How to Set Up Amazon QuickSight Q for Analytics

Amazon QuickSight Q lets anyone on a team ask data questions in plain English and get back a chart or table in seconds, with no SQL required. To enable it, you need a QuickSight Enterprise account, at least one dataset loaded into SPICE, and a Q Topic that defines which columns Q can search. The full setup can be completed inside the AWS console in under 30 minutes.

What You Need Before You Start

QuickSight Q requires a QuickSight Enterprise Edition subscription. Reader pricing starts at $18 per user per month billed annually, and Q is included at no extra charge as of early 2026.

You also need:

• An AWS account with permissions to create QuickSight analyses and datasets
• Data accessible by QuickSight: this can be a CSV uploaded directly, an S3 file, a MySQL or PostgreSQL database, or another supported source. QuickSight connects to over 30 data sources.
• At least one SPICE dataset. SPICE is QuickSight's in-memory engine. Q only works with SPICE; it does not support direct query mode.

If you are starting with a CSV file, the simplest path is to upload it directly inside QuickSight without configuring any external data source.

Step 1: Connect Your Data and Load It Into SPICE

1. Sign in to QuickSight at quicksight.aws.amazon.com and select your AWS region.
2. From the top navigation, click Datasets and then New dataset.
3. Choose your data source type. For a CSV, select Upload a file and drag in your file. For a database, choose the appropriate connector, enter your host and credentials, and click Validate connection.
4. After QuickSight previews the data, click Edit/Preview data to review column names and data types. Rename any columns with unclear abbreviations now. Q uses column names to interpret natural language questions, so "Rev Q4" is harder for Q than "Q4 Revenue".
5. When the preview looks correct, click Save and publish. In the dialog that appears, make sure Import to SPICE for quicker analytics is selected, then click Save and publish again.

QuickSight will import the data. For files under 1 GB, this usually completes in less than two minutes. You will see a green "Dataset saved" confirmation when it is ready.

Step 2: Create a Q Topic

A Q Topic is a configuration that tells QuickSight Q which dataset to search and how to interpret its columns. You can have multiple Topics, one per domain; for example, one for sales data and another for operations data.

1. From the QuickSight home screen, click Q in the left sidebar.
2. Click Create new topic and enter a name. Use a short, descriptive label such as "Sales Performance" or "Monthly Operations".
3. Under Datasets, click Add dataset and select the SPICE dataset you created in Step 1.
4. Click Save.

QuickSight will scan the dataset and auto-detect column types. Review the column list. For each column, you can:

• Mark as a dimension (categorical: product name, region, sales rep) or measure (numeric: revenue, units sold, profit margin)
• Add synonyms so Q understands informal phrasing. If your column is labeled "contract_start_date", add synonyms like "start date", "contract date", or "sign date" so users can ask "show me deals by sign date" without getting a blank result.
• Hide columns that are internal keys or IDs users would never query directly.

These configurations directly improve Q's interpretation accuracy. According to AWS documentation, topics with synonyms configured for key columns return correct first-result answers significantly more often than topics with no synonym setup.

Step 3: Add the Q Search Bar to an Analysis

1. Open an existing analysis or create a new one. Click New analysis and select your dataset.
2. In the analysis editor, click Add in the top toolbar and select Add Q search bar.
3. A search bar panel will appear on the canvas. Resize and position it at the top of the sheet.
4. Click Select topics in the search bar panel and link it to the Q Topic you created.
5. Publish the analysis as a dashboard: click Share and then Publish dashboard.

Once published, anyone with reader access to the dashboard will see the search bar and can type questions directly.

Step 4: Test and Train Q with Sample Questions

Before sharing with your team, test Q yourself by typing questions that match real business queries your team would ask. Good starting questions follow this pattern:

• "What were total sales by region in February?"
• "Which product had the highest return rate last quarter?"
• "Show me monthly active customers for the past six months."

When Q returns a correct answer, click the thumbs-up icon to mark it as verified. Verified answers become reference examples Q uses to calibrate future interpretations. If the answer is wrong, click Edit and adjust the question interpretation manually, then verify the corrected version.

Add five to ten verified questions per topic before rolling it out to your team. AWS reports that topics with verified sample questions resolve ambiguous queries at a substantially higher rate than untrained topics.

Tips for Improving Answer Quality

A few practices that consistently improve Q accuracy in real use:

Use full column names at setup. Column names like "mrr_usd" are harder for Q than "Monthly Recurring Revenue (USD)". You can rename columns in the dataset editor without changing the underlying source.

Add date semantics. In the Topic configuration, set one date column as the "Default date field". This lets Q interpret phrases like "last month" or "year to date" correctly.

Break long CSVs into clean flat tables. Q performs best with wide, denormalized tables rather than deeply relational schemas with many joined IDs. If your data requires multiple joins, create a flat export before importing to SPICE.

Limit topics to one domain each. A single topic covering sales, HR, and inventory creates ambiguity. Separate topics for each domain give Q a smaller, more coherent vocabulary to work with.

What QuickSight Q Cannot Do

Q is a strong tool for exploratory questions, but it has clear limits that business users should understand before deploying it widely.

Q cannot write SQL for you or export query logic to another system. It cannot answer questions that require real-time data, because SPICE is a cached copy of your data refreshed on a schedule you configure, not instantaneously. It does not support complex multi-step reasoning, such as "If we increase prices by 10%, what happens to volume based on last year's elasticity?" For those questions, a separate modeling tool or a data analyst is still required.

Q also does not support direct database query mode. If your dataset is too large for SPICE or your organization requires live queries, Q is not currently an option without architectural changes.

Sharing with Your Team

Once the dashboard is published, share it from the QuickSight console under Manage QuickSight then Manage users. Assign reader accounts to team members who need access. Readers can use the Q search bar on any published dashboard linked to an active topic. If you want someone to ask questions but not modify dashboards, the reader role is the correct permission level.

For teams that already have an upload and prep step in their data workflow, tools like VSLZ AI can handle the initial data cleaning and exploratory analysis before a dataset moves into QuickSight for ongoing dashboarding.

Summary

Setting up Amazon QuickSight Q involves four steps: loading data into SPICE, creating a Q Topic with well-labeled columns and synonyms, adding the Q search bar to a dashboard, and training the topic with a small set of verified sample questions. The console-only path described here requires no SQL or code and can be completed in under 30 minutes for a clean business dataset.