Do Long Tail Keywords Convert Better?

Last week, Joe passed an article around the SageRock office from Marketing Charts called Online Search Queries Get Wordier. It was a short article so I’ll just post the whole thing here:

The number of words in the average US search query is on the rise, with longer search queries – averaging five or more words in length – increasing 10% overall in January 2009 vs. January 2008, according to data from Hitwise.

The most notable change was in search queries that were eight or more words. These were up 22%. Over the same time period, shorter search queries, averaging 1 to 4 words in length, have decreased 2%, Hitwise said.

Though Hitwise does not offer any explanation for the increasing length of search strings, the phenomenon could potentially be because internet searchers are growing more sophisticated in their searching habits, or might be attributable to the fact that longer searches may now be  necessary to find more specific information amid growing web clutter.

But, being the devils advocate he is, Greg countered with “Long Tail is Rubbish!” from Science for SEO. That post is a bit longer, but this snippet sums it up pretty nicely:

For some SEO’s the take is that you go for long-tail when you can’t rank for the competitive short queries.  They also maintain that long-tail does not convert high enough to be worth it.  I’m reading very conflicting things about this.

Considering I’m more on the paid search side of things rather than an SEO expert, I decided to use some data that was available and more familiar to me with the thought that if I have a large enough sample then the data could be applied to SEO as well.

I took one of SageRock’s larger clients and ran a Search Query Report for them in Google AdWords. To be fair, I excluded branded keywords since those traditionally convert better than non-branded keywords. My data was from May 2, 2007 (the earliest any Search Query Report can begin), through almost all of February 2009. The sample included over 40,000 clicks after those wonderful “other unique queries” were taken out. The client is a bank and a conversion in this case is a completed mortgage application.

# of Words Clicks Avg CPC Cost Conversions Conversion Rate Cost / Conversion
1 Word 1350 $6.29 $8,497.91 37 2.74% $229.67
2 Words 7119 $6.93 $49,320.22 87 1.22% $566.90
3 Words 17106 $7.04 $120,369.44 196 1.15% $614.13
4 Words 10457 $6.57 $68,754.49 99 0.95% $694.49
5 Words 5707 $7.08 $40,401.87 111 1.94% $363.98
6 Words 1296 $6.49 $8,411.13 30 2.31% $280.37
7 Words 350 $6.84 $2,392.65 12 3.43% $199.39
8+ Words 88 $7.01 $616.94 5 5.68% $123.39
Totals 43473 $6.87 $298,764.65 577 1.33% $517.79

As you can see, the Cost per Conversion figures take somewhat of a bell curve shape with the highest figures in the middle and lows on either side. This data suggests that as you target longer tail keywords your Cost per Conversion will go down, therefore increasing ROI.

Interestingly enough to note though, single keywords converted at a higher rate and lower cost per conversion than six word phrases. This makes me think that instead of the traditional view of the long tail spectrum…

…there’s actually 2 long tails situated around a bell curve like I mentioned above.

I got to thinking about why this would be and I remembered that long tail keywords are not comprised of just 5+ word phrases, but there’s also the misspellings of single words such as “morgage” or when searchers forget to put spaces between words and come up with “homeloaninterestrates” (and yes, someone actually typed that).

Of course, keywords like that rarely ever get searched for and are basically rendered useless for SEO purposes with search engine features such as search suggest and “Did you mean…?” So I decided to break down the same keyword list from before into groups of how popular they are from the top 10% of keywords that received the most clicks and on down.

Clicks Avg CPC Cost Conversions Conversion Rate Cost / Conversion
Top 10% 4309 $8.09 $34,872.10 82 1.90% $425.27
11%-20% 4374 $7.91 $34,617.59 94 2.15% $368.27
21%-30% 4340 $7.39 $32,065.99 51 1.18% $628.74
31%-40% 4386 $6.54 $28,694.80 56 1.28% $512.41
41%-50% 4416 $6.53 $28,841.24 60 1.36% $480.69
51%-60% 4418 $6.07 $26,836.27 63 1.43% $425.97
61%-70% 4403 $6.21 $27,322.49 58 1.32% $471.08
71%-80% 4301 $6.53 $28,100.59 44 1.02% $638.65
81%-90% 4400 $6.70 $29,463.19 37 0.84% $796.30
91%-100% 4126 $6.77 $27,950.39 32 0.78% $873.45
Totals 43473 $6.87 $298,764.65 577 1.33% $517.79

This chart tells a little bit of a different story than the last and is more in line with the findings on Science for SEO. The cheapest conversions generally can be found in the top 20% most popular keywords and conversions become around twice as expensive as you hit the long tail phrases that may only garner 1 or 2 searches every blue moon.

The results? For SEO, I still agree with the quote from Science for SEO that if you can’t compete with the behemoths on the most popular phrases then you should aim a little further down the long tail and pick a 4-6 word phrase. My advice is just to make sure this longer phrase can be used naturally when reading. There’s no use targeting a certain phrase if it reads all wonky and drives people away. SEO might get them there, but usability is king!

For paid search, I’d save the super long tail phrases for when your campaign is more mature and you have a better idea of what phrases work for you. To start, stick to the basic 2-5 word phrases that are the most popular searches done for your product or service. Once you’ve hit a stride with those and feel like you’ve reached your limits, then it’s a good idea to expand into the one word common misspellings and the edges of the long tail. Remember though, as you do this you’ll be giving up cost efficiencies for quantity of conversions.

And because Greg told me I have to add this video to my post, enjoy…

Long-tailed Macaque photo credit: Mike (NO captive birds) in Thailand

The Long Tail graph credit: TimWilson

Bell Curve graph credit: tomhe

Most recent posts by Nate


  1. This is an amazing post man. Thanks for all your thoughtful analysis!

  2. Thanks for the compliments! But really I just wanted an excuse to use a picture of monkeys…

  3. Pingback: Long Tail Keywords
  4. i dont understand one thing in your post. you said “So I decided to break down the same keyword list from before into groups of how popular they are from the top 10% of keywords that received the most clicks and on down.” bu the table shows the click count going up and down up and down from the top tenth percentile to the bottom tenth. what am i misunderstanding here?

  5. That’s really impressive, Nate.

    That’s a perspective that I have never seen before.

    I wonder if you could do another analysis of another client to see if you find similar results.

    Maybe that’s your next post? 🙂

Leave a Reply

Real Time Web Analytics