AdWords is now Google Ads. Our new name reflects the full range of advertising options we offer across Search, Display, YouTube, and more. Learn more

4.3K members online now
4.3K members online now
Welcome to the Official Google Ads Community
Find optimization tips and how-to guides and videos by the Google team, Google Ads Experts and other industry experts
star_border
Top Contributor

Keywords and the Pareto Principle

Vilfredo Federico Damaso Pareto was born in Paris in 1848 and became known for his contributions to the worlds of economics, engineering, sociology, political science and philosophy, so he made good use of his time.  He is possibly most famous for his discovery that income follows a distribution where a large percentage of income is received by a small percentage of the population (don't we know it!).  

 

This observation was found to apply to a surprising range of distributed data sets, so much so that it became known as the Pareto Principle.  To greatly simplify this principle, it can be said that for any given set of data, roughly 80% of the effects come from just 20% of the causes.  While Vilfredo spins in his grave over my simplification, let's look at AdWords...

Vilfredo_Pareto.jpg

 "Vilfredo Pareto". Licensed under CC BY-SA 3.0 via Wikimedia Commons

I review a great many Accounts and one of the most common things I see in those Accounts is that there are usually far too many Keywords.  Many advertisers appear to operate under the impression that they need to include Keywords that can cover every possible search term that might imaginably be relevant to their product or service.  This isn't just the concept of "long-tail" Keywords but rather an attempt to draw people in who may only be peripherally interested in the product, using words that perhaps are only vaguely related to the core business.  In most cases there is the potential for a conversion to occur for these terms, but in reality, those conversions hardly ever happen.  We're not talking small numbers here; many of these Accounts contains tens or even hundreds of thousands of Keywords and a couple many more than that...

 

So what does the Pareto Principle have to do with Keywords?  One Account I reviewed last year contained more than 3.3 million active Keywords (yes, that's just the enabled ones in enabled Campaigns/Groups).  I think anyone is likely to suggest that's too many, but it becomes even more interesting when you know that in the 12 months prior to my review only 41,000 had resulted in a conversion of any kind, that's just 1.2%.  In effect, that company could (probably) have paused 98.8% of all their Keywords and made the same income.  These figures are so extreme they're not even really a good fit for the Pareto Principle, but I find that generally speaking the idea behind this principle holds true:

 

Where an Account contains a large number of Keywords, usually only 20% or less of those Keywords are responsible for 80% or more of the income.

 

So, what are the problems associated with running an Account where most of the Keywords do not appear to be contributing to the cause?

 

  • Cost:  Obviously, if the 80% of non-converters are receiving clicks, that's just money down the drain.  If they're receiving anything close to 80% of the clicks, that's horrible.  If that were the case, your ROAS (Return on Advertising Spend) could be five times what it is currently.  Of course, in reality, the 20% will probably be responsible for a larger proportion of the spend, but any spend on the 80% is an opportunity for improvement in ROAS.

  • Time:  Time is money and managing AdWords effectively takes time.  I shudder to think at the enormity of the problem of effectively managing even many tens of thousands of Keywords, let alone millions, and the effort required is probably largely wasted if 80% of those managed Keywords are actually returning no income.  Even large scale tools like Scripts and the API will require a lot of work to manage large Keyword counts (in fact, at the time of writing Scripts can only process 50,000 Keywords in a single iterator and with more complex operations will time out long before this limit is reached) and for very large counts the cost of developing automated solutions could be substantial.

  • Flexibility:  When your Accounts are fit and lean, they are easier to work with for experiments and short-term promotions.  It's easier to keep track of changes and analyse their effects, easier to adapt targeting for locations, devices and demographics, create short-term promotions and Ad changes, etc.

  • Adapting to change:  We all know the AdWords program itself can change and some of those changes require user action.  If your Account is efficient, those change may be trivial, if it is bloated and unmanageable, a large change to operation could require weeks or months of additional work.

The only real "upside" to such large Keyword numbers is the opportunity for "assists".  It may be that some Keywords do not convert frequently but do assist frequently, and you should check for such situations.  However, I'd bet those assists will still be trivial compared to the 80% that convert directly.

 

So, how can you check your Account?  Fortunately, it's very simple.  

 

  1. Log into your Account and ensure you're showing only enabled Campaigns, Ad Groups and Keywords, then click the Keywords tab.
  2. Select a good long date range - at least 6 months probably.  Make a note of the total number of Keywords (rows) - look at the bottom right of the screen.

  3. Create a new filter, and choose Converted Clicks >= 1 (or another filter, depending upon your measurement of success).  Apply this filter and make a note of the new number of rows.

  4. Divide the second number by the first, and multiply by 100 and that's the percentage of Keywords that have converted in your time period.

Whatever figure you find, consider what it means and play with the filters.  It may be that you have a lot of Keywords that have converted just once, with a much smaller number that have converted more than 10 times, so spend some time examining various options.

 

What do I do if I find I should pause 80% of my Keywords?

The most important thing is not to make any rash decisions.  Even if you are sure you can pause these Keywords, you need to ensure you have a way back to the previous state before you do so, and the capability to measure the effect closely for a trial period after the change (so don't do it on a Friday afternoon and then go on holiday for two weeks).

 

I'd suggest using Labels to label each Keyword, Ad Group and Campaign you pause (bear in mind that pausing all these Keywords may cause some Ad Groups to have no enabled Keywords, and possibly some Campaigns to have no "enabled" Ad Groups).  You can then use the Label to recover all these elements and re-enable them if the experiment fails.  You may wish to use several Labels; one for those you're absolutely sure can be paused, another for those you're not sure about, and another for those with assists, etc.

 

Take a look at your Account now and do this analysis and if it turns out only 20% of your Keywords are creating 80% of your conversions, thank Vilfredo Pareto for spotting this trend in a program that was developed almost 80 years after his death.

about Jon Gritton

AdWords Management Consultant, a Google Partner, based in Dorset, UK, but working worldwide... I'm very proud of having been a Top Contributor for AdWords since 2006 and enjoy being able to help others improve and develop their AdWords usage. I now work primarily as an AdWords Manager & Consultant. You can read my eBook "AdWords Preschool", just search on Amazon.

Comments
Ratan-Jha Top Contributor
August 2015

Now that's what I call an article! A great read indeed.

 

Jon you made a complex phenomenon look too simple.  "You want to save time and keep things simple for yourself, that's why you are not inlcluding those many keywords for us" - I have dealt with this statement several times. Thanks for this article, it's going to be a great reference for the subject. 

 

Thanks again for elplaining the relevance of Pareto Principle in AdWords.

 

Thanks
Ratan Jha

semseoguy
August 2015

@Jon_Gritton

 

You've mention a very important issue and brought it to the point! Smiley Happy

 

THX and greetings from Germany

 

cu in SFO Kai

Justin F
August 2015

Really great article@Jon_Gritton

 

We've long had debates of more or less keywords for Accounts.  I find that it often comes down to preference of the manager.  But when I say "more" I do NOT mean those staggering amounts you mentioned, and unfortunately have seen some accounts with way too many as well.  And there certainly is such a thing as way too many!

 

I'm interested to know how exhaustive their negative keyword lists are =)
And I shudder at the thought of pouring over those search term reports.

 

I'm sure a lot of the "long-tailed" could be condensed into "long-tailed" broad-modified keywords

 

I've also seen multiple match types for the same keywords in campaigns - sometimes every single keyword!  And while multiple match types sometimes can make sense, I've seen keywords that already get so little activity spread across all match types.

 

Thanks for the post!

 

-Justin Fuchs

Adrian B
August 2015
My opinion is that landing pages generate the sales as keywords and Ads only capture the attention of the user for 2 seconds until they click. So if I see 80÷ of keywords not generating conversions or assisted conversions , first I look at the landing pages before stopping the keywords. In a badly managed account there could be perfectly relevant keywords associated with Ads that point to completely irrelevant landing pages or worse all keywords draw traffic only to the home page.
Sumanth Sridhar Rising Star
August 2015

What a way to explain keyword optimization! Hats off Jon Smiley Happy

Alexey C Rising Star
August 2015

I am constanly looking how to apply 80/20 in keyword management and I also observed so many times how small amount of keywords (sometimes even 1-2) are giving 80%+ amount of conversions.  I am gratefull to Jon for opening this discussion, should mention that I liked 2 points from your article: 
1) Assisted conversions report. We all know about it but sometimes I tend to forget about it checking only on search query report. 

2) Labels. I am not really using them but it sounds like a should give them a try. 

 

An important point arises from this discussion (Jon is already started to conver it) - pause or reduce bids on those thousands of longtail keywordswhen your budget is not highly limited? Lot's of them seems to be highly relevant, got more that 70% impression share, are in good positions but got only 4 clicks in last 6 month. What if the 5 click will bring conversion? Paying less for each clic can outcome in positive ROAS if longtails will convert. But what about this observation that conversion rate falls down when your ads are on lower position? did you made such observation as well @Jon_Gritton (my data is limited so it may be coincidence). 

Jon_Gritton Top Contributor
August 2015

Hi @Alexey C, I'll try and cover both your points (let me know if I've missed anything!).

 

If you have Keywords that appear relevant, and aren't getting many clicks in an unlimited budget, what you do with them depends largely upon how the rest of the Account performs for that product/service.  If the rest of the Account is producing 20 conversions a day, is a single conversion from that Keyword really worth it?  Could you spend the same time on the converting Keywords and get them to produce 25 a day?  You do need to give Keywords a chance to perform, and that time period you give them also depends upon the nature of the Account.  I have some Accounts that convert infrequently, but do so at a high conversion value, so I'll give those Keywords maybe 2 months to perform.  I have other Accounts where conversions are measured in the 100s per day, so I can tell within a few days if a Keyword is measuring up.

 

Conversion rate vs position is an interesting discussion, one we've had many times!  In theory it can be argued that position should have no impact upon conversions since (as @Adrian B has said) it's the landing pages that generate the conversions so a click from position #1 and a click from position #6 which both land on the same landing page should have the same chance to convert*.  The counter-argument is that clicks from different positions can sometimes indicate a different intent, or a differnt type of customer, perhaps especially once you take into account whether it's the first click or a click in a longer conversion path.  Personally, I've always felt that #1.0 runs the risk of "happy clicking" where the searcher simply clicks the first result in the SERPs without actually reading anything, so I tend to avoid #1.0 unless I'm bidding on brand terms, but beyond that it's really down to your own observations and data and, as always, that's going to give different results from Account to Account.

 

*I should add that this assumes all the Ads are highly relevant, and position is based purely upon the bid amount, not the quality of the Ad.

 

Jon

Alexey C Rising Star
August 2015

Thanks, thanks for feedback, @Jon_Gritton. You gave me additional point of view on this topic - I would rather work on my best keywords getting max value from them first of all. I also prefer to be on 2-4 position however I didn't think about blindness clicks before so you gave me few more usefull insight.

Lakatos Top Contributor
June 2016

That's an awsome article, Jon. Thanks a lot for sharing your valuable insights.

Labels