Data Studio Conversational Analytics, set up on data you can trust

What is Data Studio Conversational Analytics?

It is a way to query data in Data Studio by asking questions instead of building charts first. The questions are answered by data agents, which are built in BigQuery and then published to Data Studio for people to use.

Google made it available to all users on April 16, 2026, around the time Looker Studio was renamed back to Data Studio, and it became generally available on July 30, 2026. For a business, the important part is where the agent gets its data: from BigQuery. Companies whose reporting still runs entirely on direct connectors and Google Sheets need a BigQuery layer before the feature is useful to them.

It does not replace dashboards. A well-designed report still shows the same agreed numbers to everyone, every week. Conversational Analytics is better suited to the follow-up questions a dashboard prompts, the ones that would otherwise become a request in an analyst's queue.

How do data agents in BigQuery work with Data Studio?

An agent is set up in BigQuery against specific tables, then published to Data Studio. The agent is the bridge between a question someone types and the tables that hold the answer.

That makes BigQuery the place where most of the real work happens. The agent needs to know which tables to use, what each column means and how metrics are defined. If revenue lives in three tables with three slightly different definitions, the agent has no way to know which one your finance team means.

What has to be true about your data for answers to be right?

The data model has to be clean, documented and unambiguous. This is the part most teams underestimate, and it is the part we spend most of our time on.

Before we connect an agent, we check the model against a short list. Where it falls short, we fix the tables first.

• One table per subject with a clear grain, such as one row per order line or per day per campaign.
• Readable table and column names, with descriptions written in BigQuery for every field.
• One agreed definition of each key metric, stored once rather than recalculated in many places.
• Consistent dimension values: one spelling per channel, region and product category.
• Dates, currencies and time zones handled the same way across every table.
• Access rules set in BigQuery, so the agent only reaches data the person may see.

Do you need Data Studio Pro for Conversational Analytics?

Gemini in Data Studio is part of Data Studio Pro, which costs $9 per user per project per month and only needs licences for creators, editors and managers. Viewers do not need a Pro licence.

Pro also brings content owned by the company through a Google Cloud project, team workspaces, audit logging and data residency options, which matter when an assistant can reach business data. We look at who on your team will build and manage agents and reports, and size licences to that, not to every viewer.

What does our setup service include?

We take a company from existing reports to a working agent its team can rely on. The work starts with the data, not the feature.

First we review which questions people actually ask, usually from the requests your analysts field every week. Then we build or tidy the BigQuery tables that hold those answers, write column descriptions and metric definitions, set access, create the data agent and publish it to Data Studio. Before anyone else uses it, we test it with a set of real questions whose correct answers we already know from existing reports, and fix the model wherever answers differ.

After launch, the test set stays useful. Whenever a table changes, a new source is added or a metric definition is updated, we rerun it, so a change in the warehouse never quietly changes the answers people get. Clients on our managed plan have this done as part of the monthly cycle.

Who is Data Studio Conversational Analytics a good fit for?

It suits teams that already have, or are ready to build, a modeled BigQuery layer, and whose people ask many small questions that currently land on an analyst's desk. Typical examples are a sales leader asking about one region, a marketer checking a single campaign, or someone in finance asking about one customer's orders. Each question is small, but together they fill an analyst's week.

It is a poor fit, for now, if your data lives in a dozen disconnected spreadsheets or if nobody agrees on metric definitions. In that case the warehouse comes first, and a dashboard may answer most questions on its own. We will tell you which situation you are in during the free review.