Pվ��Ƶ

Author: Saskia Hoving

Editor-in-Chief

Academic libraries have long relied on usage data to demonstrate value. But as AI changes how researchers discover and interact with content, many of the traditional indicators used to justify investment are becoming less reliable. Downloads, views and cost-per-use still matter, but do they capture the full contribution libraries make to research success? This question is becoming increasingly important for collection teams, library leaders and institutional stakeholders alike.

A famous 1970s study at the University of Pittsburgh, , revealed that nearly 40% of the library’s newly-acquired books were not being used at all. As Administration called for budget cuts, the Library Committee discovered the real truth behind the figures. “Usage” was defined in the study as an item being checked out – whereas in reality, students and researchers were routinely consulting the books in the building. 

Following our initial discussion on the new priorities for academic libraries, this post draws on the experience and perspectives of three librarians involved in usage metrics on a daily basis: Amy Lewontin, Collection Development Librarian at Northeastern University; Lisa Redlinski, UK-based Information Strategy Consultant; and Niklas Willén, License Manager from the Bibsam Consortium at the National Library of Sweden.

Why traditional library metrics no longer tell the full story

Traditionally, libraries have relied on measures such as COUNTER statistics, downloads, views and cost-per-use to show that resources are being used. These measures are still important today, but the questions universities are asking are becoming broader. As universities are increasingly focused on research excellence, research offices need to demonstrate impact and institutional leaders want to understand how their investments contribute to wider strategic outcomes. Libraries are therefore under growing pressure to demonstrate not only how much content is being used, but how their resources and expertise contribute to these broader institutional goals.

This is changing how usage data needs to be interpreted. Researchers are increasingly discovering and accessing content through AI tools, bypassing library and publisher platforms. This “zero-click” behaviour can make usage less visible. As the Pittsburgh study suggests, apparent declines in usage do not necessarily mean that demand for research content is falling. Activity may just be taking place outside the systems that libraries have traditionally measured.

There is also a more fundamental question about what usage statistics actually tell us. Niklas Willén challenges the assumption that a download or unique item request necessarily represents meaningful engagement with the content. “That unique item request doesn't tell if you actually read the article or just skimmed through the abstract or came to an article and browsed onwards. The actual need for reading the full text is way less than the statistics show you, I believe.”

This raises a bigger question for libraries: if traditional measures capture only part of the picture, what evidence should they use to demonstrate their contribution to research and institutional priorities? The issue is no longer just how much content is being accessed, but what researchers actually need, how library resources support research excellence, and ultimately what difference that support makes.

How AI is changing research discovery and library metrics

Niklas also points out that AI tools surface open access material more readily than paywalled content, which may become less visible as a result and cause a fall in demand. So, access conditions themselves may increasingly influence the visibility and value of content. If researchers increasingly rely on AI-generated summaries, research that is not readily accessible to AI tools risks becoming less visible in discovery workflows and, as a result, having less influence.

Libraries have traditionally seen their role as providing access to information. But Niklas questions what is actually needed. “I think librarians in general don't have an up-to-date view on how usage is driven anymore. If the use of subscribed material is falling, you need to base your renewal decisions on data that gives you more detail,” he says.

The future may require a more nuanced distinction between access that is desirable, access that is necessary and access that genuinely creates value. The metrics are therefore likely to broaden beyond just activity, to encompass a much wider evidence framework, including:
 

• Usage: What is being accessed?
• Need: What content genuinely requires full-text access?
• Visibility: What information are researchers finding through AI and other discovery tools?
• Alternatives: Could open access, perpetual access, scholarly sharing or document delivery meet the need?
• Research outcomes: What does library support enable researchers to achieve?
• Research impact: Does library support help research reach policy, practice or wider audiences?
• Research quality: Does it improve source evaluation, reproducibility, citation accuracy and research integrity?
• User capability: Are researchers becoming better equipped to navigate and evaluate information?

Amy Lewontin’s experience at Northeastern bears this out, demonstrating that the library's value is increasingly extending beyond just providing content. “We get more and more requests from researchers for help with research impact, and with demonstrating influence and visualising research outputs,” she says. “And we respond with research analytics tools, so we have Overton, for example.”

Measuring value requires new data and technology partnerships

As new metrics frameworks evolve, collection assessment can no longer sit just within the library. Lisa highlights the growing importance of technical services in the modern library world of electronic collections, metadata, discovery platforms and repositories. As research, and information consumption, becomes increasingly digital and AI-mediated, these technical functions become ever more integral to measuring library value.

Amy agrees - her library now works much more closely with IT, data and analytics teams. “People ask us to connect up AI tools to the licensed resources, so then we need assistance from the IT team.” It’s no longer just about the content itself but AI functionality, reliability, integration and technical performance as well. And as research becomes increasingly data-intensive, libraries need new ways to fully understand the value of datasets. Now that the library has become both a consumer and an interpreter of institutional data, it needs to draw on the expertise of technical and analytics specialists to understand complex patterns of information consumption.

Looking ahead, any future metrics framework will need reliable ways of distinguishing human users from automated activity. Technical measures such as bot identification, AI traffic classification, metadata, attribution and usage reporting cannot solve this problem alone – collaborative efforts are needed. This makes the relationship between library, IT and data teams more than just operational. And in Amy’s experience, libraries are already becoming more integrated into institutional strategy.

The growing partnership between libraries and research offices

Five years ago, research offices were primarily focused on grants, compliance, assessment and publication reporting. Today, their remit is increasingly broadening to include evidence of policy impact, public engagement, collaboration, societal impact, responsible research and open science. This expanding role is creating greater overlap with areas of established library expertise, such as bibliometrics, research analytics, open access, scholarly communications and research visibility.

In practice, a researcher seeking to demonstrate how their work has influenced policy discussions might need evidence of this for institutional reporting. The research office identifies the reporting requirement, while the library supports the researcher in gathering the evidence, using tools such as Overton, citation databases and other research analytics services to trace policy mentions, citations and wider research influence.

“I feel like the library is pretty involved,” says Amy Lewontin. “I’m dealing with colleagues in the research office more than I used to 4 or 5 years ago as we’re more embedded in the research process itself. We’re also working more together on things like open access agreements.”

Collaboration at the heart of research success

Ultimately, this more collaborative relationship reflects a wider shift in how institutions view research value. “How much did researchers use what we bought?” is giving way to “what information did researchers actually need, how did they find and use it, what did the library contribute and what difference did that make?”

Whilst Niklas Willén, Lisa Redlinski and Amy Lewontin approach these questions from different perspectives, they all agree that traditional usage statistics still have an important role to play. However, downloads, views and cost-per-use are increasingly just one part of a much broader picture. As AI and new discovery pathways reshape how researchers access and interact with information, libraries are being challenged to look beyond activity alone and build a fuller understanding of value.

That shift is also changing who libraries work with. Demonstrating value increasingly requires collaboration with research offices, IT teams, data specialists and institutional leadership. These groups are helping institutions better understand not only what research is being accessed, but how it is discovered, supported, shared and translated into outcomes.

In this evolving landscape, libraries are well positioned to connect evidence, expertise and institutional priorities. Their expertise in collections, research analytics, scholarly communication and researcher support positions libraries as important partners in helping institutions understand and demonstrate research value.

Related content

• The growing role of librarians in research data and cybersecurity
• Cybersecurity, data, value and partnerships: The new priorities for academic libraries
• Emerging librarian roles in supporting ethical and rigorous scholarship
• Exploring AI futures in libraries with the Futurescape Workshop community
• Evolving roles of academic librarians adapting to change and new technologies

Get the latest news & blogs, subscribe to The Link Alerts!