Relevance Versus G-Relevance
Relevance is a pivotal, key concept of AdWords, Google's flagship advertising program. The California-based giant emphasizes at various places that its advertising system is designed to ensure positive user experiences by displaying relevant ads to searchers. However, what relevant means has been and continues to be a headache for thousands of advertisers worldwide.
According to the Merriam-Webster dictionary, the adjective relevant means "having significant and demonstrable bearing on the matter at hand," which might well prove an important but elusive definition for our rather practical purposes. On the other hand, the AdWords Glossary states, "Relevance refers to the usefulness of information to a user (such as an ad, keyword, or landing page)." Could an automated system ever come anywhere close to understanding the psychological notion associated with relevance and usefulness, or is it just walking around in circles with definitions?
Google does not disclose details about precisely how relevance should be interpreted; and one of the first explanatory notes came from AdWordsPro, a manager of the AdWords Team, in the official AdWords forum last year. A member wanted to improve the ad position for his keyword "stuffer," associated with his website featuring stuffers, i.e., field goose decoys made from a taxidermist-preserved skin. AdWordsPro says, "Taking the keyword 'stuffer' as an example, if you search for that word on Google you will see that there are over two million search results returned. Looking at the first 30 or so, it is very clear that the most common usage of the word 'stuffer' is in the context of either a 'sausage stuffer' or a 'stocking stuffer,' neither of which pertain at all to your product. This tells me that if you used the keyword 'stuffer,' your ad would mostly be seen by folks who are NOT looking for what you have. Instead they are looking for something that is 100% unrelated to what you offer."
Based on the above hints we might have the courage to continue like this. If you Google the search term "chevron," all the natural listings displayed on your first search results page will be associated with the oil company. However, the dictionary's primary meaning of "chevron" is a badge consisting of stripes meeting at an angle, worn on the sleeve by noncommissioned officers, police officers, etc., as an indication of rank, service, or the like.
Now, if you create a campaign to promote your website that markets such striped badges, you might, of course, want to use the keyword "chevron." Based on the dictionary's meaning, you might set up very tight and relevant adverts and landing pages for this keyword. However, no matter what you do, Google will find your keyword "chevron" irrelevant to your landing page, because of the accrued statistics. In plain English, Google likely will think that users searching for "chevron" are interested in nothing but the energy giant. Chances are that's what historical data indicate.
In a recent ad group level experiment we used two keywords, "company" and "companies." Being a singular and a plural, their close semantic relationship was more than manifest. This is how the system evaluated them.
Keyword: "company" - "companies"
Quality Score: 3/10 - 7/10
Relevance: Poor - No Problem
All in all, we have good reasons to believe that it's not dictionary or semantic relevance that matters, it's what we should term G-relevance. The calculations are based on historical Google-wide CTR stats along with similar factors, and not on preliminary mapping of semantic fields. This simplified version of artificial intelligence is a plausible explanation of why we often see different relevance and QS values for semantically highly related keywords. Chances are good that Google has differing stats on those words when associated with adverts and websites similar to the ones we created for the keywords in question.
From time to time, when checking on the quality scores of your keywords, you may find yourself in a situation similar to one of the scenarios described above. You may assume the story is about Google-wide stats. Remember, chances are, it's not worth wondering about it a lot. Instead you're advised to focus your attention on trying to identify new keywords that are equally favored by Google and meeting your requirements.
B Lakatos is one of the Top Contributors selected by Google at the official AdWords Help forum. In cooperation with a few other Top Contributors, B Lakatos runs a website that helps people looking for practical assistance with AdWords. A real must have for newbies and beginners.
11:10 | ป้ายกำกับ: G-Relevance, Relevance, Versus | 0 Comments