How to Use Gemini in Looker

Gemini in Looker revolutionizes how you interact with your business data, allowing you to ask questions in plain English instead of building reports manually. This article walks you through exactly what Gemini is, how to enable it, and how to use natural language to create reports, build dashboards, and uncover insights faster than ever before.

Graphed

Build AI Agents for Marketing

Build virtual employees that run your go to market. Connect your data sources, deploy autonomous agents, and grow your company.

Build Your First Agent

What is Gemini in Looker?

Gemini in Looker is the integration of Google's powerful AI model directly into the Looker business intelligence platform. Think of it as a conversational data analyst built right into your dashboards. Instead of clicking through menus, selecting dimensions and measures, and configuring chart settings in the "Explore" interface, you can now simply chat with your data.

You can ask questions conversationally, such as "What were my total sales last quarter by product category?" and Gemini will interpret your request, query the underlying data model, and generate an appropriate chart or visualization for you automatically. It's designed to lower the barrier to data analysis, making it more accessible to non-technical users and more efficient for experienced analysts.

Key Features Included with the Gemini Integration:

• Conversational Analytics: Get instant answers and create reports by chatting in natural language.
• Automatic Chart Generation: Gemini intelligently selects the best visualization type (like a bar chart, line chart, or table) based on your question.
• Suggested Questions: The AI provides proactive suggestions for follow-up questions to help you dig deeper into your data.
• Calculations on the Fly: You can ask Gemini to create simple calculations without needing to create new LookML fields, such as "What's the ratio of new customers to returning customers?"

Getting Started: Enabling and Using Gemini

Before you can start chatting with your data, Gemini needs to be enabled for your Looker instance. This is a quick process handled by a Looker administrator and requires a specific new Looker plan.

AI Agents for Marketing

Crash Course

Graphed

Free PDF · the crash course

AI Agents for Marketing Crash Course

Learn how to deploy AI marketing agents across your go-to-market — the best tools, prompts, and workflows to turn your data into autonomous execution without writing code.

Download the PDF

For Looker Administrators: How to Enable Gemini

To turn on Gemini features, you'll first need to be on a Looker Google Cloud core edition that supports it (like Enterprise or Embed) and purchase the "Gemini for BI" add-on. Once that's sorted, enabling it is straightforward.

1. Navigate to the Admin panel in your Looker instance.
2. Under the "Platform" section in the left-hand menu, find and click on Gemini.
3. Toggle the feature on. You might see several related settings allowing you to control who has access to the feature or other GenAI capabilities.

It's important to note that Gemini’s performance is directly tied to the quality of your underlying LookML model. A well-structured, clearly-named model will produce much more accurate and reliable results. Make sure your fields, views, and Explores use intuitive, human-readable names that your business users will naturally use when asking questions.

For Business Users: Where to Find and Use Gemini

Once your administrator enables it, Gemini will appear in a few key places across Looker. You’ll primarily use it from within an Explore or a Dashboard.

• In an Explore: You'll see a new chat-style interface where you can type your question to generate a report.
• On a Dashboard: A Gemini button allows you to "ask about this dashboard," giving you the ability to query the combined data from the dashboard's tiles without having to hunt down the source reports.

Simply click the prompt box, type your query, and hit enter. Looker will process your request and generate a visualization in a few seconds.

How to Create Reports with Natural Language

Using Gemini is as simple as asking a question. The key is to be clear and specific in what you're asking for. Let's walk through some examples, from basic requests to more complex queries.

Graphed

Build AI Agents for Marketing

Build virtual employees that run your go to market. Connect your data sources, deploy autonomous agents, and grow your company.

Build Your First Agent

Asking Your First Question: A Simple Example

Start with a straightforward, high-level question. Gemini works best when you provide three key ingredients: a metric, a dimension, and a time frame.

For example, if you're analyzing sales data, you could ask:

Show me total revenue by month for the last 6 months

Gemini will parse this sentence, identifying "total revenue" as the metric, "month" as the dimension for grouping, and "last 6 months" as the time filter. It will then automatically query your data and generate a line or column chart displaying the results. You just went from a question to an insight without clicking a single filter or field.

Building More Complex Visualizations

The real power of conversational AI is refining and layering your analysis. Once Gemini generates your initial chart, you can ask follow-up questions to dig deeper or change the visualization entirely.

Following the example above, you could ask:

• To add more detail: Now break that down by product category
• To change the chart type: Show this as a stacked column chart
• To dig into anomalies: Which category had the biggest drop in May?

You can also ask more complex questions from the start by combining elements. For example:

Compare marketing qualified leads vs sales qualified leads from our HubSpot data over the last quarter and show it in a bar chart

Gemini is designed to understand this type of multi-part query, fetching both metrics, applying the date filter, and creating the specific chart you requested.

Tips for Writing Effective Prompts

While the AI is smart, clear instructions lead to better results. Here are some tips for phrasing your questions to get the most accurate answers from Gemini:

• Be Specific: Instead of asking for "best-selling items," ask for "top 10 products by units sold" or "top 5 services by revenue." Vague terms like "best" are subjective and harder for the AI to interpret.
• Clearly State Time Frames: Always specify the period you're interested in. Use phrases like "last 30 days," "this quarter," "year-to-date," or "in 2023."
• Specify the Chart Type: If you have a specific visual in mind, just ask for it. Append "...as a line chart," "...in a pie chart," or "...as a table" to the end of your prompt.
• Use Your Model's Language: While Gemini can handle synonyms, it works best if you use names that correspond to the fields and dimensions in your LookML model. If your model calls it "User Signup Date," using that phrase is more reliable than "when did people join."

AI Agents for Marketing

Crash Course

Graphed

Free PDF · the crash course

AI Agents for Marketing Crash Course

Learn how to deploy AI marketing agents across your go-to-market — the best tools, prompts, and workflows to turn your data into autonomous execution without writing code.

Download the PDF

What Makes Gemini genuinely Useful for Your Team?

The addition of Gemini transforms Looker from a tool for data specialists into an information hub for everyone in the organization. The benefits are clear and immediate.

Primary Benefits

• Data Democratization: Team members in marketing, sales, product, or operations can now self-serve their own data requests. They no longer have to wait for an analyst to build a custom report, which removes bottlenecks and speeds up decision-making.
• Unmatched Efficiency: Even for seasoned Looker experts, typing a question is often much faster than clicking through the Explore UI. This frees up data analysts from handling routine, repetitive reporting requests so they can focus on more strategic, in-depth analysis.
• Reduced Learning Curve: Onboarding new team members to Looker becomes significantly easier. As long as they understand the business and know what questions to ask, they can get value from Day One without needing hours of training on LookML or the Explore interface.

Important Limitations to Keep in Mind

While powerful, generative AI for data analysis is not a magic bullet. It’s important to understand its current limitations:

• It's Only as Smart as Your Data Model: The principle of "garbage in, garbage out" has never been more relevant. If your LookML model is confusing, poorly structured, or uses ambiguous naming conventions, Gemini will struggle to provide accurate answers.
• Not a Replacement for Data Analysts: For nuanced, multi-layered investigations that require deep business context, the expertise of a human analyst remains irreplaceable. Gemini is an incredibly powerful assistant, but it can't (yet) replicate the strategic reasoning of a seasoned data professional.
• Can Misinterpret Ambiguity: Vague or poorly worded questions can sometimes be misinterpreted. Users should always review the generated report - especially the filters and fields used - to confirm that the AI understood their request correctly.

Final Thoughts

Gemini in Looker powerfully demonstrates the future of business intelligence, successfully bridging the gap between sophisticated data tools and the people who need data-driven answers. By enabling anyone on your team to generate reports and uncover insights using simple, conversational language, it makes data truly accessible for everyone.

While tools like Gemini mark a major step forward within a single ecosystem like Google Cloud, we know that most businesses' data is far from centralized. Your marketing, sales, and transaction data is likely scattered across a dozen different apps like Shopify, Salesforce, Facebook Ads, and Google Analytics. That's why we built Graphed - to instantly connect all your tools and give you an AI data analyst that can answer questions about your entire business, not just one part of it. You just connect your apps and get immediate, unified insights using natural language, all without needing to hire a data team or write a single line of code.