Data analyst
Asks “why did sales drop?” — turns messy Excel into a table you can trust, not an ML slide.
Income range: 4,000,000–28,000,000 UZS / mo
Takeaways
- An analyst is not a coding champion — they sharpen the question and turn messy data into a table you can trust.
- Uzbekistan has no official “analyst salary” series; the role hides inside ICT, finance, or trade. Do not treat the programmer median as your offer.
- Entry: Excel + SQL + one BI tool. Python later. Portfolio is a dashboard, not an “ML model” slide.
What a data analyst does
A data analyst takes a business question — “why did sales drop?”, “which region returns more?” — and answers with a number. The work is finding the source (CRM, 1C, payments, ad cabinets), cleaning, aggregating, showing, and concluding carefully.
It is not a data scientist or ML engineer: a model is not the main product. It is not always a BI developer: cubes and ETL may not be your core job. In a small company the edges blur — you own all of it.
A fit if you like order, distrust a lonely average, and can talk to people. A poor fit if you only want pretty charts or refuse conversations.
The day and the stakeholders
Morning: check yesterday’s dashboard still loads. Then ad-hoc: marketing asks about “yesterday’s campaign,” finance waits on month-close. Afternoon SQL or Power Query; evening a short note in Slack/Telegram.
Stakeholders are half the job. Re-ask: which dates, which currency, VAT on or off. Without written acceptance you enter a “redraw it” loop.
Where the role sits
| Context | Questions | Tools |
|---|---|---|
| Product / IT | Activity, funnels, A/B | SQL, Metabase/Looker, optional Python |
| Marketing | CAC, ROAS, channel | GA4, Ads, Excel |
| Finance / 1C | Margin, receivables, budget | 1C, Excel, Power BI |
| Ops / marketplace | Delivery, returns, SKU | SQL, internal admin |
Path from zero
01Make Excel serious
Tables, Pivot, VLOOKUP/XLOOKUP, a simple chart. Bookkeeper or operator experience is an edge.
02SQL: SELECT through window functions
Mode/SQLBolt/a sandbox. JOIN and GROUP BY are 70% of interviews.
03One BI tool
Power BI (closer to finance/1C) or Metabase/Looker (product). Another human must open the dashboard and get it.
04Portfolio: two stories
A public dataset (sales, not another COVID dump). Question, cleaning, conclusion, limits. Python is optional around month five.
Skills and interviews
Hard: SQL, Excel, BI, simple stats (mean vs median, sampling). Soft: a short conclusion, saying “I do not know,” not lying with a chart.
Interview: write a JOIN; two hypotheses for “why did AOV rise”; spot a lying chart. A take-home CSV is common — ask the timebox and write the limits.
Uzbekistan: where the role exists
Ads: hh.uz “аналитик”, “data analyst”, “marketing analyst”, sometimes “business analyst” (that is another job — requirements/BPMN). Employers: marketplaces, banks, telecom, agencies, public digitization.
Language: SQL in English, reports in Russian/Uzbek. 1C is a strong filter in finance analysis. There is no dedicated series; the ICT mean of 17.41M (H1 2026) does not describe analysts. Junior pay often sits near the programmer mode (8M) or below — read the ad.
Search: hh.uz, LinkedIn, Telegram, internal posts. International remote wants strong SQL, English, and a portfolio.
Honest talk about money
International and AI
Global: SQL + Tableau/Looker + written English. Do not take the data-scientist title early. Relocation is that country’s visa.
AI: ChatGPT drafts SQL and suggests cleaning. It does not know your schema, can leak secrets, and can lie in a JOIN. You own the question, the check, and the metric dictionary. The scarce skill is defining what a KPI means.
Who distrusts a number and builds the table
Mistakes: charting dirty data; “A/B won” with no caveat; pasting PII into personal ChatGPT. Growth: analytics engineer, product analyst, lead of a small pod. Moving to engineering is easier with strong SQL, but engineering is another job.
From Excel to SQL — what people ask
Do I need Python?
Not for entry. SQL + BI is enough. Python helps on repeatable cleaning and large files.
How is this different from a business analyst?
A BA often owns requirements and process; a data analyst owns numbers and queries. Read the ad, not the title.
Should I learn 1C?
In finance/trade analysis — yes, a strong filter. In pure product IT — optional.
How long to learn?
If you already have Excel: 4–8 months of SQL+BI. From zero: 8–14. Year one on the job is the real school.
Where do I get data?
data.gov, Kaggle, or an anonymized export from your company (with permission). Real column names matter more.
Should AI write the report?
A draft is fine; do not send numbers you did not check. The source line stays yours.
Can I handle English SQL docs?
To read SQL and tool docs — yes. For an international team — written B2.
Do I need a statistics degree?
No. Economics, math, or self-study work. Some banks still filter on a diploma.
Figures draw on 2026 sources: Uzbekistan National Statistics Committee (official sector averages, Q1/H1 2026), Flexa and hh.uz vacancy medians, and public IT Park / product-team guides. Not a guarantee — contract type, city, tax, and level change the range. Compare live vacancies yourself.