The specific signals an experienced data analyst hiring panel grades on during the eight-second scan.

• SQL named with the specific dialect (Snowflake, Postgres, BigQuery)

  Generic 'SQL' parses cleanly but dialect specificity is the senior signal.

• BI tool named — Looker, Tableau, Mode, Metabase — not 'BI tools'

  Specific products are hard keyword filters. Listing one you've genuinely shipped in beats listing five you've sampled.

• dbt + transformation layer for mid+ candidates

  Analytics-engineering keyword. Signals you own data models, not just dashboards.

• Analysis paired with the decision it informed

  'Built the churn model whose findings reshaped the Q3 retention roadmap' beats 'built a churn model.'

• Dashboard ownership, not 'reporting support'

  Owning a dashboard is verifiable — what's on it, who uses it, what decisions it drives.

• Five to seven tools in skills, not twenty-five

  Twenty-five tools signals sampling. Five to seven you've genuinely shipped in signals depth.

1. 1

   Lead with the data surface you've owned

   Hiring panels triage analyst candidates by surface ownership first. If you own a dashboard that the Growth team checks every Monday, that information belongs in the first line of your summary. 'Analyst on the Growth team; own the weekly active-users dashboard and the conversion-funnel deep-dive cadence' is twenty words and tells the panel exactly what to ask about.

   For entry-level analysts, the surface might be a capstone project or an internship dashboard. 'Built and shipped the marketing-attribution dashboard during my analytics internship — now used by the founder in weekly fundraising updates' is concrete. 'Familiar with dashboard building' is not.

   Avoid leading with tool names. 'SQL + Python + Tableau analyst' is the resume equivalent of a tool inventory; it tells the panel nothing about the work. The tools belong in the skills section; the summary leads with what you've shipped.

2. 2

   Quantify the analysis by the decision it informed

   Most analyst resumes describe the analysis ('built a churn model,' 'analysed user behaviour'). The differentiator is naming the decision the analysis drove. 'Built the churn model whose findings reshaped the Q3 retention roadmap' tells a panel that the work mattered. 'Built a churn model' tells them nothing.

   The structure: [analysis verb] [specific surface], [result], [decision or outcome]. Examples: • Built the segmentation model for the SMB tier; identified the activation gap that became the H2 roadmap's primary theme. • Analysed the experiment cohort for the pricing v3 rollout; the recommended packaging won board approval and shipped in Q4. • Owned the weekly active-users dashboard; the trend analysis I surfaced in week 8 triggered the lifecycle program rebuild.

   If the analysis didn't drive a decision, soften the claim. 'Provided weekly reporting on user metrics' is honest; inflating it to 'drove strategic decisions' reads as overreach in interview.

   Numbers help, but they're not the only metric. 'Identified the activation gap' is a qualitative signal that interviewers can ask about. 'Reduced churn by 12%' needs a counterfactual you can defend.

3. 3

   Show SQL fluency through the work, not the skills list

   Every data analyst resume lists SQL. The differentiator is showing SQL fluency inside the experience bullets. 'Wrote SQL queries' is filler. 'Authored the weekly cohort-retention query in dbt; runs across 47 million events nightly and feeds the company's primary retention dashboard' is evidence.

   The pattern: name the surface, name the tool stack inline, give a scale detail (event count, table size, query frequency). Examples: • Authored 14 production dbt models across the marketing-attribution stack; runtime stabilised under 90 seconds despite a 6× growth in event volume. • Wrote the multi-touch attribution SQL (PostgreSQL + dbt) feeding the company's weekly board pipeline review. • Optimised the slowest 10 queries in our Looker instance; mean query time fell from 18s to under 4s.

   Query optimisation work is undervalued on most analyst resumes but heavily weighted by senior data engineering hiring panels. If you've done it, name it. 'Tuned three slow queries that were blocking the daily refresh' is a real signal.

4. 4

   Name the partners — PM, finance, marketing — not 'the business'

   'Partnered with stakeholders' is the analyst-resume equivalent of soft skills filler. Every candidate writes it. The signal that gets read is naming specific partner functions and the artifact you produced with them.

   The pattern: [did thing] with [partner function] [resulting in concrete artifact]. Examples: • Partnered with the Head of Growth on the activation deep-dive; the synthesis became the H1 OKR for the growth team. • Co-built the pricing experiment with the PM owning the monetisation surface; the analysis informed the board-approved v3 packaging. • Worked with the FP&A team on the unit-economics model; my contribution (CAC payback by acquisition channel) is now part of the monthly board pack.

   This is also how a hiring panel decides whether you're an analyst who can partner with the business or one who only writes queries when asked. Naming finance, product, growth, marketing, ops as specific partners — with specific artifacts — is the senior signal.

5. 5

   Cut the tool inventory

   The single biggest mistake on data analyst resumes is a 25-tool skills section. 'Skills: SQL, Python, R, Tableau, Power BI, Looker, Mode, Metabase, dbt, Snowflake, BigQuery, Redshift, Excel, Google Sheets, Airflow, Dagster, Prefect, Stitch, Fivetran, Hightouch, Census, Hex, Streamlit, Plotly, Matplotlib, Seaborn...' reads as overreach.

   List what you've actually written code in for at least three months in the last twelve. For most mid-level analysts, that's: • SQL (with the specific dialect — Snowflake, BigQuery, Postgres) • 1 BI tool (Looker, Tableau, or Mode — pick the one you've shipped most in) • 1 transformation layer (dbt is the standard) • 1 scripting language (Python pandas, R tidyverse — pick one) • 1 warehouse (the one your dbt actually runs against)

   Five items. Maybe 7-8 if you genuinely shipped work across more. That's the credible skills section. Twenty-five items signals you've touched things; five signals you've shipped in them.

Each follows the [verb] [object] [number] structure hiring managers grade against. Copy them as a starting point, swap in your own numbers, and read the annotation to understand why each one works.

• Dashboard

  Built and own the weekly active-users dashboard in Looker; the trend analysis I surfaced in week 8 triggered the lifecycle program rebuild.

  Why it works: Names the specific dashboard, the BI tool, and the decision the trend triggered. 'Surfaced in week 8' is precise enough that an interviewer can ask 'what did you notice in that week?'

• Analytics Eng

  Authored 14 production dbt models across the marketing-attribution stack; runtime stabilised under 90 seconds despite 6× growth in event volume.

  Why it works: Names the surface (marketing attribution), the tool (dbt), the count of models (14), and the performance metric. 'Stabilised under 90 seconds' is the senior signal — most analysts don't track query performance.

• SQL

  Wrote the multi-touch attribution SQL (Snowflake + dbt) feeding the company's weekly board pipeline review.

  Why it works: Stack named inline (Snowflake + dbt), and the work feeds a board-level artifact — which is the highest-leverage data work an analyst can do. 'Weekly board' tells the panel the cadence.

• Modelling

  Built the segmentation model for the SMB tier; identified the activation gap that became the H2 roadmap's primary theme.

  Why it works: Segmentation work paired with a roadmap outcome. 'H2 primary theme' is specific enough to be checked in interview. Naming the segment (SMB tier) gives the analysis a clear scope.

• Performance

  Optimised the slowest 10 queries in our Looker instance; mean query time fell from 18s to under 4s.

  Why it works: Query optimisation work that most analysts don't surface. Precise before/after metric. 'Slowest 10' is the kind of detail that proves the work was systematic rather than incidental.

• Partnership

  Co-built the pricing experiment with the PM owning monetisation; the analysis informed the board-approved v3 packaging.

  Why it works: Names the partner function (PM) and the artifact (v3 packaging). 'Board-approved' is the verifiable outcome. 'Co-built' is precise credit-sharing.

• Experimentation

  Designed and shipped 12 experiments end-to-end on the growth team (Statsig + dbt); 5 reached statistical significance and shipped to 100%.

  Why it works: Names the tool stack (Statsig + dbt), the count of experiments (12), and the success criterion (5 hit significance). 'Shipped to 100%' is the conversion from experiment to production.

• Business

  Owned the unit-economics dashboard for the SMB segment; CAC payback breakdown by channel now part of the monthly board pack.

  Why it works: Unit-economics work crosses into finance — naming the partner artifact (monthly board pack) signals seniority. The specific metric (CAC payback by channel) is verifiable in interview.

• Process

  Authored the analytics-engineering style guide for our 6-analyst team; standardised dbt naming + testing across 140 production models.

  Why it works: Process work generalises across the team. 'Style guide' is concrete; '140 production models' gives scale. Naming the team size (6 analysts) tells the panel about the org context.

• Self-serve

  Cut the time from 'PM has a question' to 'analyst has an answer' from 3 days to 4 hours via a self-serve cohort tool.

  Why it works: Cycle-time metrics are rare on analyst resumes but exactly what mid+ hiring panels weight. Before/after (3 days → 4 hours) plus naming the artifact (self-serve cohort tool) makes this verifiable.

• Churn

  Analysed the churn cohort for accounts >$50k ACV; the synthesis drove the enterprise-success team's quarterly intervention list.

  Why it works: Names the segment (accounts >$50k ACV) and the partner team (enterprise success). 'Quarterly intervention list' is the kind of operational artifact that proves the analysis stuck.

• Post-mortem

  Ran the post-mortem on the Q1 revenue miss; identified the leading indicator (top-of-funnel SQL volume) we now monitor weekly.

  Why it works: Post-mortem work signals operational rigor. Identifying a leading indicator is the kind of analytical contribution senior analysts make. 'Monitor weekly' shows the finding stuck.

Same intent, two phrasings. Read why the right column lands on the keep-pile and the wrong column doesn't.

Patterns our writers see most often when reviewing data analyst resumes — each one disqualifies candidates faster than weak experience does.

• Mistake

  25-tool skills section. 'SQL, Python, R, Tableau, Power BI, Looker, Mode, Metabase, dbt, Snowflake, BigQuery, Redshift, Excel...'

  Fix

  Five to seven tools you've actually shipped in for at least three months in the last twelve. Twenty-five reads as overreach and confuses the ATS parser (which weights frequency, not breadth).

• Mistake

  Describing the analysis without naming the decision. 'Built a churn model.'

  Fix

  Pair every analysis with the decision it informed. 'Built the churn model whose findings reshaped the Q3 retention roadmap.' If the analysis didn't drive a decision, say what it surfaced — vague verbs lose the bullet.

• Mistake

  Generic SQL claim. 'Experienced with databases and writing SQL queries.'

  Fix

  Name the specific dialect and the surface. 'SQL (Snowflake + Postgres). Authored 14 production dbt models on the marketing-attribution stack.' Dialects parse as keywords; scale (14 models) is verifiable.

• Mistake

  Listing Python and R when you've only used one. Or listing both when you've only really used one library in each.

  Fix

  List the one you've genuinely used. 'Python (pandas, NumPy, plotting)' is honest. 'Python + R for statistical analysis' from someone who's only ever ggplot'd a course assignment is overreach.

• Mistake

  Vague stakeholder claims. 'Partnered with stakeholders to deliver insights.'

  Fix

  Name the specific partner function and the artifact. 'Partnered with the Head of Growth on the activation deep-dive; synthesis became the H1 OKR.' Naming finance, product, growth, marketing, ops as specific partners is the senior signal.

• Mistake

  Two pages of resume with fewer than 6 years of experience.

  Fix

  One page. Data analyst hiring panels move fast — a second page is rarely opened unless page one earned the read. Cut the internship paragraph, trim education to a single row, and let the most recent role breathe.

• Mistake

  Burying the dashboard or analysis surface inside a long paragraph description.

  Fix

  Name the dashboard as a noun in the first sentence of the role. 'Own the weekly active-users dashboard.' Dashboards are the load-bearing surface for data analysts — surface them prominently.

• Mistake

  Confusing dataset size with analysis quality. 'Analysed 10 million rows of data.'

  Fix

  Replace dataset size with analysis specificity. 'Analysed the experiment cohort for the pricing v3 rollout; recommended packaging won board approval.' Hiring panels know row count doesn't equal insight quality.

Strong verbs lead strong bullets. Replace generic openers (worked on, helped with, was responsible for) with the specific verb that matches what you actually did.

ATS pipelines weight your Skills section as a structured list. Include 15-25 of the items below if they match your experience — not soft skills.