Research in Industry: From Generating Insights to Driving Decisions
Reflection on Lecture Talk by Dr. Prakriti Parijat @ Instacart
🛒
Dr. Prakriti Parijat | Head of Research (User + Market) | Instacart
In academia, research is often treated as the goal itself. In industry, as Dr. Parijat emphasizes, research serves a different role. It becomes one input among many in decision making. That shift changes how research is designed, conducted, and communicated.
One of the key distinctions is the role of rigor. In academic settings, rigor is often the goal. In industry, rigor is a means. It must compete with time, resources, and organizational priorities. The objective is not to design a perfect study, but to generate enough evidence to support a confident decision. This makes research feel far more pragmatic and iterative.
Another important takeaway is the range of methods used in industry. Strong research often combines qualitative, quantitative, and behavioral approaches, depending on the question at hand. At the same time, no method is perfect. Self reported data, for instance, often reflects what people say rather than what they actually do. Recognizing those limitations and selecting methods accordingly is a critical skill.
Dr. Parijat also highlights the broader role of research across the product lifecycle. Research is not only used to evaluate solutions. It can shape strategy, identify emerging problems, measure impact, and scale customer understanding across teams. In that sense, research becomes a way to build shared intuition inside an organization.
One of the most powerful insights is that insights alone do not drive change. Influence does. Even rigorous findings will not shape products unless they are communicated effectively and connected to decision making processes. This is where storytelling matters. It means making data visible, linking findings to real user experiences, and sometimes reframing the problem itself.
The idea of iteratively reframing the problems is especially compelling. Some of the most valuable moments in research happen when we realize that the original question was not quite right. That requires researchers to move beyond execution and engage critically with the broader context of the work.
Finally, Dr. Parijat shared her experiences about how the rapid growth of AI application in the industry have impaced and reshaped work patterns. AI systems often begin without a clearly defined problem, which increases ambiguity and shortens iteration cycles. In this environment, researchers need to act as signal detectors, distinguishing genuine user value from novelty and adapting continuously as tools and methods evolve.
Takeaways
- 🎯 Research in industry supports decisions rather than being the decision itself.
- ⚖️ Rigor is balanced with time, resources, and business needs.
- 🔬 Mixed methods are essential but should be applied pragmatically.
- 👀 Self reported data has limitations and should be complemented with behavioral data.
- 🗣️ Influence and storytelling are critical for impact.
- 🔄 Reframing the problem can be more valuable than refining the solution.
- 🚀 Adaptability is a key skill in rapidly evolving environments, especially with AI.
Personal Reflection
This talk resonates with my own transition from academic research toward more applied problem solving. In my current work, I often focus on building models or designing systems, but this perspective reminds me that the ultimate goal is not the model itself. The goal is the decision that the model supports.
In fatigue prediction, for example, high model performance is important, but it is not enough. The more meaningful question is how the model changes action. Does it help supervisors intervene earlier? Does it change how workers manage fatigue? Those outcomes matter more than technical performance alone.
I also find the idea of minimum viable evidence especially valuable. Rather than waiting for perfect validation, it may be more impactful to deliver timely and actionable insights that can guide decisions and then improve through iteration.
Finally, the emphasis on storytelling and influence points to an area I want to strengthen further. The ability to connect data to human experience and communicate that clearly seems just as important as the technical work itself.