1 of 26

Meg McMahon

SAA User Experience Section | April 2026

Exploratory UX Research

at Harvard Library

2 of 26

Framing

2

I want to take a little time to frame a bit about my team and introduce exploratory research. I’m a part of a small team at Harvard Library called UX & Discovery.

On my team we have to sides, one is digital collections focused and one side, is UX focused.

On the UX side, we have me and 2 other full time UX dedicated roles. I also manage two part time UX Research Assistants along 3 Library UX interns in the summer for 10 weeks. I am extremely fortunate to be in such a large Library UX team and have this be my full time job. I lead around 30 UX studies a year within my role both evaluative and exploratory research relating to library goals and questions my colleagues across the library bring to our team.

I recognize that that isn’t feasible for many in this room, but that shouldn’t dissuade you from doing UX work in your archives. Today I hope to share actionable ways to think about research within your context.

3 of 26

Background

Role: Harvard Library UX & Discovery team
Strategic goal: Simplify & enhance online special collections discovery
The challenge: researchers navigate 7+ systems: HOLLIS, HOLLIS Images, HOLLIS for Archival Discovery, Harvard Geospatial Library, CURIOsity, and more

4 of 26

Answer: Exploratory Research

Research designed to surface real problems before we try to solve them
It's research designed to help you learn what you don't know you don't know

5 of 26

Types of UX Research

Exploratory Research

Used when we don't yet know the right questions to ask

Surfaces real problems before solutions

Other Research Types

Descriptive, how much / how often?

Evaluative, does this solution work?

6 of 26

Research Context & Approach

6

7 of 26

Core Questions

How can we make our systems more transparent?

At what point do researchers feel "done" with discovery?

When do users choose online vs. in-person access?

8 of 26

A Multi-Method Approach

Survey

Broad patterns across many users
Who are special collections researchers?
What do they do?
Feeds recruitment for interviews

Contextual Inquiry

Watching researchers search in real time
Asking about their experience
Captures the "why" behind behaviors
Surfaces emotions & workarounds numbers can't

After deciding on our core research questions we decided on a multi-method approach We decided to conduct a survey first because we wanted to understand more about the breakdown in audience coming to us with special collection research needs? How do those needs differ between audience gorup? A survey can collect this at scale.

We also selected this multi-method approach because we then could recruit from a survey to a contextual inquiry. My team often uses this two pronged approach because it gives us access to participants and it gives us both qualitative and quantitative data that makes our future analysis more rich.

For the second part of the approach we decided on contextual inquiry because it focuses on watching researchers do a task as naturally as possible. We asked participants to show us how they searched for their most recent research interest in our systems. Pausing when it made sense to ask clarifying questions on their approach. This usually goes on for around 20-30 minutes of watching them complete a single task, that often has no easy finish state, in our case asking someone to research, shows us the start and middle of their process. When then could ask about final steps at the end of the interview.

9 of 26

Survey to Contextual Inquiry Tips

Use the survey to screen and segment, not to gather all your data
Close with an open invitation, not a hard ask
Design survey questions to generate good contextual inquiry prompts
Leave deliberate gaps
Send a short "bridge" communication between survey and session

1. Use the survey to screen and segment, not to gather all your data

Resist the temptation to ask everything upfront. The survey's job is to identify who to invite into the contextual inquiry sessions and to surface meaningful variation in your user population, experience level, collection types used, institutional context, frequency of archival visits. You're building a purposive sample, not running descriptive statistics.

2. Close with an open invitation, not a hard ask

End the survey with a low-friction opt-in for follow-up: "Would you be willing to be observed working with archival materials for about an hour?" Keep it vague enough that people don't self-select out because they imagine something more formal than it is. Save the details for a follow-up email once you've identified strong candidates.

3. Design survey questions to generate good contextual inqury prompts

As you write survey items, ask yourself: would an unexpected answer to this question make me curious? If someone says they "rarely" use your finding aid search system but they use your archives, that's a thread to pull in the contextual inquiry session. Build your interview guide by mining survey responses for anomalies and tensions, not just patterns.

4. Leave deliberate gaps

Don't ask about things you can observe. If you're going to watch someone search a catalog, don't ask them to describe how they search a catalog. Survey questions about behavior are often inaccurate anyway, people report idealized versions of themselves. Save behavioral questions for direct observation and use the survey for attitudes, background, and logistics.

5. Send a short "bridge" communication between survey and session

After selecting participants, send a brief message that references their survey responses: "You mentioned you often work with unprocessed collections. We'd love to observe a session where you're doing that kind of work." This does two things: it signals you actually read their answers and it begins building the rapport that makes contextual inquiry sessions more candid and productive.

10 of 26

Contextual Interview Flow

Introduction & framing
Consent & logistics
Background warm-up questions
Task introduction
Observation with concurrent inquiry
Retrospective probing
Participant reflection
Debrief and close

1. Introduction and framing

Open by explaining the purpose of the session in plain terms, like "we're here to learn from you, not to test you." Emphasize that there are no right or wrong actions, and that silence or struggle on their part is useful data for you, not a problem to fix.

2. Consent and logistics

Cover recording permissions, how data will be used, confidentiality, and their right to stop at any time.

3. Background warm-up questions

Ask a handful of brief, low-stakes questions to establish context and get the participant talking comfortably. These should cover who they are in relation to the work, their research focus, how often they visit archives, what they're working on today. Keep it to five minutes or less.

4. Task introduction

Transition into the observation portion by asking them to do something they would naturally do, not a contrived scenario. Ideally they bring their own real task. You might say: "Go ahead and do archival research as you normally would. I'll watch and ask questions as we go." If they don't have a natural task ready, ask them what their last research topic was, and use that as inspiration.

5. Observation with concurrent inquiry

This is the core of the session. Watch what they do and ask questions in the moment, as things happen, not afterward. Use a light touch: "I noticed you paused there, what were you thinking?" or "You switched search terms just now. can you tell me why?" The master-apprentice model is your guide here: you're trying to understand their tacit knowledge, the stuff they do without thinking about it.

6. Retrospective probing

If something happened during the observation that you didn't want to interrupt return to it here. You might say: "Earlier when you did X, I'd love to go back to that moment." This is also where you can ask slightly more interpretive questions: "Is this how this usually goes, or was today atypical?"

7. Participant reflection

Invite them to step back and reflect broadly: "Is there anything about how you work with archives that you don't think we saw today?" People often surface their most candid observations when given explicit permission to editorialize at the end.

8. Debrief and close

Thank them, explain next steps, and give them a chance to ask questions of you. If you're doing multiple sessions, avoid sharing findings from other participants at this stage, you want each session to stay uncontaminated. Let them know how (and whether) they'll hear about what comes from the research.

11 of 26

Contextual Interview Tips

Follow the researcher, not your process
Ask "why" more than "what"
Treat the reference interview as a warm-up, not the session
Pay attention to workarounds
Remember that context includes the physical and emotional environment

If you are considering this approach in the future here are a few tips.

1. Follow the researcher, not the finding aid

Observe users navigating your collections in their natural workflow. Resist the urge to guide them toward the "correct" path. The detours reveal where your organizational logic breaks down.

2. Ask "why" more than "what"

Archivists are trained to document what exists. In contextual inquiry, the goal is to understand why a user makes each move, why they abandon a search term, why they request a box and then set it aside, why they photograph one page and skip the next. That interpretive layer is where UX insight lives.

3. Treat the reference interview as a warm-up, not the session

A reference interview captures stated needs; contextual inquiry captures actual behavior. Use the initial conversation to establish rapport and consent, then let the work itself do the talking. Watch for gaps between what users say they're doing and what they actually do.

4. Pay attention to workarounds

If a user pulls out their own notebook, switches to a personal spreadsheet, or photographs a box list rather than using your catalog, that's signal. Workarounds are informal systems users build when the formal one doesn't meet their needs and they're gold for identifying gaps in your finding aids, catalog records, or reading room workflows.

5. Remember that context includes the physical and emotional environment

Archives have unique contextual layers gloves, time limits, no-camera policies, distant repositories. Note how these constraints shape user behavior. A researcher rushing because the reading room closes at 4pm is in a fundamentally different cognitive state than one with an open afternoon. Capturing that environmental pressure is part of capturing the full context.

12 of 26

What We Found

12

13 of 26

“Not only does Google get me into some sites in a way that is more effective than those sites' own search tools…”

Undergraduate Student

14 of 26

Fragmented discovery, no single search system is trusted

Researchers routinely bypass library systems entirely at the start.
Example they search Google + "HOLLIS" or "Harvard" to land on Harvard resources, not our own search tools.

What this means: our systems need to be legible and findable from the open web, not just navigable once you're already inside them.

15 of 26

“The biggest barrier to getting into the archives is choosing the material you want to look at.”

Graduate Student

16 of 26

Discovery Is Driven by Keyword Chasing, With No Finish Line

Researchers follow names, provenance trails, and networks from one source into the next.
Novice researchers are especially overwhelmed, topics are broad, materials scattered, and there's no signal they've found everything relevant.

Filtering feels efficient but risks excluding something valuable by constraining too early VS No filtering returns hundreds of results with no way to make sense of the volume

17 of 26

“Most of my search terms gives me 100s of objects. And I always lose my place.”

Faculty Researcher

18 of 26

Researchers Want to Track Progress and They're Improvising

Users want to mark what they've seen, save their place, and build working collections from materials they find.
Current workarounds: printing PDFs of finding aids to annotate by hand, using ⌘F on downloaded files instead of online search tools.

19 of 26

“I’ve been here 4 years and didn't realize we had Arabic manuscripts.”

Graduate Student

20 of 26

Collections Are Effectively Invisible, Users Can't Browse What They Don't Know Exists

Participants repeatedly described being unaware of entire collections at Harvard.
Awareness relied almost entirely on word-of-mouth and librarian relationships.

21 of 26

“Working with a collection specialist is having a translator that understands the collection, and is willing to look at it through my eyes.”

Faculty Researcher

22 of 26

Human Expertise Fills the Gaps Our Systems Leave

Librarians and collection specialists function as essential navigators, not a nice-to-have or last resort, but a core part of how research actually gets done.

UX improvements should amplify staff expertise, not assume technology can replace it.

23 of 26

Discussion & Q&A

23

24 of 26

Questions

What surprised you most? Does any of it match experiences with your own users?
Harvard's users relied heavily on staff expertise to fill gaps in discovery systems. Is that a feature or a workaround at your institution?
This research used surveys and interviews together. What feels like an accessible starting point in your own work?
How do you see UX research as a tool for living SAA's values and where might there be tensions?
If you could ask your users one question right now about their discovery experience, what would it be?

25 of 26