Keyword phrase research is perhaps the most important part of any Search Engine Marketing campaign. The ability to pinpoint the phrases used by searchers seeking a product or service is invaluable both in paid Search and Organic SEO. Unfortunately, the majority of the available services are lacking when it comes to bet-the-farm-on-it certainty of search numbers and predictions.
Most SEOs and Paid Search marketers will cite WordTracker is the default best available software out there, although some competitors are starting to make headway, such as Trellian's Keyword Discovery tool.
Members of popular search engine marketing forums such as the SEW Forums consistently look for ways to find keywords in an automated fashion, as evidenced by yet another post today asking for advice on a particular keyword research tool.
The consensus in these types of threads is that WordTracker is the most close to accurate, and we in fact use it at our agency. Overture (now Yahoo! Search marketing) has had a tool out there for perhaps the longest time (although it is reportedly down at this time), but savvy search marketers have known that this tool is subject to inaccuracy due to automated searches included in the count, plus the fact that it only includes searches conducted in the Yahoo! network.
Various search marketers have come up with ways to try and better estimate keyword totals, including Aaron Wall of seobook.com. He and others have used the Overture tool as the basis for the total predictions, and added a percentage to that number to account for Google and other search engine users. Google also has a keyword prediction tool, but it does not provide a number, only listing possible related keywords.
Trellian's tool is interesting, because they claim to get their data directly from ISPs. The main argument against them has generally been that the ISPs may not be the “best” of the 7000+ available in the U.S, alone these days.
Until the search engines themselves are willing to give search marketers direct access to their search records, the most accurate way to predict future searches of terms has been to pay for broad match keyword terms in all sponsored search portals for a period of time, with an unlimited budget. Given the impression data derived from these accounts (using the search network numbers alone and omitting the contextual networks), as well as scouring the keyword-specific referrer logs of the target sites to find the long tail terms, one can make the most likely prediction. In cases here this has been done until a statistically confident sample is reached, predictions by Overture and WordTracker have proven to be inaccurate.
The answer? Use a combination of the above methodologies/tools, as well as someone with insight into the particular industry and a sample of likely searchers to get the best possible prediction of future searcher behavior. No one tool or method is likely to be available any time soon to help with this important task.
Twitter Canada MD Kirstine Stewart to Keynote Toronto
ClickZ Live Toronto (May 14-16) is a new event addressing the rapidly changing landscape that digital marketers face. The agenda focuses on customer engagement and attaining maximum ROI through online marketing efforts across paid, owned & earned media. Register now and save!