Data analyst postings are keyword-dense in a way that helps you: they name the query language, the BI tool, and often the exact warehouse. That specificity cuts both ways — 'Python' on a resume invites a question about what you actually did with it, and 'machine learning' on an analyst resume invites a harder one. List the tools you have shipped work in, not the ones you have watched a course about.
Hard skills
These are what a technical screen tests. Postings differ sharply on how much statistics they expect, so read the posting before deciding how much of this section applies.
- SQL — The single most-requested analyst skill and the one most likely to be tested live — put it where it is unmissable.
- Data cleaning — Honest about where the time actually goes, and postings that mention messy data are asking for exactly this.
- Data visualisation — Distinct from building dashboards: it is about whether the chart answers a question. Spell it the way the posting does.
- Dashboard development — Names a deliverable rather than a skill; pair it with who used the dashboard and for what decision.
- A/B testing — Common in product-analytics postings; only claim it if you can talk about what made a result significant.
- Statistical analysis — Broad, so make it concrete in a bullet — which test, on what question.
- ETL — Signals you moved and shaped data rather than only querying it; frequently the line between analyst and analytics engineer.
- Data modelling — Appears in warehouse-adjacent postings; means schema design, not machine learning — do not conflate them.
- Reporting automation — Turns a routine task into evidence of judgement; strong when you can say what it replaced.
- Requirements gathering — The half of analyst work that is talking to people; underrated on resumes and named in most senior postings.
Use only the keywords that are true for your experience. Screening software matches words, but a human reads them next — and a claim you cannot back up in an interview costs more than a missed keyword.
Tools and platforms
Matched literally. Write the tool the posting names, with its real capitalisation, and only if you have used it on real work.
- Python — Expected in many postings, but say what for — pandas for analysis reads differently from scripting.
- pandas — More specific than 'Python' and closer to what analyst work actually is.
- Tableau — One of the two default BI tools; postings usually name one, so match it.
- Power BI — The other default, common in Microsoft-stack companies — spell it with the space.
- Looker — Names a modelling-first BI approach; suggests LookML experience, so only claim it if you have it.
- Excel — Still named in a large share of postings; leaving it off to look advanced costs you literal matches.
- BigQuery — Warehouse-specific; name the warehouse you used rather than writing 'cloud data warehouse'.
- Snowflake — The other common warehouse; matched literally and increasingly a posting requirement.
- dbt — Signals modern transformation work; lowercase, and a genuine differentiator where postings ask for it.
- Git — Shows your analysis is version-controlled rather than living in one notebook.
Soft skills
For analysts these are mostly about being understood. They are worth including only where a bullet shows the outcome — a chart nobody acted on is not a result.
- Stakeholder communication — The skill that separates analysts who get promoted; show it by naming who you presented to.
- Data storytelling — Named in product and marketing analytics postings; means the finding changed a decision.
- Attention to detail — Weak as a claim, but relevant here because analyst errors are expensive — earn it in a bullet about validation.
- Problem framing — Senior postings want someone who questions the request, not just answers it.
- Cross-functional collaboration — Analysts sit between teams; say which teams rather than using the phrase alone.
Bullet verbs
Lead with the action, and prefer verbs that imply a decision followed. 'Analysed' with no consequence is the most common weak bullet on analyst resumes.
- Built — Concrete and true of pipelines, dashboards and models; stronger than 'worked on'.
- Automated — Implies something used to be manual — say what, and how often it ran.
- Identified — The right verb when the finding was yours rather than requested.
- Reduced — Works for query time, error rates and manual hours; needs a before and after.
- Modelled — Precise for schema and forecast work where 'analysed' is vague.
- Presented — Shows the work left your screen; name the audience.
How to use these
Put each term where it is actually true — in your skills section if it is a capability, in a bullet if it is something you did, in your summary if it is what you are known for. Match the wording the posting itself uses, since screening software compares strings rather than meanings. Do not repeat a term to raise a count: it does not help you with the software and it reads badly to the person after it.