"Composed 15 SQL queries to support reporting needs." That bullet tells a recruiter you wrote some queries. It doesn't say whether those queries saved time, unlocked revenue, or just sat in a folder. "Composed" sounds finished, polished even—but it hides the work that matters.

'Composed' vs 'drafted' — and which belongs on your resume

Both words describe creating something, but they signal different stages. "Drafted" implies you built a first version—rough, needing review, waiting for someone else to approve or refine it. "Composed" suggests you finished the thing and shipped it. For a data analyst resume, neither verb does much work.

If you drafted a report template that your manager revised, "drafted" is honest. If you built a Looker dashboard that the sales team uses daily, "built" or "designed" is stronger because it emphasizes the technical artifact and its adoption. "Composed" often appears when someone wants to sound formal but ends up sounding vague. A recruiter reading "composed a cohort analysis" has to guess: did you write the SQL, design the methodology, present the findings, or all three?

Here's the test: if the thing you made required technical decisions—choosing a join strategy, defining a metric, structuring a data model—use a verb that shows you engineered it. If you wrote prose (an email, a memo, a summary deck), "drafted" or "wrote" is fine. But if you built logic, dashboards, or pipelines, go with "built", "designed", or "automated." For data work, the artifact is code or a data product, and "composed" undersells that.

Most analysts default to "composed" because it feels safe. It's not wrong, but it's forgettable. Swap it for a verb that tells the recruiter what you actually did.

13 more synonyms for 'composed'

Synonym When it fits Resume bullet
Built You created a dashboard, model, or query from scratch Built a Snowflake data model tracking 340K monthly active users across 12 product surfaces, cutting report refresh time from 14 hours to 22 minutes
Designed You made structural or methodological decisions Designed a cohort retention framework in dbt analyzing 89K trial signups, surfacing a 19% week-2 drop linked to onboarding friction
Developed You iterated on something technical over time Developed a churn prediction model in BigQuery ML with 0.81 ROC AUC, scoring 54K accounts monthly and flagging 230 at-risk enterprise customers
Engineered You solved a technical problem with a durable solution Engineered a Looker Explore consolidating 6 legacy views into one source of truth, reducing duplicate metric definitions by 71%
Automated You replaced manual work with code Automated weekly A/B test readouts using Python + Looker API, eliminating 8 hours/week of manual exports across 4 analysts
Constructed You assembled something with multiple parts Constructed a unified attribution model joining Google Analytics, Segment, and Salesforce data across 1.2M sessions, reconciling a $340K revenue gap
Produced You delivered a repeating artifact (report, dashboard) Produced a daily revenue operations dashboard in Tableau tracking 18 KPIs, adopted by C-suite and shared in weekly board updates
Generated You created outputs (reports, insights, forecasts) Generated a demand forecast model in R reducing MAPE from 22% to 9% across 450 SKUs, improving inventory turns by 1.3x
Formulated You defined a framework, methodology, or metric Formulated a customer health score weighting usage, NPS, and support tickets, improving churn prediction accuracy by 34%
Assembled You combined disparate pieces into a whole Assembled a cross-functional KPI dashboard pulling from 5 data sources (Stripe, HubSpot, Zendesk, Mixpanel, Postgres), refreshing every 15 minutes
Created You made something new that didn't exist Created a SQL-based funnel analysis tracking 14 conversion steps from signup to activation, identifying a 41% drop at email verification
Scripted You wrote code to automate or transform data Scripted a Python ETL pipeline migrating 3.2M rows from MySQL to BigQuery nightly, with automated schema validation and Slack alerts
Structured You organized data or logic into a repeatable format Structured a naming taxonomy for 200+ dbt models, reducing onboarding time for new analysts from 3 weeks to 5 days

Three rewrites

Before:
Composed monthly performance reports for leadership team.

After:
Built a monthly exec dashboard in Looker tracking MRR, CAC, and LTV across 8 acquisition channels, reducing report prep time from 6 hours to 12 minutes.

The rewrite swaps "composed" for "built" and names the tool, the metrics, and the time saved. "Composed reports" could mean anything; "built a dashboard" is a technical artifact a recruiter can picture.

Before:
Composed analysis of customer behavior trends.

After:
Designed a cohort analysis in SQL examining 67K users across 9 product features, surfacing that users who enabled notifications had 2.4x higher 30-day retention.

"Composed analysis" is a black box. "Designed a cohort analysis" tells the recruiter you made methodological choices, and the finding (2.4x retention) proves the work mattered.

Before:
Composed data queries to support various business needs.

After:
Automated 22 recurring SQL queries in dbt, eliminating 14 hours/week of manual pulls and standardizing metric definitions across the analytics team.

"Composed queries" sounds like you wrote some SQL once. "Automated recurring queries" shows you replaced repetitive work with infrastructure, and the 14-hour savings quantifies the impact.

When 'composed' is the right word

If you literally wrote prose—an email update, a summary memo, a one-pager for leadership—"composed" or "wrote" is fine. "Composed a weekly data insights email summarizing A/B test results for the product team" is honest and clear.

If your work involved assembling narrative alongside data (a board deck with commentary, a findings memo with SQL appendix), "composed" can work, though "produced" or "delivered" often land harder.

If you're describing a creative or communication artifact rather than a technical one, "composed" won't hurt you. But for dashboards, queries, models, and pipelines—anything that lives in code—reach for a verb that signals engineering work.

Quantifiable vs qualitative verbs

Verbs split into two camps. Quantifiable verbs—reduced, increased, accelerated, automated, scaled—create an expectation that a number follows. If you write "reduced report runtime" and don't include a percentage or time delta, the recruiter notices the gap. Quantifiable verbs commit you to proving impact with a metric.

Qualitative verbs—refined, improved, enhanced, streamlined—let you describe better outcomes without hard numbers. "Improved data quality" or "refined the customer segmentation model" are vague but not dishonest. The problem is that qualitative verbs let you hide. A recruiter can't tell whether "improved" meant a 2% lift or a 10x change, so they discount the claim.

"Composed" lives in qualitative territory. It says you made something but doesn't commit to how good it was, how fast it ran, or who used it. That's why swapping to a quantifiable verb—built, automated, reduced—forces you to add the number that makes the bullet credible. If you can't add a number, your verb choice won't save the bullet. But if you have a number and you're using a qualitative verb, you're leaving signal on the table.

Data analysts live in quantified outcomes. Your resume should show the skills that produced those outcomes, and the verb-number pairing is how you prove both. Pick verbs that demand a metric, then deliver it.

Sorce auto-tailors your resume bullets per application. 40 free swipes/day.

For more: communicated synonym, completed synonym, conceived synonym, configured synonym, convinced synonym