Mastering Data Analysis: Methodology, Team Building, and Career Insights

What Is Data Analysis

In one sentence, data analysis is the process of discovering business value from data through analytical techniques. It starts with obtaining data and ends with realizing business value, passing through stages such as data acquisition, cleaning, processing, modeling, result presentation, value discovery, and value realization.

Before detailing each stage, I first explain my understanding of the two core concepts: data and business value.

Data : Data is more than raw numbers. It must include the source, measurement method, unit, and especially the business context. For example, "170, 172, 180" are just numbers, but "average heights of three regions" becomes meaningful data, with implied units (centimeters) and source credibility.

Business value : Analysis that does not serve business is lifeless. Value is realized only when analysts present insights that business stakeholders understand and can act upon, turning analytical conclusions into concrete business actions.

Detailed process description:

Data acquisition : Requires a clear analysis goal to limit the massive data pool. The output is a data subset, either physically copied or logically defined (e.g., time range, dimensions, metrics). Challenges include understanding incomplete or changing data source documentation.

Data cleaning : Handles anomalies, missing values, consistency transformations, and encoding replacements.

Data processing : Aggregates or transforms data to make it suitable for modeling.

Data modeling : Uses statistical analysis or machine learning to describe data or predict the future. Most analysts focus on trend comparison, dimension breakdowns, and impact analysis.

Analysis result presentation : Goes beyond charts; the goal is to ensure business users fully understand the findings. Communication—both one‑way and two‑way—is essential.

Business value discovery : Analysts propose value, but it must be recognized by the business side to become actionable.

Business value realization : Although the business ultimately controls implementation, analysts must stay deeply involved to bridge gaps between analysis and execution.

Key reminders before starting any analysis:

The process is not strictly linear; you can loop back to earlier stages when new issues arise.

The process is iterative; the end of one analysis often becomes the start of the next.

Not every stage is required for every analysis.

Close collaboration between analysts and business stakeholders is essential throughout.

How to Set Analysis Goals

Goals typically come from two sources: the business side and the data team itself. In many projects both influence the goal, with varying weight.

Business‑driven goals : Business units ask questions to diagnose past performance or forecast future trends. These goals are often vague and expressed in business terms, requiring analysts to perform extensive requirement analysis and translate business language into data metrics.

Data‑team‑driven goals : An independent data team can propose goals based on a mature metric monitoring system. However, without clear business relevance, such reports may be ignored.

A well‑defined goal should:

Reflect a business perspective and pinpoint pain points.

Be supported by data.

Be quantifiable (e.g., "why did production drop from 10,000 to 5,000?").

Map to specific measurable indicators.

What Makes a Qualified Data Analyst

Qualifications span work objectives, responsibilities, and skill requirements, which vary with company stage and data maturity.

Work Objectives Vary by Company Phase

Start‑up: Provide industry and competitor insights to guide strategy.

Rapid growth: Monitor performance, build metric systems, uncover hidden issues.

Stable phase: Focus on efficiency, cost, and fine‑grained operations.

Plateau or bottleneck: Analyze market supply‑demand shifts and competitor responses to find new growth points.

Core Responsibilities (Data Acquisition → Value Realization)

Analysts spend more time on acquisition, cleaning, and metric building in early data‑maturity stages, and shift toward modeling, presentation, and value realization later.

Skill Requirements

Two categories: general abilities and technical abilities, each split into business and data dimensions.

Business abilities :

Micro: Understand day‑to‑day operations and company challenges.

Macro: Grasp industry trends, company positioning, and strategic direction.

Data abilities :

Know all foundational data sources and relationships.

Build and relate operational metrics.

Translate analysis results into actionable business recommendations.

Communicate, influence, collaborate, and iterate the analysis‑to‑execution loop.

Technical abilities :

Database and data‑warehouse tools (SQL, Hive, etc.).

Statistical and machine‑learning algorithms; tools like Excel, Python, R.

Data‑visualization tools (Excel, R, PowerPoint, etc.).

Effective analysts consistently demonstrate three traits: a big‑picture view, balanced data‑and‑business thinking, and strong communication skills.

Which Companies Need Data Analysts

Having data does not automatically require analysts. Simple reporting tools (GA, Tableau) can serve many needs. Analysts are needed for complex, customized, insight‑driven tasks.

Companies typically need analysts when:

Decision‑making requires data‑backed reasoning and impact prediction.

Business scale and complexity demand rigorous metric definitions and validation.

Rapid business changes require continuous metric maintenance.

Fine‑grained operations need detailed KPI decomposition and causal analysis.

Data silos cause inconsistent metrics and low trust.

Data infrastructure is in its early stage and needs metric design, source integration, and process definition.

Business needs are vague and require translation into measurable indicators.

How to Build a Data Analyst Team

Key questions for establishing a team:

What is the purpose of the team?

Who are the primary stakeholders?

What are the main responsibilities?

What is the optimal team size?

How will performance be evaluated?

Answers guide recruitment, headcount, skill requirements, and evaluation criteria.

Data Analyst Team Structure and Collaboration

Common pain points: scattered goals, chaotic work, and weak influence. Solving “scatter” requires a company‑wide analysis framework based on a unified business model rather than isolated departmental views.

Steps:

Define a high‑level business model (e.g., users, products, and the shopping experience).

Derive analysis questions from the model (company‑level goals, supporting factors, and their interactions).

Group analysts by these model‑based questions, ensuring shared objectives and reusable data work.

Benefits include aligned goals, macro‑level thinking, concrete business‑level deliverables, and cross‑group knowledge sharing.

What Makes a Good Analysis Report

A report has a clear topic, analytical process, and conclusion. Quality depends on the audience. Readers can be classified by seniority (decision vs. execution) and business familiarity (aware vs. unaware), yielding four categories.

For each audience, standards differ in topic selection, data granularity, analytical depth, conclusion placement, and visualization style.

Overall, report quality hinges on topic relevance, data choice, analytical rigor, clear conclusions, logical structure, and effective visualizations, always tailored to the reader’s needs.

Data Analysis Trinity: Trend, Method, Technique

Trend (势)

Assess whether data analysis aligns with industry and company momentum. Factors include industry data volume, market dynamics, competition, and the company’s data culture, infrastructure, and leadership support.

Method (道)

Refers to the analytical framework, purpose, and value chain. Simpler business models and stable, clear requirements make analysis more effective and faster to deliver value.

Technique (术)

Encompasses analytical thinking and technical methods. Key practices include:

Choosing qualitative vs. quantitative analysis.

Identifying topics from metric monitoring or business problems.

Applying mathematical modeling where possible.

Innovating metrics to capture nuanced business signals.

Balancing macro and micro perspectives.

Introducing additional dimensions when low‑dimensional analysis stalls.

Formulating bold hypotheses and rigorously testing them (e.g., A/B testing).

These principles help analysts turn raw data into actionable insights.