one word longer "Cheap Life Insurance Quotes" costing £15-£25 for the same positioning. With all other things being equal, it is easy to see how CPAs and margins are going to be widely different depending on your keyword strategy.

9 months ago

Wesley Parker

Wesley Parker, Managing Director at Clicteq Limited

Hi Oren, I see your point here, I don't disagree with you here that the change that you are talking about will improve performance (in fact its something that I do for my clients) and I often do see similar results.

But the crux of the argument that I am trying to make in the article is that if you already have a strong base of word 4 keywords such as "Best takeaway in Aberdeen" in broad match modifier for example. By adding longer keywords that are 5 or 6 words in length such as "takeaway in Aberdeen open now" in exact match won't have a higher CTR than the search query "takeaway in aberdeen open now" when it was triggered by the broad match modified "Best takeaway in Aberdeen" ad group. Simply because you can't make the adverts any more specific over 4 words in length in most cases if that makes sense?

9 months ago

Wesley Parker

Wesley Parker, Managing Director at Clicteq Limited

Hello Phil, Apologies this should read 4+ words long. (I will speak to the editor and get this changed).

The point that you raise about ROI is an important one and should not be overlooked. As you have very rightly pointed out some 4 word keywords such as "cheap life insurance quotes" may be much cheaper that there counterpart head terms. It is then down to your own best judgement as which keywords to choose depending on your target cost per lead. If your target CPA is low and can't be achieved by focusing on head terms then using longer tail variants is going to be your best option.

9 months ago

Ivan Palii

Ivan Palii, Marketing Manager at Boosta Ltd

Great structured overview, Wesley.
And I also want suggest our internal instrument, which most of our collegues love to use every day for functions you are writing. It's about free keyword suggestion and research tool - http://kparser.com
I'll be happy to know you opinion about that tool.

8 months ago

Krishna Soni

Krishna Soni, CTO at PassiveLead

Thanks Wesley !
Very good article, with lot’s of awesome information! We are looking to start PPC campaign and it is what we found at the right time. Also going to share on Linkedin.

4 months ago

Comment
No-profile-pic
Save or Cancel
Daily_pulse_signup_wide

Enjoying this article?

Get more just like this, delivered to your inbox.

Keep up to date with the latest analysis, inspiration and learning from the Econsultancy blog with our free Digital Pulse newsletter. You will receive a hand-picked digest of the latest and greatest articles, as well as snippets of new market data, best practice guides and trends research.