Wednesday, March 19, 2008

How DaniWeb IT Discussion Community Uses Google for Site Search

The DaniWeb IT Discussion Community recently implemented Custom Search Business Edition on its website. We asked Dani Horowitz, the owner of the site, about their business and experience using Custom Search.

Google: Please give us an overview of your existing business as well as some information around the business pain that caused you to seek out Google Custom Search Business Edition.

Dani: DaniWeb IT Discussion Community is an interactive, online community for everyone in the IT industry from software developers to web developers to Internet marketers to tech enthusiasts. A quarter of a million members strong, and increasing by nearly 1,000 a day, we've been around for about four years with features such as forums, blogs, IT news, a code snippet library, and more.

As with most large forums, the all-important site search has always been a concern for us. We outgrew our forum system's internal search feature a couple of years ago. Since then, we have been playing with multiple options such as MySQL Fulltext Search and SphinxSearch. When Google offered a SOAP-based API that would allow us to outsource search while still providing Google-quality results, I thought it was the end to our problems. However, the 1,000 queries per day limit was frustrating, and the program was closed down just as we began to exceed the limit. We were forced to migrate back to SphinxSearch and, as our web traffic grew, eventually offloaded it onto our mail server. While SphinxSearch was rather fast, our main concern was how literal the results were, often returning no results at all despite a significant amount of information within our database on the query submitted. Obviously this wasn't an ideal situation. We also played around with AdSense for Search, and while the results were spot on and it was nice to make money on search result pages, it wasn't as flexible as we needed.

Somewhere along the way I heard that Google was now offering XML-based results, for a fee, so I decided to search around Google's API and Product Info pages for more information. It seemed as if XML results would be comparable to the SOAP-based results Google used to offer. I soon discovered Google Custom Search Business Edition, but we didn't seem to fit into its default pricing model. I called up my Google AdSense rep who hooked me up with a CSBE representative who was able to get us live before the end of the same week, and the rest is history.
Google: How was the implementation process?Dani: The implementation process couldn't have been easier. I was able to gain access to the XML-based results roughly a day after signing the contract. Once that happened, it didn't take more than an hour to write a PHP script to fetch the results. However, I had wanted to take that a step further, and it took another few hours to customize the results by integrating them with DaniWeb's internal database.

Google: What technology did Custom Search Business edition replace and why?

Dani: Google replaced every other internal search script or algorithm that I was able to get my hands on or code myself over the past few years, from Google's SOAP-based API, to MySQL Fulltext search, to SphinxSearch, to AdSense for Search.

Google: What benefits have you derived from using CSBE?Dani: Pretty much the only direct benefit that I was looking for in CSBE was XML-based results. This allows us to have the same amount of flexibility as we did with in house solutions such as MySQL Fulltext and SphinxSearch, while still having amazingly accurate Google-powered results. This gives us the ability to not just spit out search results, but, as a database-driven website, actually heavily integrate the results with information we already know about each page.

As a huge site which covers just so much content, navigation and usability has always been a huge problem for us. It's great to be able to actually take Google results and have the flexibility to replace the URLs of each result with a site-specific navigational breadcrumb. This allows a searcher to always be aware of exactly where they are in the hierarchy of our website. We've also been able to filter out navigational pages from the search results, ensuring that searchers are always taken directly to the content-based forum thread or article pages that directly contain the information they're looking for.

Something else that we've been able to implement, thanks to CSBE's flexibility, is the ability to highlight search keywords and keyphrases on the landing page for each search result. This allows a searcher to not only know that what they are looking for is contained somewhere on the page, but specifically where on the page it is.

Additionally, we're using Google AdSense custom search units on our search result pages, thanks to my AdSense rep, so that plays a major role in offsetting the cost for us. We also use Google Analytics, and their new Search feature makes it really easy to see exactly how search performs for us. I can now say with confidence, thanks to CSBE and Google Analytics, that visitors who use our site search feature visit an average of 10 times the number of pages and spend ten times as long on the site. And thanks to a few thousand people searching a day, it adds up to a noticeable increase in retention rate across the board.

Google: What features of CSBE do you appreciate most and why?

Dani: My favorite features of CSBE include the high quality of the results, the ability to include and exclude individual pages and directories by using wildcards (although RegEx would be a nice future addition), XML-based results, a flexible pricing model, and integration with Google AdSense for Search (thanks to my AdSense rep). If any one of these features were missing, CSBE would most likely not suit my needs and/or be within my budget.

Google: What's been the overall impact of using Google CSBE on your website?

Dani: CSBE has only been running for a couple of days so far, but the results have been overwhelmingly positive. I'm excited as I look to DaniWeb's future without the constant worry that an internal site search is a greater problem the bigger we grow. Being able to have an outsourced solution with all of the benefits of an in house solution definitely is one less thing to have to worry about. It's also amazing to witness that Google provides better results than search algorithms that have direct access to our content database.