THE RKGBLOG

WMQA – Proper Use of Google Matchtypes

Dear WMQA —

I’ve read bloggers on line saying to only run terms on Google using exact match. And I’ve read the opposite — only run terms on broad. Which idea is more correct?

Good question, as we’ve discussed recently (Broad Match a Change for the Worse, Broad Match Controls), matchtypes are an important and complex piece of the puzzle. We believe the answer to your question is to use both broad and exact matchtypes. Done correctly, you can capture the benefits of more precise targeting, while also getting the most exposure on likely relevant searches that don’t match your keywords exactly.

It’s no surprise that exact match keyword variations typically hold higher CTRs and CRs since the ads creative and destination url are all geared towards that identical user query. It’s also easier to manage these keywords from an efficiency standpoint for the same reasons. The downside to using exact matchtype exclusively is traffic volume. No advertiser will ever be able to capture all variations of keywords that appropriately lead consumers to their site. In fact, anyone that has studied search query logs can tell you there are thousands of non-exact queries using misplaced characters, superlatives, and an overwhelming amount of misspellings. (Let’s face it, the English language is complex!)

This leads us to the use of broad and even phrase matchtypes. The clear pro of the expanded matchtype versions is gaining incremental qualified traffic. As I mentioned before, it’s impossible to generate every keyword variation, and much of the traffic that would be “lost” is captured by expanded matchtypes. The con is gaining incremental unqualified, unwanted traffic. Google’s systems do their best to determine meaningful, appropriate matches to user’s queries, however these don’t always match the advertiser’s desires. Google understands this effect, and last year released their Search Query Performance Report to help advertisers identify both qualified and unqualified search queries triggering your ads. Run these reports, add unwanted search queries as negatives and add meaningful search queries as new keywords to your campaigns. This will allow you to maximize the visibility into wanted and unwanted traffic.

George is lobbying Google to get more and better controls for matching alternates at today’s Google SEM Council of Advisors meeting.

An interesting note about the use of these two more expansive matchtypes used in conjunction with one another. Google cautions against using broad and phrase matchtypes for the same keyword simultaneously.

Let’s assume you’re running the keyword ‘discount furniture’ on each exact, phrase, and broad matchtypes. The exact version will work appropriately when paired with the phrase or broad matchtype, and Google will choose to serve the exact version when the query is identical to the keyword. When discussing phrase and broad matchtypes competing against one another, it becomes much more difficult for any algorithm to serve one ad version over another. For our ‘discount furniture’ example, suppose the user queries ‘best discount furniture’. Both the phrase and broad matchtype versions of the keyword meet Google’s standards for display. To avoid any confusion in tracking results by keyword and matchtype, use broad or phrase, but avoid the use of both simultaneously.

- WMQA

Technorati Tags:

Comments
8 Responses to “WMQA – Proper Use of Google Matchtypes”
  1. Rehan says:

    With respect to broad match vs. phrase match, Google says at http://adwords.google.com/support/bin/answer.py?hl=en&answer=66292 :
    “The more restrictive match type will always trigger the ad, regardless of CPC bids.”

    Based on that, the phrase matchtype would be the one selected for the ‘best discount furniture’ query.

  2. Marc Adelman says:

    Post for rkg

    Match type

    I join in chorus, that the there should be greater levels of control for broad match.

    KWs are like individual stocks, even down to the match type level for the same string.
    Each KW match type has its own competitive landscape.
    If we further look at KWs as investment products (which they are) we can see that the different match types yield different returns and results for the same KW string.

    Exact is highly targeted, highly search relevant, and hopefully if paired well with a targeted landing page, highly converting.

    Broad has the largest net, capturing the most search queries (potential visitors). Its value is more along the lines of revenue per new unique visitor and almost more importantly, as mentioned in your post, its ability through Search Query or data on referring URLS, to farm alternative KW strings that convert & if you want the whole bucket, also don’t convert (new revenue opportunities).

    Phrase plays a hybrid role between these having limited aspects of the other two match type’s strength.

    If the value and performance of each match type can be uniquely defined, then surly each match type should be treated differently with the aim to accomplish different results from each.

    While exact and phrase are static in their function, Google has made refinements on broad match (or changes dependent on how you view it). So in harmony to RKG’s battle call for deeper control for Adwords, here are a few suggestions for improvements in user control for broad match.

    1. Optimize broad match similar to the optimize option for adtext rotation. Help auto filter irrelevant search queries without the hassle of ever expanding negative builds. Create a function that limits the impressions of broad variations that have a low CTR.

    2. Have a 4 function selection for broad matching

    a. Extended match (everything but the kitchen sink) Canon for Nikon and Nikon for Canon (the anti targeting option).

    b. Full Exposure without extended match (ok a bit better).

    c. KW string in any order, with unlimited character differences per string, but without other words added to the string. This would be aimed at capturing misspellings. More of a hard coded non-dynamic function.

    d. High Relevancy Broad Match (#1 on my wish list): Create a relevance score that looks at the relevancy of the user search query to the content on the advertiser’s landing page and gives a relevance score similar to a quality score. This option would only show impressions for variations above the poor quality range by default and to drill down even further, one could select the range of score to show for (Example: on a scale of 1-10, only show for variations with a relevancy score > 7).

    I would love to know any further suggestion y’all might have on taming broad match.

    Marc

  3. Rehan,

    I chatted with a Google product specialist, and he stated that their system should work in the manner you’ve described, but acknowledged that they might not always be able to properly serve a phrase matchtype over a broad matchtype version of the same keyword.

    I’ll try to have Google clarify the situation a bit further.

  4. Marc Adelman says:

    I join in chorus, that the there should be greater levels of control for broad match.

    KWs are like individual stocks, even down to the match type level for the same string.
    Each KW match type has its own competitive landscape.
    If we further look at KWs as investment products (which they are) we can see that the different match types yield different returns and results for the same KW string.

    Exact is highly targeted, highly search relevant, and hopefully if paired well with a targeted landing page, highly converting.

    Broad has the largest net, capturing the most search queries (potential visitors). Its value is more along the lines of revenue per new unique visitor and almost more importantly, as mentioned in your post, its ability through Search Query or data on referring URLS, to farm alternative KW strings that convert & if you want the whole bucket, also don’t convert (new revenue opportunities).

    Phrase plays a hybrid role between these having limited aspects of the other two match type’s strength.

    If the value and performance of each match type can be uniquely defined, then surly each match type should be treated differently with the aim to accomplish different results from each.

    While exact and phrase are static in their function, Google has made refinements on broad match (or changes dependent on how you view it). So in harmony to RKG’s battle call for deeper control for Adwords, here are a few suggestions for improvements in user control for broad match.

    1. Optimize broad match similar to the optimize option for adtext rotation. Help auto filter irrelevant search queries without the hassle of ever expanding negative builds. Create a function that limits the impressions of broad variations that have a low CTR.
    2. Have a 4 function selection for broad matching

    a. Extended match (everything but the kitchen sink) Canon for Nikon and Nikon for Canon (the anti targeting option).

    b. Full Exposure without extended match (ok a bit better).

    c. KW string in any order, with unlimited character differences per string, but without other words added to the string. This would be aimed at capturing misspellings. More of a hard coded non-dynamic function.

    d. High Relevancy Broad Match (#1 on my wish list): Create a relevance score that looks at the relevancy of the user search query to the content on the advertiser’s landing page and gives a relevance score similar to a quality score. This option would only show impressions for variations above the poor quality range by default and to drill down even further, one could select the range of score to show for (Example: on a scale of 1-10, only show for variations with a relevancy score > 7).

    I would love to know any further suggestion y’all might have on taming broad match.

    Marc

  5. Indeed, Rehan, this is exactly the bug we brought to Google’s attention. The system doesn’t work the way they intended it to, and we have received assurances that they’re working to fix it.

    http://www.rimmkaufman.com/rkgblog/2009/01/21/google-broad-match/

  6. Marc, your suggestions are fabulous. I will make sure they get in front of the right folks at Google. Bidding by relevance score…love it!

  7. Jeremy Campbell says:

    So can I infer from this article that there would be no problem with including all keywords as all match types, except Google won’t always show a phrase match keyword over the same broad match keyword?

    Are there any implications to this besides that you might pay more for the broad match keyword?

  8. It’s not that you’d pay more for broad matched traffic, it’s that the value of that traffic is less. Launching on all three match types is fine as long as you’re differentiating the bids wisely, bidding more on the exact matched version than the phrase matched and more on the phrase matched than the broad matched. Having all the versions without differentiating the bids serves no purpose, you might as well just have the one broad matched ad.