Tracking Onsite Search

Onsite search is a topic that has been slowly gaining attention in the web analytics community over the last few years. However, I believe this is one of the most often overlooked areas of rapid and immediate improvement (low hanging fruit) possible for many sites. First, many visitors leverage onsite search as their preferred method of navigating the site. Additionally, the data discovered from analysis of onsite search activity can often provide valuable insights into the motivations of your site's users that could not otherwise be collected and analyzed.
The following discussion will focus on what steps need to be taken to establish a solid set of key performance indicators for evaluating the success of onsite search and to provide an overview of the tools and options available.

Important Measures and Key Performance Indicators (KPI) for Onsite Search
Below is a list of measures and KPI's that Stratigent has successfully used in various situations to help our clients optimize their onsite search activities. This list should not be considered to be an exhaustive list nor should it be assumed that everything on this list will apply to all companies. Instead, this list is meant to be a starting point in the process of creating the appropriate measures and KPI's for evaluating your onsite search features:

• Average number of searches per session
• % of searches that do not return a result
• % of search sessions that click on search results
• % of sessions that exit from search results pages
• Conversion ratio of visitors who use onsite search
• Top search terms and corresponding conversion rates
• Top keyword groupings
• Top decomposed keywords
• Top keywords that do not return search results
• Top locations on the site where onsite search is accessed from
• Number of searches per session / number of items added from search results
• Session duration for all sessions that included searches
• Average time spent on search results pages
• Average time spent before searching
• Average order value for customers who used the onsite search
• After evaluating these metrics for all of your site traffic, you may be able to gain additional insight by also evaluating your visitor segments. Armed with this information, you may be able to customize the search experience to best serve your visitors.

How to Track

In order to successfully track your visitor's use of onsite search you will need to leverage one of three main methods that are currently available.

• Using your web analytics application: This is generally the most comprehensive option. It allows you to track the visitor's complete session and combine this data with the search data to compile a complete picture of the visitors and their sessions. Normally, this involves either placing tracking code within the search page or modifying the search application to make the needed variables available within the URL in the query string of the request.
   
• Leveraging the data contained within your search logs: This method requires a fairly flexible web analytics application to accept the log files or the creation of custom database to conduct the analysis. However, this approach can sometimes streamline the data collection process and the ongoing management issues involved with option 1. In most cases it will require significant expertise to integrate the onsite search behavior with the actual customer ID and purchase history. It is a slightly less common method of conducting analysis.
   
• Leveraging a dedicated search analytics application: Most onsite search engines feature some type of internal reporting that can be leveraged to supply the data needed to measure some of the KPI's outlined above. A common obstacle is that it is often impossible or impractical to relate this data to the overall visitors of the site and fully correlate a visitor's lifetime value behaviors with their respective onsite search behaviors. In addition to the standard reporting that most onsite search applications provide, Mondosoft (www.mondosoft.com) makes a specific onsite search analytics model (Behavior Tracking) which plugs into many of the leading search engines. It allows for the easy analysis of a majority of the high level measures that are outlined above and has tools to assist with the overall optimization of the onsite search application's performance.

In addition to the information above you can check out Hurol Inan's newly published book: "Search Analytics - A Guide to Analyzing and Optimizing Website Search Engines" It approaches the topic from an academic perspective and contains the results of a very insightful survey that was conducted in 2005. Overall, I highly recommend this book as I believe the author and I share a common viewpoint: the optimization of onsite search is often undervalued, despite the fact that it is a highly productive area to focus resources in the quest for a competitive advantage online.