Live Connection vs. Extract in Tableau: A Certification Guide

A Tableau live connection sends queries to the underlying database when Tableau needs data for the view. A Tableau extract is a stored, compressed snapshot of source data that Tableau uses until it is refreshed. Choose live when the decision requires current source results and the database can support the query workload; choose an extract when a governed snapshot, faster analytical response, or controlled refresh schedule better fits the use case.

Verified August 6, 2026: the definitions, refresh behavior, and limitations below were checked against current Tableau Help. The scenarios are original practice cases, not released certification questions.

Start with the question: current compared with what?

Candidates often reduce this choice to “live equals slow and extract equals fast.” That shortcut is wrong because neither label answers the actual requirement. First ask what “current” means to the stakeholder. A call-center supervisor monitoring open cases during a shift may need the database result at the moment the dashboard is viewed. A finance team reviewing yesterday’s closed transactions at 8 a.m. may need a reliable morning snapshot, not every transaction posted during the meeting.

Tableau's Refresh Data Sources documentation says a live connection sends queries to the database and updates the view based on the results. It also says that refreshing an extract queries the source from which the extract was created and rebuilds the extract. That creates a useful rule: a live view’s freshness is tied to the query and source availability; an extract’s freshness is tied to its last successful refresh.

What a live connection does

A live connection is a direct analytical path from Tableau to a data source. When a viewer filters a dashboard or opens a sheet, Tableau sends the relevant queries to the database. The precise fields Tableau queries are defined as part of the connection, and a data-source refresh can update new or changed fields in Tableau Desktop.

The advantage is freshness. A warehouse manager who needs to see current inventory exceptions can query the operational or analytical source and make a decision from the present result. The trade-off is dependency: response time depends on database performance, network conditions, drivers, query design, concurrent demand, permissions, and the source’s ability to answer the query. “Live” does not mean instant.

Use live carefully when a dashboard sends heavy queries to a production system. Tableau’s performance guidance recommends testing close to the data, using native database drivers, and understanding how the environment behaves before building a complex workbook. A live dashboard that forces a slow source to execute expensive calculations for every interaction may be technically current but operationally unhelpful.

What an extract does

Tableau describes an extract as a compressed snapshot of data stored locally and loaded into memory as needed to render a visualization. It is not a permanent connection to a database. The value is that Tableau can work from a prepared copy designed for analytic queries, often reducing dependence on a slow or heavily used source.

An extract has a freshness contract. If the source changes at 10:00 a.m. and the extract refreshes at midnight, the dashboard may correctly display the snapshot but not the 10:00 a.m. change. A candidate should be able to say that clearly to a stakeholder. “The dashboard is refreshed daily” is meaningful only when the reader knows the cutoff and whether the last refresh succeeded.

Extracts also invite deliberate data reduction. Tableau’s extract performance guidance notes that extract filters can exclude data you do not need. That can reduce volume and focus a workbook, but it can also remove records a future question requires. Document the filter’s business purpose. Never silently exclude a region, date range, or product line and call the result a company total.

Compare the two options by requirement

This is a decision aid, not a rule engine. A current operational dashboard may use a carefully optimized extract if a few-minute lag is acceptable. A monthly executive report may use a live connection if governance requires the source to remain authoritative. State the service expectation instead of arguing from product labels.

Worked scenario: inventory exception dashboard

An operations manager asks, “Which products are below reorder point right now?” The data changes as shipments arrive and orders are allocated. The word right now is a freshness requirement. Begin by testing a live connection against the approved inventory source. Check how long the query takes, whether filters are pushed to the source, and whether the database owner approves the expected traffic.

If the source cannot tolerate interactive demand, do not automatically choose a nightly extract. A nightly snapshot could hide a shortage that began after the refresh. Explore an appropriate governed refresh cadence or an optimized source view with the data team. The chart choice comes after the data contract: a quick bar chart of shortage quantity is misleading if its timestamp is stale.

Worked scenario: weekly executive sales pack

An executive team reviews closed sales every Monday morning. The agreed reporting cutoff is Sunday at 11:59 p.m.; late adjustments are handled in the next cycle. An extract refreshed after the cutoff is a strong fit because it gives every reader the same snapshot and avoids repeated queries during the meeting. Display the refresh timestamp and use a monitored schedule so a failed refresh does not quietly become last week’s report.

This is not “extract because executives are not technical.” It is extract because the agreed question is about a period-close snapshot. The technical choice follows the business definition.

Refreshes, publishing, and changing the connection type

Tableau’s web extract documentation explains that published data sources can be live or extracts and that the Data Sources tab shows the type. It also documents ways to change an embedded data source from live to extract after publishing. However, details differ by environment and source. The documentation notes limitations: credentials may need to be embedded; some embedded sources cannot be extracted in particular configurations; and file-based sources do not gain a performance benefit from adding an extract.

These details matter because “switch it later” is not a complete plan. Before publishing, identify the data owner, credentials approach, refresh owner, failure notification, and expected latency. If a credential expires or a refresh fails, an extract dashboard may remain visible but increasingly stale. If a live source changes schema, Tableau can surface missing-field problems after refresh. Both options need monitoring.

A practical certification reasoning checklist

1. State the required data freshness in business language.
2. Identify the source and the expected query workload.
3. Test a simple view before building a large dashboard.
4. Decide whether a snapshot can answer the question.
5. If using an extract, define refresh cadence, owner, and failure response.
6. If using live, test source load, permissions, and interaction performance.
7. Communicate the last-refresh time or source-update expectation to readers.

The current Salesforce Tableau Data Analyst prep guide includes connecting and transforming data, analysis, creating content, and publishing/managing content. A strong answer connects the technical mechanism to a stakeholder’s decision. “Extract” alone is not reasoning; “extract because the report is a Monday period-close snapshot and the refresh is scheduled after the agreed cutoff” is.

FAQ: Is a Tableau extract always faster than a live connection?

No. Extracts are designed for analytic response and can help when source queries are slow, but performance depends on data volume, workbook design, calculations, filters, infrastructure, and the source. Test the actual use case.

FAQ: Does a live connection always show real-time data?

No. It queries the database for the view, but the result depends on how current the underlying source is and how the connection and query behave. “Real-time” needs a defined latency expectation.

FAQ: Can I change a published source from live to extract?

Often, Tableau provides a way to change the type, but the current web documentation lists environment and credential limitations. Confirm the route for your Tableau Cloud or Server configuration before promising a change.

Continue the data-connection practice

Use the Tableau certification complete guide, then practice LOD calculations and Tableau chart types. The Tableau study guide connects this decision to the wider skill path. For a second opinion on a source, refresh, or dashboard scenario, use the SimpuTech Tableau study coach.