Don’t Let Machines Write Your Keyword Lists
As anyone who has seen the Terminator or Matrix movies can attest, if machines rule the world humans are in deep trouble. Similarly, if you allow machines to control the PPC keyword creation process you are guaranteed to both leave opportunity on the table and flush money down the toilet.
The reason is simple: language is really hard. The best language software in the world, Google’s, is awful with language. My last SEL post related to bid management in PPC; the AdSense ad running next to it was for a contracting firm looking to bid on construction projects.
Consider the following:
Example 1: Playboy Classic Cigarette Girl Adult Women’s Costume
A smart human would tread carefully here. A dumb robot would throw out some fairly eye-opening keywords. Now, in the real world someone would catch this, right? Maybe. In fact, we found that for this retailer, an agency using robotic keyword construction was running the keyword “Playboy” as a search term landing on this page and ran the same keyword on content match!
Example 2: Mountain House Beef Stew
Now, the smart analyst realizes that this is backpacking food and will attach the appropriate modifiers whether or not those words appear on the page. The robot? Not so much.
Machines will also put together words that don’t make any sense as keywords, and carve up phrases that should not be separated. Some example of the nonsense we’ve seen from “world class keyword generating tools”: “nnn”, “33”, “longsleeve”, “clothes”, “women”, and many more. The retailer for whom these were built sells sporting goods. These lists not only came with landing pages, but bid suggestions, too!
Granted, for websites with 20,000 products, smart humans can’t keep up with keyword construction manually. However, they can use tools that allow humans to control the process in a way that both prevents wasteful spending and capitalizes on the analyst’s understanding of industry jargon, appropriate synonyms, common misspellings, and likely abbreviations to build a much more robust list.
More importantly, while a controlled use of tools is valuable to build out and maintain the long tail, there is no substitute for human understanding in the more general, higher traffic category and subcategory terms. Machines leave gaping holes and invariably use synonyms that are completely inappropriate – think about that “costume jewelry” category. Cleaning this dangerous stuff out of a massive list takes more time than building the list carefully in the first place through machine assisted but human driven processes.
Finally, a good many folks talk about the power of competitive analysis and seek to find their competitors’ keyword lists. However, the fact that a given keyword works for your competitor doesn’t mean it will work for you. They may have a wider selection or lower prices, and the keywords will often relate to brands and products you don’t even carry. Moreover, you really don’t even know if these terms “work” for your competitors; they may be hemorrhaging money on them for branding purposes.
Your time is best spent focused on what works for you and the best ways to describe your products. Don’t let the machines run this process; it’s far too important for that.
Opinions expressed in this article are those of the guest author and not necessarily Search Engine Land. Staff authors are listed here.