Common Pitfalls in Problem Identification and Analysis Thinking for Data Analysts

As analysts, we must be accountable for data and conclusions, adopting a mindset of "bold speculation, careful verification." The article shares common pitfalls in problem identification and analysis thinking, illustrated through real‑world scenarios.

Scenario 1: A business colleague reports an abnormal bounce‑rate on an activity page. The correct first step is to clarify the metric definition (answer C), distinguishing between app bounce‑rate and page bounce‑rate to avoid misinterpretation.

Scenario 2: The same colleague sees a declining "visit‑to‑pay conversion rate" despite a successful campaign. The situation exemplifies Simpson's paradox: overall metrics drop while each user segment improves, often due to changing segment proportions such as more silent or new users.

Scenario 3: A colleague claims that selling more flowers increases sales of razors and massage chairs. This is a correlation‑causation fallacy; the observed correlation likely stems from seasonal events rather than a causal relationship.

Analysis Thinking: To avoid chaotic analysis, the article recommends three common frameworks:

1. Business‑process driven: Map the result to its contributing steps, identify relevant factors at each step, and use data to validate hypotheses (e.g., diagnosing stock‑out issues by tracing supply‑chain nodes).

2. Department‑function oriented: Align analysis with responsible departments, allowing clear accountability and actionable recommendations (e.g., evaluating activity performance by department).

3. Strategy‑goal oriented: Clarify the business objectives and strategies before analysis, ensuring the investigation aligns with the intended outcomes (e.g., reviewing promotional activities against defined goals).

In summary, the article highlights typical traps when defining problems and choosing analysis paths, and provides structured methods to guide analysts toward disciplined, insightful investigations.