≡ Menu

How to trim the long-tail fat from your paid search program

Let’s face it. Enterprise paid search advertising is a big mess of big data. A lot of the manual techniques that can be used to optimize small to medium-sized paid search programs are simply not scalable when you’re dealing with thousands upon thousands of individual keywords and ad groups.

This is especially true when it comes to the famous “long tail” of search.

Unlike head and torso terms, which have a relatively high amount of search volume and therefore result in a relatively high amount of clicks and conversion data, many long-tail terms sit there gathering dust. They only result in clicks every once in a while, which makes them notoriously difficult to evaluate from a conversion and ROI standpoint. Moreover, there are so many of them that attempting to evaluate on a one-by-one basis is a fool’s errand.

But with a bit of craftiness – and a penchant for slicing and dicing big data – it is possible to identify and eliminate unproductive long-tail keywords en masse.

I can’t give away the exact recipe that we use to identify long-tail clunkers, because that would just be wrong. However, what I can do is provide some key reporting dimensions, metrics, and cadences that you can use to build out your own scalable approach to managing the long-tail:

  1. Establish a minimum threshold of clicks/spend without a conversion – we have a unique recipe for determining this at HSN but you can experiment with different thresholds until you determine the right threshold for your business/client. In addition to that threshold, decide upon a time frame in which you will query conversion data to determine non-converters (e.g. no conversion after x clicks/spend over a six-month period)
  2. Compare different attribution models – A good place to start is by comparing last-click conversion data vs. first-click conversion data. The goal here is to ensure that the clunkers you identify are true clunkers no matter how you look at it (Note: If you don’t know what the difference between first-click and last-click attribution is or how to gather that data, find out in a hurry because your business is missing a huge piece of marketing perspective)
  3. Query your keyword data on a regular, ongoing basis to identify new clunkers – Quarterly, monthly, maybe even weekly. Whatever floats your boat. The key is to continually mine for keywords that are simply not worth keeping in your portfolio
  4. Develop a strategy for reallocating the spend that you “save” by pausing unproductive keywords. This is key because the ultimate idea isn’t to simply remove unproductive keywords. The real goal is to take those savings and reallocate them into productive areas of the program so that you can maximize your paid search budget (or your overall marketing budget across all channels)


Comments on this entry are closed.

  • http://twitter.com/content_muse Anthony Pensabene

    Hugo have more thoughts, but sleep deprived and my fingers won’t type my thoughts well..

    #4 immediately resonates..in this case ‘unproductive,’ meaning, ‘not keeping close watch on leveraged terms.’
    Last night, my dad excitedly showed me a favorite golf site of his.
    A: I noticed he typed in the name every time B: the site was ranking for their brand name (organically) in the top spot. However, G was also serving up paid ads, which my dad clicked on (each time)..
    “Dad, if I was there ad guy, I’d probably turn that ad off/modify for the brand term”
    “Because of dudes like you. 🙂 ”

    • hugoguzman

      Ha! Get some sleep, Anthony. Interestingly, that anecdote is funny but not necessarily representative of majority search behavior. Moreover, I’ve seen data that suggests that in some cases (not all but some) not bidding on your brand term can allow competitors to occasionally swoop in and steal your customers.

      Definitely interested in hearing your other thoughts, but get some sleep first!

  • Pingback: Marketing Day: September 7, 2012

  • http://www.facebook.com/people/Steven-Werner/612913547 Steven Werner

    I’ve been using this method recently to cull wasteful keywords. I’ve run into a problem however, and I’m not sure how to get around it. Lets say you run the following filter on all keywords: “Has spent more than X and had no conversions, over time period Y”. The problem here is you aren’t accounting for keywords that haven’t run for the full amount of time Y, right? You could run a report over the last 6 months, but what if some of the keywords may have only been in the account for 2 weeks. Is there any way to run a filter that says “look only at keywords that have run for at least time Y”? There is a campaign start date filter, but not a keyword start date filter.

    • hugoguzman

      Hey Steven! Thanks for chiming in. One easy away around your challenge is to establish a minimum threshold of clicks without a conversion. That way, even if you aren’t sure that the keyword has satisfied the time threshold you can be sure that it reached the minimum click threshold. Let me know if that helps.

      • http://www.facebook.com/people/Steven-Werner/612913547 Steven Werner

        I’m slapping my forehead. That’s obviously a good solution. I suppose in this case a minimum click threshold is more important than a minimum time threshold.

        • hugoguzman

          Glad you found that tip helpful, Steven! We all live and learn together.

          P.S. Looks like you’re a banjo man. Would love to jam sometime (I’m a guitar/base kind of guy)