What are people searching for on our site?

It's a simple question, but the answer is HUGELY challenging.  In library-land, our online users come visit our web presence and swiftly leave to visit thousands of sites, services and information sources. We're full of info riches, so sometimes answering a simple question just isn't an easy feat, it's a headache.

At U of M, there are currently six search forms on our library homepage:

• Site search
• MNCat (library catalog)
• Article search (EBSCO, PubMed, etc.)
• Journals search (MetaLib or Primo)
• Video and Media (scoped Primo search)
• Course Resources

Each of these search forms will move a user from our website into another application or vendor product. To answer the question of what people are searching for, we need to dive into data held in at least six different places.  Not all of these apps/products have Google Analytics (GA) attached.  Some of them might record the user query in a database, while others will only have the query in their log files.

It's a bit messy, so we've worked to make it much simpler.  Our first open source project on GitHub is called GA Track Form Submit.

Using this code, and taking advantage of event tracking in GA, when a user submits a form where we've added "ga-track" as a class attribute, we'll capture the user query, the radio or drop-down input values and record them (as JSON) along with the Form Submit event in GA.

It's important to note we DO NOT capture any personal information here -- our GA account anonymizes IPs, you know.  We're just recording the form input values for internal use.

For a simple form like our library catalog:

<form name="mncatplus" class="ga-track"...>
  <input name="request" type="text"/>
  <input type="image" src="../path"/>
  <input type="radio" name="type" value="keyword"/>
  <input type="radio" name="type" value="title"/>
  <input type="radio" name="type" value="author"/>
</form>

We'll push an event to GA that records info like this:

Form

Submit

{form: "mncatplus", request: "hemingway", image: "Go", type: "author"}

Using this technique, as time goes by, we'll gain a great sense for the most common queries people perform on the site.  But more importantly, we will begin to understand the queries we handle poorly and better focus our efforts to improve the user experience.

So... what are people searching for on our site? Just give us a little bit of time to capture the data and we'll let you know.