Interview with Wordtracker Founder Andy Mindel
By Scott Buresh 2003-09-04

Search engine optimization experts have long known that choosing keyphrases is the single most important step in a search engine optimization campaign. Not long ago, however, much of the keyphrase selection process was done by trial and error, or relied upon tools that provided data of questionable value.

Then came Wordtracker, a powerful keyphrase evaluation tool that gave popularity figures for individual search terms based upon actual search engine activity. Many search engine optimization experts, this author included, immediately found Wordtracker research to be an invaluable part of the search engine optimization process.

I recently interviewed Andy Mindel, Wordtracker’s founder and president, to learn more about the London-based company and the service that it provides.

Scott Buresh: Tell us a little bit about your background and the origins of Wordtracker.

Andy Mindel: We initially entered the search engine positioning field. Whilst carrying out some work for ourselves and friends we only had access to the Overture suggestion tool. We achieved a number of top ten rankings and waited for the traffic to flow. The Overture suggestion tool reported hundreds of visitors a day from each keyword, so we braced ourselves. Whilst testing we realized something must be wrong with the search predictions. We were expecting over ten thousand visitors for the week and we received about ten. So we carried out our own research and utilized a keyword source available on the web at the time. The source was taken from metacrawlers and the results we received from these engines were very different. We started to use these words and found them to be far more accurate. So initially a simple tool was created for our own use. We then started to sell this tool to other positioning experts, who requested additional features. Slowly we integrated all these and created what you see today.

Scott Buresh: Where exactly does the popularity data in the Wordtracker Database come from?

Andy Mindel: Popularity data is taken from the largest metacrawlers on the web, Dogpile and Metacrawler. After much testing, we found that results from these engines were more accurate than other sources. We examined keywords from other engines and noticed a distortion from position checkers and hard coded queries. One thing we notice is that the top keywords always fall into a certain pattern – these usually consist of google, hotmail, sex, mp3, etc. When this pattern changes then we know something’s up- and often it’s because the engine’s database is being used at another site (for example gambling or shopping sites).

Scott Buresh: For each phrase entered, the Wordtracker database gives both a "count" and "predict" number. For those unfamiliar with Wordtracker, can you explain exactly what these figures represent?

Andy Mindel: In a nutshell, the count is the total number of times that a keyword has been looked up in the past 60 days using our 350 million keyword database. This database is the *complete* log of all requests made at the Metacrawler/Dogpile Metacrawlers (we don't use search engines because of software robots/position checkers distorting the results but the lookups are very similar).

The predict column, however, attempts to predict the total searches of that keyword in the next 24 hour period for all search engines/directories/pay per bids to give you a rough idea of whether it's a good choice or not. It utilizes the predicted total number of searches made on the net each day (see the first article referenced below). Sometimes the count/predict will be very similar. This is because the predicted total number of daily searches may be the same as our total database size.

The count is from keywords in our database. This database is taken from the major metacrawlers which only get about 2/3% of the total search market (this number constantly changes). Using the formula above we work out a predict that represents all engines on the web. When we work out this number it is often much higher than the count, as we are taking into account all the searches on the web and not just the major metacrawlers.

Please look to the following sources for detailed explanations of count and predict:

 

• 1) http://www.wordtracker.com/articles, in particular 'How many queries are performed on the web each day' which explains how we calculate count and predict.
• 2) Click on the count/predict columns when you do a search on Wordtracker. Or go directly to http://www.wordtracker.com/help/counthelp.html

   

Scott Buresh: How has the increased monetization of search changed Wordtracker, if at all?

Andy Mindel: In our current version we offer the ability to find niches within the PPC engines. This will be expanded upon in Wordtracker’s new release. There is a much greater importance placed on these engines and Wordtracker will be catering for these needs.

Scott Buresh: Do your customers consist primarily of search engine optimization experts, or is there a wide mix?

Andy Mindel: Our mix has been getting wider and we now cater for marketing sites and those just starting out looking for the idea and inspiration for a new product. However our predominant base is still positioning specialists.

Scott Buresh: What are the limitations of the free Wordtracker trial (as opposed to the full product)?

Andy Mindel: Here is a summary:

 

• a) The full Wordtracker system returns 300 related words per search (and another 300 from the thesaurus) whilst the free trial returns 15.
• b) The full database returns up to 500 keywords for each popularity search. You may also enter any number of your own keywords. The free trial returns 15.
• c) You can store up to 5,000 keywords with the full membership. You also clear your basket and delete your last keyword. With the free trial you can store up to 30 keywords.
• d) You can find out how many people misspell your chosen keywords (e.g. Altivista, alttavista). This feature is not available on the free trial.
• e) With the full subscription we provide you with a number of different ways to search our keyword database - including word stemming (tie, ties, tieknot), upper and lower case separation or compression, and pluralisation. You also have the option to include or exclude adult search terms.
• f) The full database allows you five projects for each account. They can be implemented as you need them, cleared of keywords, renamed or deleted, the choice is yours. This option is not available on the free trial.
• g) There are two reports, the short term top 1000 which shows you data from the last 36 hours, and the long term top 1000 which shows you data from the last 8 weeks. This allows you to spot trends, sudden surges of interest or keywords which stick around consistently from day to day.

Scott Buresh: What future changes or additions do you anticipate for Wordtracker, in the short and long term?

Andy Mindel: For the last year and a half we have been working on a new and updated Wordtracker. We have rewritten the infrastructure to make its workings faster and smoother. We have been receiving suggestions over this period from our users, all these will be implemented. We are also releasing a webservices side to Wordtracker which will allow users to connect to all parts of Wordtracker via an API. There have been big changes, far too many to list. We are hoping to release Wordtracker in Beta mode early next year and then we will implement any other changes and suggestions throughout this period.

Scott Buresh: Thanks for taking the time for this interview, and good luck with the upcoming release.

Andy Mindel: No Worries.

Scott Buresh is managing partner of Medium Blue Internet Marketing. His articles have appeared in numerous publications, including ZDNet, WebProNews, MarketingProfs, DarwinMag, SiteProNews, PromotionData, and Search Engine Guide. Medium Blue is an Atlanta search engine optimization company that works with clients all over the country.