How to Create a Summary Report in Looker with AI

Creating a summary report can feel like a hunt for the perfect combination of metrics and dimensions. You know the insights are in your data, but clicking through menus to find them can be a grind. This article will show you how to skip the manual work and create powerful summary reports in Looker just by asking for them in plain English.

What Exactly Is a Summary Report?

Think of a summary report as the highlight reel of your business data. It’s not meant to show every single data point, instead, its job is to provide a high-level overview of an area’s performance. Its goal is to give business leaders and stakeholders a quick, clear answer to the question, “How are things going?”

These reports distill large, complex datasets into a few key performance indicators (KPIs) and critical takeaways. They are essential for:

Quick Decision-Making: Instead of getting bogged down in raw numbers, you get immediate insights to act on. For example, a weekly sales summary might show you which region is underperforming, prompting a discussion with that team.
Better Communication: It's much easier to share a clean visualization summarizing campaign performance with your boss than it is to send them a cluttered spreadsheet with hundreds of rows.
Spotting Trends: By looking at metrics over time (e.g., monthly organic traffic vs. paid traffic), you can easily spot upward or downward trends without having to manually plot numbers.

A few common examples include a monthly marketing performance overview, a quarterly sales pipeline report, or a daily summary of website conversions. Each one takes a massive amount of underlying data and presents it in a simple, digestible format.

The Old Way vs. The AI-Powered Way in Looker

Traditionally, building any report in Looker, including a summary, followed a very manual, structured process. You’d start in an "Explore," which is Looker's environment for building queries. From there, you would:

Navigate sidebars to select specific Dimensions (the "what," like Campaign Name or Country).
Select specific Measures (the "how much," like Total Spend or Number of Users).
Apply filters to narrow down the data (e.g., setting a specific date range).
Run the query to see the resulting data table.
Choose a visualization type (like a bar chart or single value) and configure its settings.
Finally, save your creation as a "Look" or add it to a broader dashboard.

While powerful, this process requires that you already know exactly what dimensions and measures you need and where to find them. It has a significant learning curve and can be time-consuming, even for experienced users. A single report could take 30 minutes to build and refine.

With Looker's integrated generative AI, this entire workflow gets turned on its head. Instead of being a builder who clicks through menus, you become a manager who gives instructions. The process is no longer about finding the right fields but about asking the right questions. You simply describe the summary report you want, and Looker builds it for you.

How to Create a Summary Report with Looker's AI: A Step-by-Step Guide

Let's walk through a practical example. Imagine you're a marketing manager, and you need a quick summary of your campaign performance from the last 90 days to see which channels are delivering the best return on investment.

Step 1: Start a Conversation with Your Data

Your journey begins in an Explore or a dashboard where Looker’s generative AI features are enabled. You’ll typically see a chat or prompt interface that invites you to ask a question. This is your command center for creating the report.

Step 2: Write a Clear and Specific Prompt

This is the most crucial step. The quality of your output depends entirely on the quality of your input. Avoid vague requests. Instead, be as specific as you can about what you want to see.

Vague Prompt: "Show me marketing data."
Good Prompt: "What was my ad spend, revenue, and ROAS by channel for the last 90 days?"
Even Better Prompt: "Show me a table with total ad spend, total revenue, and ROAS broken down by marketing channel for the last 90 days. Please sort it by ROAS from highest to lowest."

Notice how the better prompt includes the core metrics (ad spend, revenue, ROAS), the dimension (marketing channel), and the timeframe (last 90 days).

Step 3: Generate the Initial Report

After you enter your prompt, you’ll see the AI translate your natural language request into a query and generate a data table or visualization. In our example, it would likely produce a simple table listing each channel (e.g., Google Ads, Facebook Ads, LinkedIn Ads) with columns for Spend, Revenue, and ROAS, sorted as requested.

You’ve just created your first draft of the summary report in a few seconds instead of several minutes of clicking.

Step 4: Refine the Report with Follow-Up Questions

The real power of this conversational approach lies in iteration. Your first result might be close, but it’s rarely perfect. Now, you can build on it by asking follow-up questions.

Changing the Visualization: "Show this as a horizontal bar chart."
Drilling Down: "Now filter to only show paid search and paid social channels."
Adding a Calculation: "Can you also add a column for conversion rate?"
Adjusting the Timeframe: "Compare this to the previous 90 days."

Each follow-up prompt modifies the existing report, allowing you to slice, dice, and explore your data on the fly. You're having a conversation, getting closer to the perfect summary with each request.

Step 5: Save and Share Your Summary

Once you’re satisfied with the result - let's say you have a clean bar chart showing ROAS by channel - you can save it. Typically, you can add this new visualization directly to an existing Looker dashboard or save it as a standalone Look. This makes it a reusable asset that will automatically update with fresh data, ensuring your summary report is always current.

Best Practices for Prompting Looker's AI

To get the most out of Looker's AI, keep these simple principles in mind.

1. Use Specific Names for Metrics and Dimensions

Although it feels like magic, the AI is tethered to the underlying data structure (the LookML model). While you don't need to know the database column names, using the common business terms your team uses in Looker will yield the best results. For example, use "Revenue" if that's what your team calls it, not "Sales" or "Income."

2. Always Include a Timeframe

A report without a timeframe is missing critical context. Be explicit about the period you’re interested in, whether it’s "last 7 days," "this quarter," "in January 2024," or "year to date."

3. Start Simple and Build Up

Don't try to craft a perfect, paragraph-long prompt from the start. Begin with a straightforward request to get your core data. Then, use short, iterative follow-up prompts to add layers of complexity, change the visualization, and apply filters. This approach is faster and more effective.

4. Describe the Chart Type You Want

If you have a specific visualization in mind, just ask for it. Phrases like "show this as a line chart," "create a pie chart," or "make it a single numeric value" work perfectly. This saves you from having to adjust settings in the visualization editor manually.

Final Thoughts

Switching from manually building reports to conversationally requesting them makes data analysis significantly faster and more accessible. Using AI inside Looker allows anyone, regardless of technical skill, to get a clear summary of business performance and answer critical questions in seconds, not hours.

At our core, we believe the future of data analytics is conversational. We designed Graphed to connect to all your key marketing and sales tools - like Shopify, Google Ads, Salesforce, and HubSpot - so you can instantly create live dashboards and reports using simple language. Instead of spending half your day jumping between platforms and wrestling with spreadsheets, we give you a central AI data analyst that provides answers, spots trends, and empowers your whole team to make smarter decisions.