Conceptualization as the Foundation of Hypotheses in Product Development
By 2025, data-driven approaches are the norm in product development: teams form hypotheses, test them via A/B experiments, and iterate based on metrics. Yet in practice, the HADI cycle (Hypothesis, Action, Data, Insights) often stalls. After validating a hypothesis—for instance, changing a button's color to boost conversions—it's unclear what to test next. Without a step for generating fresh ideas, teams stagnate.
The solution? Shift from statistical hypotheses to scientific ones that explain causality. Instead of "enlarging the button will increase conversions," frame it as: "enlarging the button will increase conversions because users have poor eyesight and will notice it more easily." This creates chains of hypotheses, where one outcome informs the next.
The Problem of Multiple Hypotheses in Experiments
In real-world tests, isolating a single hypothesis is impossible. Per the Duhem-Quine thesis, parallel assumptions are always being tested:
- The idea is sound, but the implementation has a bug (e.g., the progress indicator resets on page refresh).
- The idea is sound, but the solution falls short (e.g., the indicator design is cluttered with text instead of a progress bar).
- The idea is flawed from the start (e.g., the indicator doesn't affect users' confidence in the checkout process).
This muddies data interpretation and demands a systemic view of the product.
What is Conceptualization in Product Work
Conceptualization means mapping core concepts and their relationships to systematically describe the product. Concepts like "page," "input fields," and "order creation" link via confirmed facts or untested hypotheses. A target concept, such as "repeat orders," integrates into the network: users choose platforms based on ease of use and low fees.
The map reveals gaps. Facts capture reliable links; hypotheses flag those needing validation. Choosing a base metaphor (the research lens) defines the concepts: it shapes hypotheses and conclusions even before experiments.
Philosophically, this is Popper's "illogical core of a theory": the metaphor sets the viewpoint, constraining and sharpening the analysis.
Practical Case: Evolution of the Conference Model
Consider a conference as a product. Base metaphor: "knowledge marketplace," where speakers (producers) trade insights for recognition, and attendees (consumers) gain professional growth.
Research uncovers mismatches:
- Attendees skip talks but hit the parties—add the "entertainment" concept.
- Some ignore both talks and parties (sent by their employer)—introduce "employer," shifting the metaphor to "team-building camp for employee motivation."
- Parties lose money—the brand used them for outreach, so revert to "listener."
The model evolves: it's not "right," but suited to the current data. Conceptualization acts like blinders: it sharpens focus while tuning out noise.
Segmentation through Conceptualization
Refining the map ditches the generic "user" for targeted segments:
- Demographic: "female user aged 25–35, developer."
- Behavioral: "user who abandons the cart at checkout."
- Latent: "user with low tolerance for complex interfaces."
This sharpens hypothesis precision and experiment relevance.
What Matters
- Conceptualization chains isolated hypotheses together, solving post-test idea generation.
- It addresses Duhem-Quine multiplicity, systematizing experiment analysis.
- The base metaphor locks in the research logic long before any data arrives.
- The model evolves with new insights, serving as a focusing tool rather than universal truth.
- Segmenting concepts makes product decisions more relevant for specific groups.
— Editorial Team
No comments yet.