AI automated 30-40% of the old analyst job. The remaining 60% pays better than ever. Here is what the role actually looks like now.
Get new posts delivered to your inbox. No spam, unsubscribe anytime.
Last year, a junior analyst on my team automated herself out of 15 hours of weekly work. She wrote Python scripts that pulled data from three APIs, cleaned it, generated the standard Monday report, and emailed it to the marketing team — all before anyone got to the office.
Her manager panicked. "If we automate the reports, what does she do all day?"
What she did was spend those 15 hours on the questions nobody was asking. She found that our highest-spending customers had a 40% churn rate in the first 90 days. That insight led to a retention program that saved the company roughly $2M in annual revenue.
She got promoted. The report still runs on autopilot.
That's the data analyst story in 2026. The routine work is disappearing. The people who used to do that work? The good ones are more valuable than ever.
Data analyst salaries have shifted significantly. Here's the 2026 breakdown from Glassdoor:
| Experience Level | Median Salary | 25th Percentile | 75th Percentile |
|---|---|---|---|
| Entry-level | $63,107 | $49,660 | $80,737 |
| Mid-level | $93,060 | $71,951 | $121,526 |
| Senior | $131,224 | $105,797 | $164,380 |
The average jumped to $111,000 in Q1 2025 — a $20,000 increase from early 2024. That's not inflation. That's the market saying analysts who can do more than pull reports are worth paying for.
On the demand side, data science careers are projected to grow 36% from 2023 to 2033, with roughly 11.5 million new jobs in data science and analytics expected by late 2026. The BLS projects 317,700 annual openings in computer and IT occupations, growing "much faster than average." There's an estimated global shortage of 250,000 unfilled data analyst roles.
The jobs are there. But they're not the same jobs they were two years ago.
Let's be direct about this because the clickbait headlines aren't helpful.
AI has automated roughly 30-40% of tasks that used to fill a typical analyst's week. The stuff that's gone or going:
The stuff AI can't do:
Harvard's career services put it simply: analysts aren't replaced by AI — they're replaced by analysts who use AI. The roles being eliminated are primarily junior report-generation positions where the primary output was recurring dashboards and scheduled queries.
If your entire job is pulling the same report every Monday, yes, you should be worried. If your job involves original thinking about messy, ambiguous business problems, you're fine. Better than fine — you're in demand.
Here's what companies actually expect you to know, based on job postings and hiring patterns:
SQL — Appears in 85-95% of data analyst interviews. Not basic SELECT statements — you need window functions, CTEs, complex JOINs, and subqueries. This is the single skill that determines whether you get past the first round.
-- This is the kind of SQL they test you on.
-- Not "SELECT * FROM users" — real analytical queries.
WITH monthly_revenue AS (
SELECT
customer_id,
DATE_TRUNC('month', order_date) AS month,
SUM(amount) AS revenue
FROM orders
WHERE order_date >= CURRENT_DATE - INTERVAL '12 months'
GROUP BY customer_id, DATE_TRUNC('month', order_date)
),
ranked AS (
SELECT
customer_id,
month,
revenue,
LAG(revenue) OVER (PARTITION BY customer_id ORDER BY month) AS prev_revenue
FROM monthly_revenue
)
SELECT
customer_id,
month,
revenue,
prev_revenue,
ROUND(((revenue - prev_revenue) / NULLIF(prev_revenue, 0)) * 100, 1) AS pct_change
FROM ranked
WHERE prev_revenue IS NOT NULL
ORDER BY pct_change DESC
LIMIT 20;
Excel — Still everywhere. Pivot tables, VLOOKUP/XLOOKUP, conditional formatting, basic macros. Don't underestimate this — it's the first tool most hiring managers expect.
One BI Tool — Either Tableau (28.1% of postings) or Power BI (24.7%). Pick one, get good at it. Power BI is easier to start with if you know Excel. Tableau gives more visual control.
Python — Pandas, NumPy, basic matplotlib/seaborn. You don't need to be a software engineer. You need to automate data cleaning, run analyses that are too complex for SQL, and build simple scripts that save you hours.
import pandas as pd
# Real analyst work: clean messy data, calculate metrics
df = pd.read_csv('transactions.csv')
# Handle the mess
df['date'] = pd.to_datetime(df['date'], errors='coerce')
df = df.dropna(subset=['date', 'amount'])
df = df[df['amount'] > 0] # remove refund artifacts
# Monthly cohort retention analysis
df['cohort_month'] = (
df.groupby('customer_id')['date']
.transform('min')
.dt.to_period('M')
)
df['order_month'] = df['date'].dt.to_period('M')
df['months_since_first'] = (
(df['order_month'] - df['cohort_month']).apply(lambda x: x.n)
)
retention = (
df.groupby(['cohort_month', 'months_since_first'])['customer_id']
.nunique()
.unstack(fill_value=0)
)
# Normalize to percentages
retention = retention.div(retention[0], axis=0).round(3) * 100
print(retention.head())
Statistics — Not textbook-level. Practical: hypothesis testing, A/B test analysis, correlation vs. causation, confidence intervals. You need to know when a number is statistically meaningful and when it's noise.
| Tool | Priority | What It's For | Time to Learn |
|---|---|---|---|
| SQL | Must-have | Data extraction, analysis | 2-4 weeks basics, months to master |
| Excel | Must-have | Quick analysis, presentations | 1-2 weeks for advanced features |
| Tableau or Power BI | Must-have | Dashboards, visualization | 3-4 weeks |
| Python + Pandas | Expected | Automation, complex analysis | 4-8 weeks |
| Statistics | Expected | Validating insights, A/B tests | 4-6 weeks |
| Git | Nice-to-have | Version control | 1 week |
| dbt | Nice-to-have | Data transformation | 2-3 weeks |
| Cloud DW | Nice-to-have | Big data querying | 1-2 weeks if you know SQL |
They tell you to learn everything at once. Excel AND SQL AND Python AND Tableau AND R AND statistics AND machine learning. Then get a certification. Then build 10 portfolio projects.
That's not a roadmap. That's a recipe for burnout.
Here's what actually happens when someone gets hired as a data analyst in 2026:
That's it. That's the bar for entry-level. Everything else is a nice-to-have that you learn on the job.
The obsession with certifications is particularly misleading. Nobody has ever been hired because they had a Google Data Analytics Certificate. They've been hired because they could demonstrate skill — and the certificate was one way they built that skill. The portfolio matters. The cert is decoration.
Week 1-4: Excel. Yes, really. Pivot tables, VLOOKUP/XLOOKUP, conditional formatting, basic charts. Use real datasets — not tutorials with 50 rows. Download a Kaggle dataset with 100K+ rows and explore it.
Week 5-8: SQL. Start with SELECT, WHERE, GROUP BY, JOIN. Then window functions, CTEs, subqueries. Practice on SQLPad, LeetCode SQL, or DataLemur. Aim for 10-15 problems per week.
Week 9-12: Pick Tableau or Power BI. Build three dashboards from real data. One should tell a complete story — not just charts, but insights with context.
Week 13-16: Build your first portfolio project. Pick a business question, find a public dataset, do the analysis in SQL + your BI tool, write up your findings. Structure: business problem, approach, analysis, key finding, business impact.
Week 17-20: Python basics + Pandas. Focus on data cleaning, transformation, and basic analysis. Don't try to learn everything — learn enough to automate the tasks that annoy you in Excel.
Week 21-24: Second and third portfolio projects. One should involve Python. One should be SQL-heavy. Both should answer real business questions.
Week 25-28: Practice SQL interview questions. Prepare three portfolio walkthroughs using the structure: business problem, approach, analysis, key finding, impact.
Week 29-32: Apply. Two to four weeks of focused interview prep is the sweet spot. Spend 1-2 hours daily on SQL practice, review stats fundamentals, polish your portfolio.
This timeline assumes you're learning part-time (10-15 hours/week) alongside a job or school. Full-time learners can compress to 3-4 months.
Here's something the boot camps don't tell you: domain expertise is becoming one of the most important differentiators in 2026.
A data analyst who understands healthcare terminology and processes is worth significantly more in a hospital system than a generalist with the same SQL skills. A data analyst in fintech who understands loan underwriting can ask better questions and catch errors that a generalist would miss.
If you're transitioning careers, your previous domain knowledge is an asset, not a sunk cost. A nurse who becomes a healthcare data analyst. An accountant who becomes a finance data analyst. A marketer who becomes a marketing data analyst. These people get hired faster and paid more because they already understand the business.
The data analyst role isn't a dead end — it's a launchpad. Here's where people actually go:
Technical track: Senior Data Analyst → Data Scientist → ML Engineer. This requires deepening your Python, statistics, and machine learning skills. Timeline: 2-4 years to data scientist.
Leadership track: Senior Data Analyst → Analytics Manager → Head of Analytics → CDO. This requires growing your stakeholder management, team leadership, and strategic thinking. Timeline: 4-7 years to analytics manager.
Specialist track: Senior Data Analyst → Analytics Engineer → Data Engineer. This requires learning dbt, Airflow, cloud infrastructure. Growing fast as companies build modern data stacks.
Lateral track: Data Analyst → Product Analyst → Product Manager. Common in tech companies. Your data skills become product intuition.
The highest-paid analysts in 2026 aren't the ones with the most tools on their resume. They're the ones who understand their business deeply and can translate data into decisions. That takes time and can't be shortcut.
I've been a data analyst. I've hired data analysts. Here's what I believe:
The entry-level data analyst job is harder to get than ever, but the mid-level and senior roles are easier. There's a glut of boot camp graduates who know basic SQL and have identical Titanic dataset projects. But companies are desperate for analysts who can actually think — who understand the business, ask the right questions, and communicate clearly. The supply of technically competent generalists is high. The supply of business-aware problem solvers is low.
Python is not optional anymore. I know some articles say "you can be a data analyst with just SQL and Excel." Technically true. Practically limiting. Every competitive candidate in 2026 has at least basic Python. The ones who don't are competing for a shrinking pool of Excel-only roles that pay less.
AI makes good analysts better, not redundant. The analyst who uses ChatGPT to write boilerplate SQL and then spends the saved time on deeper analysis is more productive than ever. The analyst who only knew how to write boilerplate SQL? That's the one in trouble.
Domain knowledge beats tool knowledge. I'd hire an analyst with solid SQL, one BI tool, and deep understanding of our industry over someone with five certifications and no business intuition. Every time.
The best career advice I can give: stop learning tools and start solving problems. Find a messy dataset, ask an interesting question, and answer it. Do that 20 times and you'll be a better analyst than someone who completed 10 online courses. The market rewards demonstrated ability, not credentials.
Build things. Break things. Show your work.
Ismat Samadov builds data pipelines and writes about the craft of turning messy data into working systems at ismatsamadov.com.