Site Search Analytics – what users want from your site, in their own words.
A little goes a long way. The Zipf distribution, shows that investment in results for most frequent search terms can go a long way. Focus on small number of most popular search terms, work on those, then pick another small chunk.
Users stop (and maybe search) when they lose the scent of information.
Search Terms or Queries
Interesting parallel between meta data and what people search for on your site. Meta data distills content found on a page, and search queries distill what user is looking for on the page.
Cluster your queries and determine which terms seem to mean the same thing? This will enable you to learn more about users’ language and tone. This also gives you a sense what might go into taxonomies, meta data and page content. Can give insight on what types of what facets you might consider for your search.
Determine content type
What kind of content is a user searching for when they use various search terms. Applications, news, reference content, contact info, instructions. This allows you to create a ranked list by content type which allows you to better understand contextual navigation. Content type can also be use to improve search performance, using content type as a search facet. Understand seasonality to determine editorial schedules.
What is unique about your queries? Product codes, course codes, proper nouns. If you see product names or unique queries you can identify, try to give search results more context, make your query string suggest sorting. When queries begin to spike, compare to current content and determine if there might be a need for more content on this topic.
You might use cookies for audience segmentation, so you can break down search and needs by different audiences. Are queries for content that doesn’t exist on your site? Try to determine what’s commonly important to all audiences.
Use numbers for proof you need to de-jargonify
Are people searching for jargon, where do search terms cross your company jargon? Focus on naming content what it is.
Key is to constantly tune your site’s performance. If you’re always looking at what users want from you, you have more of a chance of satisfying them than you do spending x number of dollars on a redesign every x number of years.
Goal-Driven Analysis
Remember to analyze failed search terms. Look at successes and failures in processes.
What are top pages where user searches from? What are the search terms for each page? Is page titled in an unclear or ambiguous way? Are you missing content from the page?
Test performance of most frequent queries. Measure using original two sets of metrics.
- Relevance (or queries with the best result).
- Precision (assess each result on a scale or relevant, near, misplaced, or irrelevant).
“Relevance will only work if you have an idea of the best results”
Getting Started with Site Search Analytics
If you use Google Analytics, you must enable search query collection in the settings.
Check out:


Recent Comments