For years, dashboards have been the de facto deliverables for many data analysis projects. They certainly have their strengths: polished, long-lived artifacts are useful when tracking established metrics or exploring trends over time.
But we also know that many dashboards start to rot as soon as they’re shipped. Unless data teams dedicate ongoing resources to maintenance and continued development, dashboards can break or lag behind evolving company needs. Stakeholders stop returning to dashboards when they have to leave their day-to-day tools and workflows, track down the right link, and relearn how to interpret the dashboard content. Perhaps most importantly, stakeholders are realizing that answers to many questions that pop up in their day-to-day work often aren’t found in dashboards: they’re increasingly found in quick chats with teammates, or through self-serve discovery enabled by AI assistants.
As a result, expectations about where and how stakeholders interact with data are changing. Here, we describe shifting user preferences and habits, and share how data teams can meet stakeholders where they are in today’s emerging AI-powered data landscape.
Users want to explore data independently in the flow of work
Until recently, it was standard practice to publish a dashboard and assume decision makers would return to it when they needed answers. But stakeholders today are unsatisfied with consuming prebuilt insights. They want to investigate ad hoc questions as they arise, without having to wait in their data team’s ticket queue.
As Observable co-CEO Julio Avalos shared in our recent webinar, this is an evolution years in the making:
We've spent decades convincing business managers, executives, and even individual contributors throughout an organization that data is meaningful, and they therefore want to participate in finding insights. They don’t feel entirely comfortable offloading data work onto specialists within their organization — they want faster answers, and more self-serve access.
Julio Avalos
With AI agents quickly democratizing data analysis, it’s not just possible for more people with diverse skillsets to jump into data work — it’s becoming the norm. AI agents are powerful for common tasks like summarizing trends, surfacing anomalies, translating natural language into queries, and drafting data visualizations. Boosted by AI, product managers, researchers, and even casual hobbyists now have the ability to independently investigate and answer questions.
Just as important as how stakeholders want to interact with data is where that interaction happens. People want to easily access and work with data without jumping out to a separate BI platform, which slows momentum. Modern tools help users stay in flow by exploring metrics inline, adjusting assumptions in context, and sharing findings instantly without saving and emailing screenshots.
Here’s Julio again:
People don't want to go and learn yet another tool. I think that's another changing expectation for stakeholders: they want to be met where they are. They want to be met in Slack. They want to be met in Teams. They want to be met wherever they're working, and they want to consume your information and participate in data work there.
Julio Avalos
To summarize, stakeholders are moving away from dashboards as the go-to place to find insights in their data. Instead, they’re turning to AI agents that let them immediately dig into new questions as they pop up, without filing a ticket or breaking out of their existing workflows.
How data teams can navigate shifting stakeholder expectations and practices
Adapting to these evolving expectations doesn’t require sacrificing rigor. As AI-powered analytics becomes more widespread, data teams play a critical role in enabling this new way of working while safeguarding trusted, high-quality practices. Here are three practical ways data teams can meet stakeholders where they are, without compromising analytical standards:
Model and teach reproducibility. Reproducibility becomes even more critical as insights are generated in AI chats and shared across ephemeral channels, because it ensures accurate, trusted answers that hold up to scrutiny. When a metric appears in a thread, it should link back to a source of truth such as a version-controlled query, a documented definition, or a computational notebook with necessary code to recreate the results. By encouraging good practices and tooling for reproducible analysis, analysts can help users across teams produce verifiable results for more confident decision-making.
Ship portable, digestible, and focused insights. Rather than adding to sprawling dashboards, analysts should move toward delivering digestible insights and data visualizations that answer specific questions. A focused chart embedded in a ticket, thread, or shared document invites more engagement than one locked inside an unfamiliar BI workspace.
Define clear standards around exploratory versus production-grade outputs. In fast-moving conversations, it is tempting to immediately share whatever number an AI agent returns. Sometimes, that might be low-risk. Or, it can have serious business consequences if AI hallucinates or misinterprets your request. Analysts can help their teammates distinguish between quick exploratory analyses, and production-grade outputs that require more rigorous testing and validation. This empowers users to practice fast, AI-powered analytics when appropriate, and know when they need to ask for additional review.
By modeling and teaching reproducible workflows, shipping digestible insights, and helping to differentiate between low-risk exploration and consequential analyses, data teams can meet stakeholder expectations for faster AI-powered discovery while promoting sound practices across their organization.
Conclusion
Dashboards aren’t disappearing, but they’re no longer the center of gravity for data work. Stakeholders expect to find insights in real time within their everyday tools, enabled by AI agents that democratize data exploration and analysis.
In this evolving landscape, data teams can meet stakeholders where they are by delivering digestible, inspectable insights directly into their existing workflows, and advocating for practices that uphold analytical standards.
Want to hear more about how AI is impacting the data analysis landscape, and how data teams can keep up? Watch the recording of our recent webinar, or read the recap blog post for highlights.