PPC Myth Week Pt 3: Kill Keywords That Dont Convert

This is Part 3 of our PPC Myth series. Please check out Organic Search Traffic is More Qualified Than Paid and Bid Higher to Appear Higher if you missed them.

There’s a lot of PPC experts out there who will tell you to look at your PPC keyword reports and get rid of keywords that don’t convert. Sounds logical, right? Why spend money on losers when you can spend more on winners? Especially when you’re under pressure to show strong ROI (or ROAS – return on ad spend) or are working with a tighter budget in these tough economic times.

But nixing “non-performing” keywords is not always a good idea.


Most analytics reports (including your Adwords report) credit the last keyword clicked before conversion. For example, your customer searched for “kids bedroom furniture” on Monday and found a Cars movie race car bed on your site. The customer searched Google for “Cars movie racecar bed” on Tuesday, clicked your paid or organic search listing and completed the purchase on your site. “Cars movie racecar bed” is credited for the $400 while “kids bedroom furniture” registers as a non-converting click. Because the credited keyword is “long tail” – perhaps that click only cost $0.50 while the more competitive “kids bedroom furniture” costs $2.50 – certainly one appears more “profitable” than the other.

Multiple keyword searches and site visits are not uncommon. According to a 2005 comScore study, people perform an average of 13 searches before converting — leaving 12 keywords out in the cold in conversion reports. (Though these keyword searches may lead a customer to other sites, not just your own). Craig from ClickEquations shares some actual data on visitor behavior on his blog.

There is much debate whether philosophically the first or last click should be credited – or credit be divided across keywords. And there are tools like Omniture SiteCatalyst that allow you to use “linear” allocation (again, Craig shares an example).

But this post does not attempt to solve the attribution/allocation dilemma. Because allocation/attribution is not the only thing messing up your keyword reports! Other reasons keywords may not receive the credit they are due:

  • Orders placed by telephone. There are ways to track telephone orders, but it is not default in any analytics package.
  • Cookie deletion. The customer clears cookies, uses another machine or browser or returns to your site after the original cookie expired. Any of these would fail to correctly credit a keyword. (Some estimates suggest 30% of web users regularly clear cookies)
  • The broad match type. For example, the “kids bedroom furniture” keyword may be matched to a search for “kids bunk beds” which you don’t sell. A high volume of searches for “kids bunk beds” and other searches that cause your ad to appear will boost a keyword’s impressions and will either dilute your click through rate (if your ad is not specific to the search term) or your conversion rate (your landing page doesn’t match the search term). If you use broad match – always use the broad match keyword exposure filter.

Before you hit delete…

1. Add this Google Analytics filter so you can see what exact searches trigger ads from your broad match and phrase match keywords. Anything irrelevant gets added as a negative keyword at the Campaign level (to prevent ads from other Ad Groups from appearing).

The benefit will be a better click-through rate (less clicks but far less ad impressions). You’ll have a lower absolute spend because you’re not paying for irrelevant keyword matches anymore, and your higher click-through rate means a lower cost-per-click. Hurrah!

2. Play customer on your own landing pages. Think about search intent – certain keywords are more “informational” than “commercial.” Would someone using the keyword in a search engine hope to find information or a product page? How can you improve your landing page to connect with that visitor? Does this keyword need its own Ad Group with its own landing page?

3. Chop at the Ad Group level. If you need to save money on PPC, figure out which product/categories are low margin, under-performing or are too expensive per click to keep bidding on and pause or delete the entire Ad Group, rather than killing individual keywords.

Again, deleting individual keywords within Ad Groups may not improve your results, because these keywords may not be getting credit for all the “assists” they’ve made to conversion. And your ads may still appear thanks to the broadness of broad match if the same search just gets matched to a similar keyword. If you do remove a keyword, be sure to add it as a negative keyword at the Campaign level.

Related Articles

8 Responses to “PPC Myth Week Pt 3: Kill Keywords That Dont Convert”

  1. Linda
    Are you aware that your rss feed link at the top of this page doesnt work? It gives an error – we really wanted to syndicate your rss feed at our new site.. http://ecommerce-sites.ning.com – it will open on Mothers Day 2009.
    but cannot because it fails.
    Please fix it and we will add you!

    • Hi Michelle,

      The feed button is working just fine on our end. I’m sorry but I’d appreciate if you don’t syndicate our stuff on your site – nothing personal. This feed gets scraped a lot and I don’t think it’s fair that other sites benefit from my hard work — this blog is property of Elastic Path Software. Thanks for understanding. I don’t mind links to our posts but not full or partial posting of the article.

  2. Dave says:

    I would add a number 4 to the list : Know your numbers : You can get a sale and lose some money at the frontend but if you know the lifetime (yeartime,etc) value of a customer and you know he will be coming back 2-3 times/year you can still make money with ‘bad converting’ keywords

  3. We are just starting our ppc campaign and will keep this bookmarked for future reference. We are pushing for a stronger organic ranking but ppc is the way to go until organic picks up some steam. thanks for sharing the info.

  4. rather than concentrating only on ‘assist’ keywords, consider site actions that are leading to your end conversion. For example, are people who watch a video or register on the site more likely to convert? If yes, then buy keywords that drive people toward those actions. Identifying the relationships between those site actions can help drive conversion rate.

  5. [...] Linda Bustos is up first, suggesting “failed” keywords contribute to the sales funnel and may help more than they hurt. GrokDotCom’s Jeffrey Eisenberg says that no matter how much [...]

  6. MartinHN says:

    I was wondering if there’s any website that sells a very small service that let’s you provide your PPC account credentials and analyzes your keywords and gives you the list of keywords that didn’t convert for the last x days?

  7. I’ve heard this topic discussed in depth recently and I absolute agree. Managing bids by only looking at the conversion count is misleading. I blame lack of tools for the lack of information in the hands of users. The SE and/or bid management tool companies need to get on the ball with attribution management.

Leave a Reply

© 2014 Get Elastic Ecommerce Blog. All rights reserved. Site Admin · Entries RSS · Comments RSS