What Does a Data Analyst Do?

What does a data analyst do all day? The short answer is that they turn messy data into decisions a business can act on. A typical week mixes SQL queries, dashboard updates, stakeholder questions, and one stubborn spreadsheet that refuses to match the warehouse. The role is less about memorizing tools and more about asking the right question, proving the number, and explaining what it means.

That distinction matters if you are hiring, switching careers, or trying to understand why an analytics team keeps asking for more time. Good analysis is not a chart factory. It is a chain of small, verifiable decisions that ends with someone changing a price, a campaign, a schedule, or a process.

The Short Answer: Evidence, Not Just Charts

A data analyst sits between raw data and the people who make decisions. They define the metric, pull the right records, check whether the data is trustworthy, look for patterns, and present a recommendation in plain language. The job title changes by company, but the core loop is stable: question, data, validation, analysis, explanation, action.

For example, a subscription business may ask why cancellations rose last month. The analyst first confirms how churn is defined, then checks whether the increase is real or caused by a reporting change. Next they segment by plan, tenure, channel, and region. Finally they show the board which segment moved and what the team should test next. The deliverable may be a dashboard, a memo, or a five-slide review. The value is the decision it supports.

That is the practical answer to what does a data analyst do: they make the next decision less uncertain.

7 Tasks That Fill a Typical Week

1. Clarify the business question

Most requests arrive as a symptom: sales are down, users are unhappy, the report looks wrong. A strong analyst turns that into a testable question with a definition, a time window, and a decision owner. They ask what will change if the answer is yes, and what would count as a false alarm. This step prevents a week of work on a metric nobody will use.

2. Pull and clean data

Analysts write SQL to join orders, customers, events, and finance records. They check row counts, duplicates, nulls, time zones, and currency mismatches before trusting a result. Cleaning is often 40 to 60 percent of the work. A fast query on bad data is worse than a slow query on good data because it creates confident mistakes.

3. Explore patterns before building a model

Before reaching for machine learning, a data analyst looks at distributions, cohorts, seasonality, and outliers. Simple comparisons often answer the question faster. If a segment has only 12 customers, a fancy model will not rescue the finding. Analysts separate signal from noise and state the limits of the sample.

4. Build dashboards people actually use

A dashboard is a product, not a data dump. The best ones answer three to five recurring questions, show definitions, and make the next action obvious. Analysts choose the right chart, remove vanity metrics, and add filters only when they change a decision. A clean weekly view beats a 30-tab workbook nobody opens.

5. Explain findings in plain language

The analyst translates coefficients, p-values, and query logic into a short story: what happened, why it matters, what to do next. They lead with the recommendation, then show the evidence. They also name uncertainty and the assumptions behind the number. Communication is not a soft extra; it is how analysis becomes value.

6. Automate recurring reports

Monthly reporting should not consume three days every month. Analysts move repeatable queries into scheduled jobs, document the logic, and add alerts for broken pipelines. Automation frees time for ad hoc questions and reduces the chance that two teams report different numbers for the same metric.

7. Protect data quality and access

Analysts are often the first line of defense for data quality. They flag missing fields, inconsistent categories, and permissions that expose customer data. They follow privacy rules, limit access to raw records, and keep an audit trail. Trust is part of the job, especially in finance, health, and customer analytics.

Data Analyst vs Data Scientist vs Business Analyst

The titles overlap, which is why job descriptions can feel confusing. The simplest split is the question each role answers most often.

Role	Main question	Common tools	Typical output
Data analyst	What happened, and why?	SQL, spreadsheets, BI tools	Dashboards, reports, recommendations
Data scientist	What is likely to happen next?	Python or R, statistics, ML libraries	Models, experiments, forecasts
Business analyst	What should the business change?	Process maps, requirements, spreadsheets	Requirements, process improvements, business cases

A data analyst usually works closer to reporting and decision support. A data scientist spends more time on modeling and experimentation. A business analyst focuses on process, requirements, and stakeholder alignment. Small companies often combine all three into one role, so read the tasks rather than the title.

7 Tools Most Data Analysts Use

You do not need every tool on this list to start. Most teams care about one query language, one spreadsheet, and one visualization layer. The rest are advantages when the data stack grows.

So what does a data analyst do with this stack? They use the simplest tool that can answer the question safely, then document the step so the next person can repeat it.

Tool	What it does	Where it fits
SQL	Queries and combines data from databases	The core skill for almost every analyst role
Excel or Google Sheets	Quick analysis, modeling, and checks	Fast ad hoc work and stakeholder handoff
Power BI, Tableau, or Looker	Builds dashboards and shared reports	Recurring metrics and self-serve reporting
Python or R	Cleans data and runs advanced analysis	Larger datasets, automation, and statistics
dbt	Organizes SQL transformations	Teams with a warehouse and version control
Git	Tracks changes to queries and models	Collaboration and reproducible analysis
BigQuery, Snowflake, or Redshift	Stores and processes large datasets	Cloud data warehouses

Learn SQL deeply before collecting tools. Then add a BI tool and a spreadsheet. Python becomes useful when you hit repetitive cleaning, statistical testing, or datasets too large for a spreadsheet. dbt and Git matter more on teams that treat analytics like software, but they are easier to learn after you understand the data model.

Skills That Matter More Than Tools

SQL fluency. You should be comfortable with joins, filters, aggregations, window functions, and common table expressions. Most technical screens test whether you can answer a business question from several tables without guessing.

Statistics you can defend. You need to understand averages, medians, distributions, sampling, confidence intervals, and the difference between correlation and causation. You do not need a PhD, but you do need to know when a result is too fragile to act on.

Business context. A data analyst who understands how the company makes money will ask better questions. Learn the revenue model, the customer lifecycle, and the three metrics leadership watches every week.

Communication. The job is full of translating. You will explain a join to an engineer, a churn number to a marketer, and a forecast to a finance lead. Clear writing and a short recommendation are competitive advantages.

Data quality habits. Check the source, document the transformation, and keep a query you can rerun. The best analysts are boring in the best way: their numbers reconcile, and their work is easy to audit.

How to Become a Data Analyst in 90 Days

A three-month plan will not make you an expert, but it can produce a credible portfolio and a clear interview story. Treat it as a sprint with evidence at the end, not a tour of every course online.

The fastest way to answer what does a data analyst do is to finish one real project, not watch another tutorial.

Weeks 1–2: Learn the business basics

Pick one industry and learn its core metrics. For ecommerce, study conversion rate, average order value, repeat purchase rate, and contribution margin. For SaaS, study activation, churn, expansion, and customer acquisition cost. Write a one-page glossary so you can explain each metric without jargon.

Weeks 3–6: Get fluent in SQL and one BI tool

Practice joins, aggregations, window functions, and date logic on a public dataset. Build five queries that answer real questions, then recreate them in Power BI or Tableau. Focus on clean labels, sensible filters, and a dashboard that fits on one screen.

Weeks 7–10: Build three portfolio projects

Choose projects that show different skills: a sales dashboard, a cohort retention analysis, and a data cleaning case. For each one, write the question, the data source, the method, the finding, and the limitation. Publish the SQL and a short README. A reviewer should be able to reproduce your result.

Weeks 11–12: Apply and rehearse

Tailor your resume to the tools in each job post. Prepare three stories: a time you found a costly data error, a time you changed a decision, and a time you explained a complex result to a non-technical audience. Expect a SQL screen, a take-home analysis, and a business case interview. For a deeper academic route, compare costs and outcomes in this guide to a data science masters degree.

If you need a structured way to organize sources and notes during the sprint, use a research toolkit rather than scattering links across browser tabs.

Salary and Career Path

Pay depends on market, industry, seniority, and whether the role is remote. The U.S. Bureau of Labor Statistics groups data analysts with data scientists in its official outlook, so treat the number as a benchmark rather than a guaranteed starting salary. Its BLS occupational data reports a median annual wage of $120,230 for data scientists in May 2025 and projected growth of 35 percent from 2025 to 2035.

A common path looks like this: junior analyst, data analyst, senior analyst, then analytics manager, data scientist, or analytics engineer. Each step adds scope. Junior roles focus on reporting and clean queries. Senior roles own a domain, design metrics, and mentor others. Managers own the roadmap and the team. Specialists go deeper into experimentation, finance, marketing, or machine learning.

To move up, build evidence of business impact. A dashboard that saved five hours a week is useful. A analysis that changed pricing, reduced churn, or exposed a costly data error is a promotion case. Keep a short record of the question, the action, and the result. For a view of how analytics and AI roles connect, see this guide to data analytics and AI careers.

Common Mistakes That Slow Analysts Down

• Starting with the tool. A beautiful dashboard cannot fix a vague question. Define the decision first, then choose the tool.
• Skipping validation. Check row counts, duplicates, and date ranges before you share a number. One bad join can erase a week of trust.
• Writing SQL nobody can read. Use clear aliases, comments, and common table expressions. Future you will inherit the query.
• Presenting without a recommendation. A list of charts pushes the thinking back to the audience. Say what you would do and why.
• Ignoring the cost of maintenance. Every recurring report needs an owner, a definition, and a failure alert. Otherwise it becomes stale quietly.

Bottom Line

What does a data analyst do? They make business questions answerable with evidence, then turn that evidence into a decision someone can act on. The tools matter, but the durable skills are SQL, statistics, business context, and clear communication. Start with one real question, one clean dataset, and one recommendation. That is the job in miniature, and it is the fastest way to prove you can do it at scale.

FAQ