Keyword Testing and Expansion Done Right, Part 1

Keyword expansion. You hear it at shows, your engine reps remind you about it, entire sessions at Search Engine Strategies are devoted to it, and your SEM (define) agency also chimes in, especially if you have a seasonal business or your product/service mix is changing.

Everyone says you must expand your keyword list. If you don’t, you’re missing opportunities the competition is profitably exploiting. Those advocating for keyword expansion may be right in recommending an expansion, or you may have already completed so many expansions over the years that very few keywords rally are missing from your campaign.

Keyword expansion gets a lot of attention. If you find a new keyword and it can deliver significant incremental high-quality traffic to your site, your business will benefit for as long as it remains a profitable part of your overall campaign. There are right and wrong ways to add keywords to a campaign, however, because every keyword, new or old, must be tested to determine if it can become an important profit driver for your business. If it can, further testing helps determine how best to maximize the positive impact each keyword has on your business.

In this two-part series, I’ll cover the right and wrong ways to discover, test, and implement new keywords. My objective is to help you find the winners fastest while determining how to eliminate the losers.

Keyword Discovery

Keyword discovery has been covered before here and in other columns; however, there are some unconventional ways to generate keywords that can be added to your standard methods of using engine-supplied keyword research tools.

Spider-based keyword extraction tools have been around for a while, for example, and can be used to generate keywords from your site. The less conventional use of these tools is to send the spider through your competitors’ sites, one by one. Clearly the more accurate, usable list is the one your site generates. But depending on how closely your business matches that of your competitors, their lists may be useful. The quality of spider-generated lists is particularly high if the sites (yours or theirs) have good SEO (define). Remember to take trademark ownership into consideration when considering keyword use, and be sure to follow keyword use editorial guidelines. Note: There may be ethical, or even legal, ramifications in crawling another person’s site.

We have access to tools built with data licensed from comScore that provide insight into keywords actually used to find and visit competitive sites, as well as a client site. We find this data superior in many ways to spider-generated keyword lists. Although the keyword lists are smaller, they focus our efforts on keywords that are much more likely to matter (and be searched). This is particularly true in respect to phrases. Spider-based keyword-generation tools generally extract individual keywords, which don’t take into account how people search.

Any list you generate should be reevaluated in respect to stemming (plurals and other forms of the word), although many engines automatically expand searches to include stemmed versions of the keyword (in both organic/algorithmic results and paid listings). The closest match is generally considered most relevant, and the search engine only puts the exact match in boldface. Bolded keywords in titles and descriptions increase visibility and CTR (define), increasingly important in all engines.

Keyword Implementation

Regardless of where new keyword suggestions originated, those keywords must be rolled into your campaign in a structured, intelligent way. Otherwise, you do a ton of work without maximizing the opportunities inherent in the new keywords. After deciding a keyword or keyword phrase is worth testing, you must decide on campaign structure, a very different decision in each of the major engines. In Google, for example, you must decide whether to add the keyword to an existing campaign or AdGroup, or to start fresh. There are pros and cons to each choice.

In part two, we’ll continue discussing implementing and testing new search campaign keywords, including how to structure a campaign around the new words, how to select landing pages, and how high to set initial bid prices. I’ll also cover how to deal with the data scarcity that occurs when a keyword has low click and impression volume (even if it has a high CTR).

Meet Kevin at Search Engine Strategies in Toronto, April 25-26, 2006.

Want more search information? ClickZ SEM Archives contain all our search columns, organized by topic.

Related reading