Why pre-screen Tableau specialists before the technical interview
A dashboard that takes twenty seconds to load gets opened once. Most of the cause sits upstream: blending where a join belongs, calculations that should be in the database, extracts that pull everything rather than what is needed. Specialists worth hiring diagnose that rather than adding filters. A short screen asks why one of their dashboards was slow and what fixed it.
What actually matters when screening Tableau Expert candidates
- 01
Technical proficiency
Check fluency with LOD expressions, table calculations, parameter actions and data blending versus relationships; ask which SQL dialects they write and how they tune extracts.
- 02
Systems and trade-offs
Probe how they structure workbooks for scale: extract refresh schedules, Tableau Server or Cloud permissions, row-level security, and when they push logic to the warehouse instead.
- 03
Evidence and rigour
Assess evidence of dashboard performance work: Performance Recorder output, query counts, render times, and how they validated numbers against source systems before publishing.
- 04
Collaboration and communication
Look for how they extracted requirements from finance or ops stakeholders, ran adoption sessions, and handled requests for charts that would mislead.
Pre-screening questions to ask Tableau Expert candidates
12 questions grouped by what they test. Ask the same set in every screen and score answers on a consistent scale, or send them as an async video screen and compare answers side by side.
Dashboards in use
3 questions01Can you describe using this tool to solve a real business problem?
Listen forA specific problem with the decision it supported, and evidence the output is still used.
Dashboards described by their features, or no decision anyone made from the output.
02What is your experience building dashboards?
Listen forDashboards designed around a question, with unnecessary charts removed rather than accumulated.
Dashboards packed with every available measure, or design driven by what looks impressive.
03What is the most complex visualisation you have built?
Listen forComplexity that served the reader, with a simpler alternative considered and rejected for reasons.
Complexity for its own sake, or visualisations that require explanation to read.
Prep done upstream
4 questions04What types of data sources have you connected to?
Listen forDatabases and warehouses used directly, with query performance considered in the connection design.
Sources limited to spreadsheet extracts, or manual data preparation before every refresh.
05Can you explain a case where you combined data from separate sources?
Listen forJoins performed upstream where possible, with blending used only where it is genuinely required.
Blending used as the default, or the performance cost of it not understood.
06How comfortable are you writing queries against the underlying data?
Listen forConfident query writing, with heavy calculations pushed into the database rather than the workbook.
All logic built in the workbook, or an inability to read the query being generated.
07What is your experience with the calculation and expression syntax?
Listen forCalculation types understood including their evaluation order and the effect on query performance.
Calculations copied from forums, or level of detail behaviour not understood.
Performance diagnosed
3 questions08What performance optimisation techniques do you use?
Listen forPerformance recorded and analysed, with the cause fixed at the data layer where appropriate.
Optimisation limited to reducing marks, or performance never measured before changes.
09Have you troubleshot a performance problem, and what was the cause?
Listen forA specific cause identified from a performance recording, with the fix and its effect described.
Slowness attributed to the tool, or problems solved by reducing the data shown.
10What is your knowledge of data extract capabilities?
Listen forExtracts used deliberately with filtering and aggregation applied, and refresh schedules managed.
Full extracts of everything, or refresh failures discovered by users seeing stale data.
Governed properly
2 questions11What is your experience managing security and permissions?
Listen forRow-level security implemented where needed, with access reviewed rather than granted broadly.
Everyone given access to everything, or sensitive data visible in published workbooks.
12What other visualisation tools have you used?
Listen forBreadth with a view on where each fits, showing the tool is chosen rather than assumed.
One tool treated as the answer to everything, or no exposure to alternatives at all.
How to score responses
Score every candidate on the same four criteria immediately after the screen. At this stage you are shortlisting for panel interviews, not making the final call.
Technical proficiency
35%5Explains FIXED versus INCLUDE order of operations unprompted, writes performant custom SQL, and tunes hyper extracts with measurable load time gains.
Systems and trade-offs
25%5Weighs live connections against extracts with cost and freshness reasoning, and describes governance choices made across published data sources.
Evidence and rigour
25%5Cites before and after load times, names the culprit (nested calcs, quick filters, cross-database joins), and shows a reconciliation process.
Collaboration and communication
15%5Describes redesigning a requested pie chart or gauge with a defensible alternative, plus concrete usage or adoption numbers afterwards.
A dashboard that takes twenty seconds to load gets opened once. A one-way video screen asks why theirs was slow.
Try it on HirevireScreening FAQ
Process basics
How long should a pre-screening round for this role take?
Ten to fifteen minutes across eight to ten questions, answered async. Enough to establish dashboards in use, test their data preparation depth, and check performance and governance.
How much data engineering should I expect?
Enough to write the query rather than compensate in the workbook. Someone who solves data problems with blending and complex calculations will build workbooks nobody can maintain.
Evaluating answers
What is the strongest signal when screening this role?
Why a dashboard of theirs was slow and what fixed it. Real answers involve moving work upstream. Anyone who says they added filters has treated the symptom rather than the cause.
How do I judge whether their work gets used?
Ask which dashboard is opened most and by whom. Real answers include usage data and a decision it supports. Anyone who does not know has built reporting nobody asked for.
Go deeper on this role
Sanat has been hiring since 2012 and watching the recruitment industry change up close ever since, and turned that screening process into Hirevire's video screening platform. LinkedIn
