Is It Getting Used Enough?

Today’s design dilemma revolves around the question of whether an existing feature in an interface is getting used enough. Our Primo VE discovery layer (locally branded as OneSearch) includes an add-on service from Ex Libris called the bX Recommender. As you view records of journal articles in your search results, the service will serve up a short list of relevant articles on the right side of each record, as can be seen in the “Related Reading” section of this journal article record.

Ex Libris provides this quick overview of where the recommendations come from:

bX harvests link resolver usage data from many academic institutions around the world. If two articles are used in the same session, the system analyzes the connection between them and stores the items in a co-retrieval network. Because bX recommendations are based on link resolver usage, they are truly platform- and content-neutral. The usage is generated through discovery systems, A&I databases, publisher platforms, and any other source that links users to full text via a link resolver. The articles may be from different journals, publishers, and platforms.

While this content is seemingly useful, having it there comes at a cost to the user and to our library system. For the user, it adds a considerable amount of text to what is unarguably a very busy, text-heavy page design. For our library system, it is an add-on service to our Primo VE subscription that we pay extra for. A question has come up if it is worth having or not. We have usage data that suggests it is indeed getting used (roughly 2000 clicks a year) but it is not clear how much use is enough to merit the cost to the university for the subscription and the cost to the user for the added cognitive load. I don’t know how much we pay for this service, but if it is substantial, that’s a serious consideration.

My UX instincts tend to push me to be wary of the lure of features lest they then turn into featuritis situations, but without enough data about user behaviors and preferences in this matter, it’s hard to know how valuable this feature actually is and to what extent there is a notable number of users who like it. I wonder if the usage numbers are the work of a handful of heavy users of the feature or more evenly distributed; maybe it’s another example of the Pareto principle at work where only 20% of the users of the feature are responsible for 80% of the usage.

In theory, deciding what to do in this case is researchable. I can think of a number of different ways to get better data on usage of the service, perceptions of it, and the user experience of it. But, there are also dozens of other openly debated questions we have at our library and across our library system that need to be resolved, but only so much time and staff availability to make evidence-driven decisions. I guess we’ll have to figure out how much research we need to do on this matter to make a decision about whether to keep it or not.