Why Patience Is A Virtue With The Long Tail

A recent article by fellow Search Engine Land contributor Matt Van Wagner on long-tail keyword management  - as well as some recent experiences with several large-scale advertisers, inspired me to write this article on quantitative long-tail management.

Too often, I have seen cases where marketers apply incorrect reactive rules on the long tail and kill it in effect. This can have a disastrous effect for a large-scale marketer with an ad spend in the hundreds of thousands to millions of dollars a month, where the tail can represent a significant portion of the business.

The Dilemma

Take a look at these two keywords. Which is better?

Keywords Clicks Cost Orders
KW 1 (Head term) 2000 $2000 40
KW 2 30 $40 0

Many are tempted to say that keyword 2 is a bad keyword. It has received 30 orders and no conversions. Surely, 30 clicks should be enough to decide if a keyword is good or bad. But is it?

If you look at the head term, the average clicks per order is 2000/40= 50 clicks. Another way to look at the data is to build a histogram of clicks per order for the head terms.


The peak Clicks/Order for the distribution is around 35 (median), and the mean clicks/order is even higher, as there is a significant fat tail to this distribution. Looking at this, one could say that 30 clicks is not sufficient to make a conclusion about the keyword. If I bid this keyword down, then I might miss out on a conversion that I might have otherwise received.


Bucketing: A Better Approach To Look At Data At A High Level

A better approach to measuring the efficacy of the long tail is to group keywords into the number of clicks they received and measure the aggregate. This is shown below:

Range of Clicks # KWS Spend Clicks Orders Clicks/Order
1 to 5






6 to 10






11 to 50






51 to 100






101 to 500






501 to 1000












One can clearly see that that head bucket is the best performer, as it has the lowest clicks per order. This is expected here as the head buckets are mostly brand terms.

The 6-10 bucket is the worst performing one that perhaps bears investigation. However, one must note that the key assumption here is that all keywords in a bucket are homogenous in terms of their properties, e.g. consumers will convert and behave with those searches in a similar manner.

Bucketing essentially has an averaging effect and gives us a high level understanding if the performance is strong or poor.

The Danger Of Short Revenue Windows

Too often, I notice advertisers using a short window when determining a keyword’s efficacy. I have seen tail management rules such as “If a keyword has 10 or more clicks in 7 days and no conversions it should be bid down”.

These arbitrary rules can easily kill the long tail, as the parameters (clicks and days) do not account for:

  1. The average number of clicks you need for a conversion
  2. The average time taken for a given tail term to get the number of clicks.

In my research, I have found that typically 40-50% of keywords that get clicks in one month do not get clicks in the next month. Moreover, a conversion event might take several months to occur for a tail term.

A rule like the one above will systematically kill tail terms, for at any given point in time a set of tail terms will appear “bad” will be bid down before they can become “good” again.

Five Tips For Effective Long Tail Management

  1. The best way to manage your long tail is to use statistical algorithms that determine the probability of conversion given the number of clicks a keyword has while using the distribution of clicks per conversion on keywords with enough data (mathematically called the “prior”).
  2. If you do not have access to such a platform then first do not use an arbitrary rule. Let the data determine your click threshold. Second, look at data over a longer period of time. A short window can misdirect your efforts.
  3. Bucketing keyword data is useful as it gives you an understanding of where performance could be off-the head, torso or the tail.
  4. Avoid drastic actions like pausing words if they don’t meet your “good ROI” criteria, a gradual bid up /bid down approach is better.
  5. Finally, save yourself from the myth of “wasted” spend.  If you sum all the keywords with clicks and no revenue you might think that you have “wasted” all this money on keywords that don’t convert. If you are seeing this data this way, then I would recommend pulling keyword reports from the previous time period and check what fraction of the keywords in the “wasted” bucket generated revenue in the previous period. The numbers might surprise you. Like I said before, the average number of clicks per conversion and the time period are critical to this analysis.

When you manage tail terms right, you can increase the performance of your marketing campaign, but it requires us to overcome our discomfort with sparse data and act in a patient, rational and statistically reasonable manner.

Opinions expressed in the article are those of the guest author and not necessarily Search Engine Land.

Related Topics: Channel: SEM | Enterprise SEM


About The Author: is Director, Business Analytics at Adobe. He leads a global team that manages the performance of over $2 BN dollars of ad spend on search, social and display media at Adobe.

Connect with the author via: Email | Twitter | Google+ | LinkedIn


Get all the top search stories emailed daily!  


Other ways to share:

Read before commenting! We welcome constructive comments and allow any that meet our common sense criteria. This means being respectful and polite to others. It means providing helpful information that contributes to a story or discussion. It means leaving links only that substantially add further to a discussion. Comments using foul language, being disrespectful to others or otherwise violating what we believe are common sense standards of discussion will be deleted. Comments may also be removed if they are posted from anonymous accounts. You can read more about our comments policy here.
  • http://www.datapop.com Frank Lee

    Nice post, Sid. I agree with points here. On “wasted spend,” I would also add that i typically look to spend 10-15% of my spend in any given month to “testing” budgets. Meaning, if i have 10-15% of my budget being spent to keywords with 0 conversions, i am okay with that (as long as my overall CPA/ROAS goals are being met) because i am trying new things.
    I agree that many folks bid down or pause keywords too soon, w/o enough data to support their decision. I’ll also throw out there that if you test for the proper message against these keywords, often times these keywords justify themselves on position 1 on Google. Once you find the optimal combination of keyword, ad and landing page, then determine the best bid price.
    Lastly, i love the idea of bucketing keyword data. in fact, that is how you can understand creative performance as well… by bucketing like intents and understanding the messages that lead to the best response points.

  • Matt Van Wagner

    Hi Sid,

    Nice article, and nice recommendation for another way to approach to difficulty of managing through the fog of long tail keywords. I agree that ‘wasted’ is the wrong way to categorize the cost of the long tail. I think of it more as a necessary overhead to be managed appropriately.

    Applying an average yield approach to determining keyword decision thresholds (bid down, bid up, pause, delete) seems like a more reasonable approach than simply applying an arbitrary pulled out-of-thin-air threshold based on time or clicks. The challenge in applying this approach, I suppose, is to balance the gain you will get by bucketing your l-t keywords by characteristic (meaning, clicks, aggregated CTR or CVR, CPA) against the cost of complexities required to manage the effort. Managing long-tail keywords by their ever-changing performance characteristics,almost by defintion would have to be done by machine, since any changes made to manage the long tail bids directly and immediately influences performance characteristics of your other keywords that compete in the auction.

    I get dizzy just thinking about it. Very glad there are companies like yours who have algorithmic systems designed to operate inside the fog of long tail.


Get Our News, Everywhere!

Daily Email:

Follow Search Engine Land on Twitter @sengineland Like Search Engine Land on Facebook Follow Search Engine Land on Google+ Get the Search Engine Land Feed Connect with Search Engine Land on LinkedIn Check out our Tumblr! See us on Pinterest


Click to watch SMX conference video

Join us at one of our SMX or MarTech events:

United States


Australia & China

Learn more about: SMX | MarTech

Free Daily Search News Recap!

SearchCap is a once-per-day newsletter update - sign up below and get the news delivered to you!



Search Engine Land Periodic Table of SEO Success Factors

Get Your Copy
Read The Full SEO Guide