Leveraging Google Analytics: Helping Users Find What They Think They Want
by Malisa Anderson-Strait
A virtual visit to the library may be initiated for many reasons. Usage data shows the visit is frequently for access to a subscription database or journal article, often after being prompted to pay for an article or report found by searching the free web. A yearlong review of Google Analytics keyword data at Emory University’s Goizueta Business Library (GBL) surfaced many common user search tactics for this type of content. Responding to users’ keyword searches with changes to metadata increases the probability they will find what they want. Additionally, these metadata changes offer librarians an opportunity to redirect users toward alternative relevant content, if the precise content they search for at the library is not available.
|Keyword searches for subscription databases were at the top of list.
The Project That Started It All
GBL has a small and diverse userbase, including 1,200 students—some of whom conduct nearly all their research on a tablet and others who are using older, work-issued laptops. Undergraduate users include junior-level and senior-level business majors. The M.B.A. population represents both full-time students, as well as working professionals in both evening and executive programs. Additionally, GBL works closely with Ph.D. students, faculty, and alumni.
In January 2014, the library launched a university-wide project to re-platform university websites from Drupal to Cascade. This project provided GBL with the opportunity to work with Emory’s team of web design experts to create a website that responded effectively to the variety of business school users’ search and discovery needs. The web team brought a more thoughtful and research-oriented approach to website design than the last few website iterations GBL had undertaken. The process included usability testing with nine user interviews, card sorts, and the development of user personas.
Not surprisingly, the usability results showed that access to subscription research databases was one of the key reasons users come to GBL’s website. Historical Google Analytics data showed that approximately 55% of all site visits were to access database pages, and the usability study reinforced the importance of efficient access. Quickly accessing a known database was key to a successful web visit. The new site was designed to meet this user requirement, with a database quick search box on the front page for known-database searches.
However, a challenge of this quick search box was that it would only return results when the user searched “correctly.” A typo would not produce a friendly Google “Did you mean?” prompt. Instead, a message that no results were found would appear. If the user did not catch the typo (or did not realize one was made), this unsuccessful search experience could lead to frustration, increasing the probability the user would return to Google for an alternative resource. Adding metadata to reflect common imperfect searches was identified as a solution to alleviate this challenge and ensure results were returned.
GBL’s initial approach to selecting metadata for the new site was admittedly less than scientific. First, the business library student workers compiled a list of typos and other terms they felt they or a fellow student would likely use to search. Second, the business librarians added terms reflecting experiences from consulting with students over the years. This two-pronged approach to compiling metadata reflected the many ways a user may be inclined to search for a database. The metadata included vendor names that are sometimes more well-known than database names and alternative common ways to spell a database name (Lexis versus Lexus; Thompson versus Thomson). After adding the metadata, the team returned to adding content to the front end of the website and launched it.
Assumptions vs. Reality
In January 2014, the site went live. For that first year, Google Analytics data was primiarly analyzed for page count hits on the most frequently accessed databases, usage spikes after classes, and similar content and page usage data. Nearing the end of the first year, GBL had more than 15,000 unique keyword searches in Google Analytics, a dataset sufficient for identifying search and find trends. Of the 15,399 keywords, 882 were used more than once. These recurring keywords reinforced that users were searching as expected. Keyword searches for subscription databases were at the top of list. “Hoovers” was searched 1,129 times, and the Hoovers database access page was returned at the top of the results. Other anticipated misspellings (such as the Thomson with a “p”) were also reflected in the keyword analysis and returned the proper database result.
However, the keyword data also revealed several searches that were not represented in the initial metadata creation. These search tactics returned no results, preventing efficient access to the desired content. Review of these unsuccessful keyword searches led to many easy-to-implement changes designed to enhance the user experience.
TACTIC 1: Imperfect Searches Need to Return Results
Partial names and slight misspellings rose to the top of the imperfect search tactics in the Google Analytics keywords review. For example, a simple search for “Marketlin” (Marketline) returned zero results. The same was true for “wet fet” (Wetfeet) and many spelling iterations of Euromonitor. Although Hoovers spelled correctly rose to the top of common searches, the data showed many failed searches for “Hover,” “Hooverds,” and “Hoobers.” A compilation of the frequently searched typos and misspellings were added to the metadata as a means of enhancing the probability of a search returning useful results.
TACTIC 2: Abbreviations— Not Just for Texting Anymore
Acronyms and abbreviations dominate texting, and they manifested as a search tactic in the keyword data. Many “sounds like” snippets and acronyms were represented, including obvious acronyms such as “BSC” for Business Source Complete (54 times) and “BoL” for Book of Lists (38 times). This also included more unique searches such as “SnP” for Standard & Poor’s (21 times).
The original metadata did not anticipate this type of search tactic. It was revised to reflect what the users were actually searching. Based on the popularity of short terms, more were proactively added for many other GBL databases that didn’t have acronyms rise to the top of the keyword data in anticipation of this trend continuing. The web redesign already included many mobile-friendly features such as responsive design, and the addition of short terms to enhance convenience for mobile searching made perfect sense.
TACTIC 3: Where Information Lives
Analytics data also identified keyword searches for business content found within databases—such as “five forces” and “5 forces”—a common framework for industry analysis. Although the students were searching for this framework, they were missing the connection to the database in which that lived, Marketline. Also common were searches for journal names, such as Barron’s and the Atlanta Journal-Constitution (found in Factiva) and Harvard Business Review (found in EBSCO’s Business Source Complete). These common business topics and journals were added to the relevant database’s metadata. Certainly, including a few journal titles in metadata is not a replacement for a robust ejournal search tool such as SFX. However, for commonly requested journal titles, including them in the metadata enhances search success. Admittedly, this tactic is far easier for a business school library with a focused user population and set of terminology than it is for a larger university or public library with a broader range of users to accommodate.
Knowing changes to the metadata would help direct students to the proper database for a journal, the GBL team was concerned about the potential for confusion when a journal name search returned a result for a database. To counter this, several content additions were made to the database descriptions. This included adding common journal names and frameworks found in databases, as well as “formerly called” notes reflecting database name changes so users would understand why they were seeing a particular page returned.
TACTIC 4: Addressing Resources the Library Does Not Provide
The Analytics keywords also revealed searches for many types of content the library does not have available, including textbooks, market research reports found on Google, and databases to which the library does not subscribe. This raised the question of how to communicate to the user about the materials not found, not because of the chosen search strategy, but because the resource isn’t in the library. The team focused on adding explanatory FAQs on the GBL blog—for example, why a specific database is not part of the library’s suite of resources and suggestions for alternative resources. Even if the user does not get the precise content desired as demonstrated by the keyword search, the opportunity to offer alternative resources or at least an explanation as to why something is not part of the library’s collection will hopefully provide the user with a more positive experience than a “no results” message.
TACTIC 5: Strange Searches
The keywords GBL chose to address in the metadata were repetitive and commonly searched, appearing 20 to 50 times in the year’s data. There were many other keywords searched less frequently but still more than once. These searches were not aligned with any obvious research need and were thus not addressed in the metadata and web content changes. Queries such as cereal (twice) and weddings (three times) were amusing, but did not lead to any actionable changes on the site. Similarly, the many unrepeated, single keyword searches were not necessarily individually actionable, but still offered insight as to the search and discovery process. Many of the single keywords looked as if they would best return results from a site search. This confirmed the usability-based decision to make the site search the default on the homepage, instead of the catalog search, as had been previously considered.
Playing Well With Google
GBL’s metadata and web content changes were focused on streamlining the user experience to facilitate a successful visit to the library website. However, a review of the Google Analytics data also showed that the library’s homepage is not the only starting point for users seeking access to a subscription database. Rather, many searches for library databases are originating directly from Google. Of 95,000 user sessions on GBL’s website between September 2014 and September 2015, 34,000 sessions were organic, with users arriving at the website via keyword searches in Google or another search engine. The common search convention is “school name database name,” or the reverse order. Many of the top most common searches identified in the Analytics data reflect the same search tactics as applied directly on the library’s website (e.g., snp Goizueta). The metadata additions enhance the probability of successful results via a user’s preferred starting point, whether that’s Google or the library’s homepage.
Access to subscription databases continues to be the most common reason users access GBL’s website. Analytics data for page views shows database pages remain the top reason users access the library; of GBL’s 314,000 page views from September 2014 to September 2015, 171,000 were for databases. An ongoing review of new keyword searches indicates that users continue to make common typos and create abbreviations for databases. The good news is now these work. Users are directed to the database they seek, rather than a message of no results returned.
During the last several months, GBL has continued to build on learnings from Google Analytics keywords and implemented a more intentional and structured approach to metadata when adding new databases to the website—a vast improvement from the first metadata added prior to launching the site. The addition of acronyms has enhanced efficiency on a broader scale than simply reflecting user search preferences. Asking students to “type ABC” for Atlanta Business Chronicle keeps instructions concise in classes and consultations, decreasing the possibility of misunderstanding an unfamiliar database name. GBL is planning additional usability testing in spring 2016 to follow up on the web redesign and iterative changes made since the initial 2014 launch. This will enable a new round of enhancements to increase opportunities for a successful and efficient customer experience using GBL’s website.