How Advanced SEMs Roll: Search Query Mining

Big data is everywhere. It’s become a fashionable topic du jour, and in an industry like search engine marketing, where data is everything, that trend holds especially true. One of my favorite quotes on the importance of data to the future of business is from Google’s Chief Economist, Hal Varian, “Datarati are companies that have the edge in consumer data insight... Data is ubiquitous and cheap, analytical ability is scarce... The sexiest job in the next ten years will be statistician.”

Though I’m no statistician, per se, I have found myself on dozens of occasions attempting to explain my line of work to friends and family. It’s easy for the lay observer to glance at a search results page, eyeball the text ads and dismiss them as simplistic. “Oh, those little ads on the page? That’s you?” has become a common type of response.

Yes, that is me (and several thousand others like me), vying for attention, clicks and conversions among the more than 12 billion searches performed each month across major search engines. “Those little ads” are driven by an incredibly complex algorithm that determines which advertiser’s ad is shown based on relevance, historical performance, keyword bid, geographic targets, time of day/day of week -- all of which is calculated in 1/100th of a second.



Though marketing smarts are still prerequisites to success, competing in the “little ads” industry has become a decidedly data-driven endeavor.  Winning in paid search today means having a complete understanding of the data returned by search campaigns. One of the best opportunities for search marketers to produce outstanding results versus the competitive set, is through a methodical analysis of keyword versus query data. That analysis is called search query mining.

Last spring I authored a column, “The Call for Smarter Search Analytics," where I identified search query mining as crucial for SEMs. In that piece I noted, “Query mining is the process of identifying raw queries which were mapped to keywords within the search auction, and then extracting long tail derivatives and negative keywords to be explicitly introduced across the programs to enhance overall performance. This is an essential tactic for advertisers who rely on broad match keyword portfolios or are launching new programs.

Think of query mining as a way to help eliminate the unqualified noise, while enhancing the keyword portfolio with more precise phrase and exact match keyword targets.”

In the time since that column was published, I’ve performed a sweep of query mining solutions available to advertisers today. There’s the default Google AdWords’ offering, which allows an export of raw user queries and the keywords and match types they were paired with. ClickEquations (which has recently been acquired by Acquisio) also has a very slick Keyword Zoom feature that allows some small-scale query mining natively within its UI.

But the solution I have become most enamored with is QueryMiner, a Web-based tool that can easily analyze several hundred thousand rows of data in a matter of minutes. I connected recently with QueryMiner founder Chad Summerhill to get his take on the importance of search query mining and vision for how SEMs will utilize technologies like his. (Disclaimer: I do NOT have any personal or professional involvement in QueryMiner.)

RD: Why did you decide to create QueryMiner?

CS: Because search query mining at scale is extremely hard. It’s difficult to recognize patterns with the naked eye, or even [Microsoft] Excel, for that matter. When you have more than a screen’s worth of data, identifying patterns is impossible.

RD: What is the value of using a solution like QueryMiner, and how does the software work, specifically?

CS: QueryMiner promises to expose hard-to-find keywords and patterns among long-tail search queries. First and foremost, it surfaces problems with AdWords campaigns. The advertiser then has three options – either write better ads, produce better landing pages, or add the query as a campaign-level negative keyword.

As far as product functionality, QueryMiner utilizes on-site conversions as a proxy for value. We unearth queries that are worthy of investigation based on their performance versus the campaign’s mean average; and that performance is determined by on-site conversion data.

RD: Is QueryMiner focused on long-tail or campaign-level negative keyword identification?

CS: Both really, though our solution really shines in identifying campaign-level negatives. We will soon be releasing a new “keyword expansion” module which will allow our users to leverage the tool to build out keyword-thin campaigns. Keyword expansion will additionally identify themes across user queries that can be used as proof points for new advertising copy… the applications for this data are really limitless, in our opinion.

CS: Google’s opinion of relevance and our opinion of relevance are seriously different. A lot of irrelevant search queries get paired to broad match type keywords. Without sound strategies and the right tools, advertisers will waste a high percentage of spend on unqualified traffic.

There’s no reason why any advertiser should be without query mining technology. We offer free analyses to demonstrate how much wasted spend exists, even across the tightest-managed campaigns.


Next story loading loading..