The bottleneck in your data organization isn't the data. It's the conversation happening before anyone touches the data and not understanding the question before answering it. I’ve learned that librarians have addressed this problem with the reference interview, a structured way to uncover the true information needs, not just what they asked for. Jenna Jordan and Amalia Child joined Tim Gasper and I on Catalog & Cocktails Honest No-BS Data Podcast to break down why this framework belongs in every data team's toolkit. This expands on their talk from Data Day Texas Before you answer, establish trust, listen without interrupting, paraphrase back, ask open questions then narrow ones, and verify at the end that the real need was met. Because more often than you'd think, it wasn't. These aren't soft skills. It’s the STRONG skills, which are much needed in this new AI world. Invest upfront, and the next time a similar question lands, you're already ready. Two additional interesting topics that came up: - It's not service vs. strategy. The reference interview is the strategy. Front-load the conversation, and you scale. - Domain expertise is non-negotiable. Librarians often carry two degrees, library science plus a subject area. Data people need the same. Less data literacy. More business literacy. IMO, this is also a differentiator in this new AI world. Also: track your unmet needs. What did people ask that you couldn't answer? That dataset tells you more about your org than most dashboards will. Check out the episode! Link in comments Have you been applying a reference interview in your data work?
You might like this paper from Corey Harper mapping librarian skillsets to the needs of AI. https://www.tandfonline.com/doi/full/10.1080/01639374.2025.2539787#abstract
The librarian's are using something similar to Open, Probe, Confirm questioning method. Very useful for discovery workshops.
Thanks for sharing this Juan Sequeda. We can learn from methodologies like Imago Dialogue, as it all comes down to people and praciting mirroring, validation, and empathy.
The business literacy point hits. In my Data Office, the questions that move the needle rarely arrive as "I need this dataset". They arrive as "something feels wrong in the churn figures". Translating that into a data question is the craft. We've started treating intake calls like diagnostic conversations... surface the decision behind the request, then figure out what data serves it. Cuts rework in half. Reference interview is a good name for what we've been doing instinctively.
When data teams operate in ticket-queue mode, they get very good at answering the wrong question at high precision. The shift happens when you build structures that force the upstream conversation - just as a process. That's actually one of the core jobs of data leadership: designing the environment where this conversation happens by default. Not coaching every analyst to ask better questions, but building decision forums where the framing happens structurally - before anyone touches the data.
This is spot on—an essential skill for better data modeling. It really comes down to staying curious, like we did as kids: learning, asking questions to validate understanding, and continuously connecting the dots to build deeper context.
The best data teams I’ve seen do two things well: they understand the question, and they understand the business behind the question. Strong point on domain expertise. Without business literacy & domain knowledge, even the best data model ends up answering the wrong question.
The conversation is the bottleneck, and the question that exposes it fastest is the one most leaders skip: how have you been burned by data in the past. Pain points and goals get a useful brief; the burn question gets the trust deficit nothing else surfaces. Stakeholders have an answer ready; the data team usually never asked.
Check out the LinkedIn live event with all the great comments https://www.linkedin.com/posts/juansequeda_data-teams-think-like-a-librarian-with-activity-7449219441655181312-1U_y?utm_source=share&utm_medium=member_ios&rcm=ACoAAAE_pWMBR7k60w6xpXEoAiARb9omer0qUaQ