Data Analyst
The agent writes the SQL, builds the dashboards, and runs the queries. Your day is the questions only a human can ask and the interpretation only a human can make. The craft moves from query-writing to question-framing.
The work
You answer questions decision-makers care about — about customers, product, operations, revenue. The agent handles substantial parts of what used to consume most of your day: writing SQL, building dashboards, running ad-hoc queries, assembling charts, basic data cleaning. You handle the question-framing, the interpretation, the storytelling, and the judgment about what's actually true.
Day-to-day, you:
- Frame the question correctly. When a leader asks "are customers using feature X?", they often mean something more specific. Your job is to translate the request into a precise, answerable question. The agent does not do this well; you do.
- Specify what data to pull and how to interpret it. What counts as "use", what counts as "customer", what time window, what segmentation. The agent runs the query against your specification.
- Curate agent-produced output. Charts, tables, dashboards — the agent builds; you select, refine, and decide what tells the truth without misleading.
- Tell the story. Findings, implications, recommendations. The data doesn't speak for itself; you make it speak.
- Handle anomalies and surprises. When the data shows something unexpected, you investigate before reporting. Bad data, mislabeled metrics, edge cases — the agent doesn't notice; you do.
- Validate at risk-graded gates. Routine reporting flows through agent-only review. Findings that drive major decisions, executive-facing analyses, customer-facing data, and methodologically novel analyses require your direct sign-off.
- Maintain the data model and definitions. What does "active user" mean in this company? What's the official churn calculation? The definitions are the foundation; you tend them.
- Partner across functions. Product, Sales, CS, Marketing, Engineering, Finance — they all ask you questions. With routine query work absorbed, you can engage substantively with each.
What success looks like
Concrete outputs at this tier:
- Decision support quality. Decisions made on your analyses hold up over time. Your analyses inform what actually happens, not just what gets discussed.
- Definition coherence. Company-wide metrics have consistent definitions. People stop arguing about whether they're looking at the same number.
- Time-to-insight. Questions get answered fast — hours to days, not weeks. The agent does most of the work; your judgment turns work into insight.
- Cross-function trust. Leaders across functions trust your numbers and your interpretation. Your work informs strategy.
- Data quality. Issues with data quality, instrumentation, or labeling get caught and addressed. The data foundation improves over time.
What does not count as success: dashboards built, queries run, slides produced, "self-service" reports nobody reads.
What makes this work interesting
The interesting part is not the SQL. It's the interpretation, the storytelling, and the strategic seat at the table.
You ask the questions only a human can ask. "What does this mean?" "Should we be worried?" "What would we do differently if we knew this?" The agent surfaces patterns; you surface meaning.
Cross-function reach widens substantially. Sales, Product, Marketing, CS, Operations, Finance — they all need you. With the query work absorbed, you have time for substantive engagement with each. Few roles see this much of the company.
Interpretation is craft. The data has many possible stories. Choosing the right one — the one that's true, actionable, and honest about uncertainty — is real skill. People who liked data work because they liked making sense of things find this concentrated.
You're a strategic partner, not a report generator. When decision-makers think about big calls, you're in the conversation. The role at T3 is closer to internal consulting than to legacy analytics.
The questions get harder. With routine reporting absorbed, the work that's left is the genuinely interesting questions. Why is this segment churning? What's the leading indicator of expansion? Which feature actually drives retention? The questions reward thinking, not just querying.
Definition work compounds. A clean data model with coherent definitions is one of the most valuable assets a company has. Building and maintaining it is craft; the work compounds across years.
You see the business in real time. Pattern recognition across many analyses gives you a perspective most functional leaders don't have. Data analysts at T3 often become advisors to the executive team well beyond their formal scope.
The career mobility is real. Data analysts who develop at T3 move into Product, Operations, Strategy, Finance, executive analytics roles. The transferable skills — question framing, interpretation, cross-function communication — are valuable everywhere.
What may not appeal. If your craft identity was rooted in SQL artistry — the satisfaction of writing a beautiful query, building a clever dashboard, hand-tuning a complex analytical pipeline — that work absorbs into the agent. Analysts who came to the role for the technical craft of analysis sometimes find the new role more diffuse and less hands-on. You also live in the discomfort of partial information; the questions you answer rarely have clean closure, and people will sometimes ignore what your analysis shows. The legacy data-analyst feeling of "I delivered the report and it's now their problem" mostly disappears at T3 — you're in the conversation about what to do, which is harder and sometimes frustrating.
Who thrives in this role
The aptitudes that matter most at T3 are intellectual, interpretive, and partnership aptitudes — different from query-craftsperson strengths.
You're genuinely curious about what's actually happening. Not just what the data shows; what's true. Analysts who chase the why outperform analysts who deliver the what.
You frame questions well. Most analytical asks are ambiguous on first request. People who can clarify before querying produce more useful analyses than people who pattern-match and dive in.
You're suspicious of clean stories. When the data tells a tidy narrative, you investigate. When the numbers seem too consistent or too dramatic, you check the instrumentation. Healthy skepticism is the analyst's protection against being wrong.
You communicate without losing nuance. Translating complex findings into clear recommendations without flattening uncertainty. Analysts who can hold "this is what we see, here's what's robust, here's what's not" produce trust.
You're patient with definition work. Building coherent definitions of metrics is unsexy but compounding work. Analysts who can't tolerate this work produce inconsistent companies; analysts who can build the foundations everyone else stands on.
You partner across functions without being captured by one. Strong analysts maintain perspective even when working closely with a single function. Analysts who become an extension of Product or Sales lose the independent-judgment value they could otherwise provide.
You write clearly. Findings, memos, dashboards with explanatory text. Analysts who write clearly are listened to; analysts who only deliver numbers without narrative are bypassed.
You handle being wrong well. Analysis is sometimes wrong — bad data, misframed question, missed variable. Analysts who can update gracefully when proven wrong build trust; analysts who defend bad analyses lose credibility.
Less essential than before: depth in any specific SQL flavor, mastery of any specific BI tool, the ability to optimize complex queries by hand, the speed of dashboard production. The agent absorbs these. Your value is in question-framing, interpretation, and judgment.
Skills to develop to get there
The aptitudes describe disposition. The skills below are what you actively build.
Question framing. Turning ambiguous asks into precise, answerable questions. How to practice: before any analysis, write the question you'll answer. Have the asker review. Where they say "that's not quite what I meant" is where your framing needs work.
Specification of analysis. Writing what to pull, how to segment, how to interpret, what counts as success. How to practice: for any analysis, write the spec before the agent runs. Track when the spec missed something; refine.
Skepticism craft. Spotting bad data, misleading visualizations, spurious correlations. How to practice: before reporting any finding, generate three alternative explanations. Test each. The disciplined version of this prevents bad calls.
Storytelling with data. Findings, implications, recommendations — written so non-analysts can act. How to practice: for each major finding, write a one-page memo. Have someone non-analyst read; refine until they can act.
Definition stewardship. Maintaining coherent metric definitions across the company. How to practice: take one metric people argue about. Write the definitive specification. Get cross-function buy-in. The discipline compounds.
Cross-function communication. Writing for Product, Sales, Marketing, CS, Finance, Engineering simultaneously. How to practice: draft a finding. Show to one person from each function. Where they get confused or have different interpretations is where the writing needs work.
Anomaly investigation. When the data shows something surprising, investigating before reporting. How to practice: track surprises. For each, write a brief on whether it was real, an instrumentation issue, or a definitional drift. The pattern is your training.
Decision-relevance discipline. Not running analyses that no one will act on. How to practice: before any new analysis, write what decision it will inform. If you can't name the decision, don't run it.
Pick the skill that maps to your most recent analytical disappointment. Practice it for a month.
How this differs from the legacy Data Analyst role
| Legacy Data Analyst (pre-AI) | Data Analyst (AI-native) |
|---|---|
| Substantial time writing SQL, building dashboards, handling ad-hoc requests | SQL, dashboard building, and ad-hoc queries absorb into agent; time goes to question framing and interpretation |
| Most work is reactive — someone asks, you deliver | Most work is proactive — you frame and prioritize what's worth investigating |
| Dashboards proliferate; few are read consistently | Dashboards are curated; the ones that exist are used |
| Definitions are inconsistent across reports | Definitions are stewarded; coherence is real |
| Best analysts are the most technically fluent | Best analysts are the sharpest question-framers and clearest writers |
| Stakeholders treat analysts as report generators | Stakeholders treat analysts as strategic partners |
| Career path: Analyst → Senior Analyst → Manager of Analytics | Career path: same, plus lateral to Product, Operations, Strategy, executive roles |
The role is not a faster data analyst. It is a different role — the technical craft absorbs and the interpretive craft expands.
Which role evolution patterns are in play
- Elevation (primary, drastic). This is one of the most transformed roles in the catalog. Value migrates from query-craft to question-framing, interpretation, and storytelling.
- Specialization (secondary). The role narrows to its irreducible human core — the interpretive and partnership work that the agent does badly.
- Convergence (partial). Boundaries with Product (analytics), Operations (business intelligence), and Specification Owner (definition work) blur as the analyst role has time for substantive cross-function engagement.
Absorption applies heavily to specific tasks (SQL writing, dashboard production, manual data cleaning). Emergence applies to some new responsibilities (definition stewardship at company scope, agent-output curation).
Related roles in the catalog
Sources & further reading
- Patel, N. (2026). From Tasks to Roles: How Agentic AI Reconfigures Occupational Structures. Data analysis is cited as a role with heavy Elevation pressure.
- Jain, R. et al. (2026). Agentic Generative AI in Enterprise Contexts. Organizational and analytical implications.
- This framework's Reference Framework and Specification Guide.
← Back to Roles · Role evolution patterns · Reference framework · Transforming your role