How to write it
Name the decision, not the dashboard
Analysts are hired to change what a business does. A resume that lists reports built describes the activity; a resume that names a decision describes the value. "Ran the markdown analysis that reshaped buying for a $40M category, recovering $2.1M in margin" is the shape to aim for.
You will not have a dollar figure for everything, and you shouldn't invent one. Where the money isn't attributable, use the decision itself: a strategy that changed, a programme that stopped, a process that was retired, a forecast that got adopted.
Prove the SQL with the shape of the work
Everyone writes SQL on a data analyst resume, so the word carries almost no information. What carries information is what you built with it: a 60-model dbt project, a semantic layer, a cohort analysis across four systems, a reconciliation that found an error nobody else had.
The same applies to Excel. "Advanced Excel" is unverifiable; "rebuilt the planning model 40 people use" is not. Let the artefact establish the skill level and use the skills section only for parsing.
Automation is the analyst's most reliable metric
Almost every analyst has removed manual work, and almost none of them put a number on it. Hours per month recovered, a reporting cycle shortened, a spreadsheet retired, ad-hoc requests halved — these are easy to quantify honestly and they read as maturity, because they show you improved the system rather than just serving it.
This is also the bullet that travels best across industries, which matters if you're changing sector.
- Hours of manual work removed per week or month
- Reporting cycle time, before and after
- Number of people using what you built
- Requests reduced by self-service reporting
- An error you found, and what it had been costing
Finding the error is a real achievement
Data quality work feels unglamorous and is enormously valuable, because a business making decisions on wrong numbers is worse off than one with no numbers. If you have found a material error — a misattributed return, a double-counted channel, a broken join in a report everyone trusted — that belongs on the page.
Write it with the consequence: "found and fixed a returns-attribution error that had overstated category profitability by 8%". It demonstrates rigour, scepticism and ownership in one line.
Show the domain, not just the tooling
Analyst roles are unusually domain-bound. A retail analyst is expected to know what markdown, sell-through and open-to-buy mean; a fintech analyst is expected to know cohort retention and unit economics. Using the language of the sector correctly is a strong signal, and it's invisible in a tool list.
If you're moving between industries, keep the domain vocabulary of your target where it's genuinely transferable, and lead with the methods and automation wins that don't depend on sector at all.