Graphed LogoGraphed
Login

How to Create a Weekly Report in Looker with AI

Cody Schneider•

Getting a weekly summary of your business performance shouldn't feel like wrestling with your data tools. While Looker is incredibly powerful, building and maintaining reports can often feel rigid and time-consuming, especially when you have follow-up questions. This guide will walk you through setting up a standard weekly report in Looker and then show you how a modern, AI-first approach completely transforms this process from a static chore into a dynamic conversation with your data.

What Makes a Great Weekly Report?

Before jumping into the technical steps, let’s agree on what a truly useful weekly report looks like. A report isn't just a list of numbers, it's a tool for making better decisions. A great one is always:

  • Focused on Key Metrics (KPIs): It tracks the handful of metrics that truly matter for your business goals. For a sales team, this might be Leads, Conversion Rate, and MRR. For a marketing team, it could be Ad Spend, ROAS, and MQLs. Trying to track everything means you end up tracking nothing.

  • Contextual: A number like "1,000 new users" is meaningless by itself. Is that good? Bad? Compared to what? A great report shows this metric in the context of the previous week, the same time last month, or your weekly goal, so you can immediately spot trends.

  • Visual and Clear: Huge tables of data are overwhelming. The best reports use simple, clear visualizations - like line charts for trends over time or bar charts for comparisons - to tell a story at a glance.

  • Actionable: Ultimately, the report should spark a question or suggest an action. It should make it obvious where things are going well and where you might need to dig in deeper.

The Traditional Path: Building Your Report in Looker

In a typical workflow, creating a report in Looker involves a series of structured steps, often managed by a data analyst or someone with technical expertise. Looker’s power comes from its modeling layer, LookML, where developers define business logic, dimensions, and measures. This ensures consistency across all reports, but it also creates a barrier for non-technical users.

The standard process usually looks like this:

  1. A business stakeholder (like a marketing manager) requests a new report.

  2. A data analyst checks if the required data is available in the LookML model. If not, they have to update the model first.

  3. The analyst builds the individual charts and tables (called “Looks”) and assembles them into a dashboard.

  4. After some back and forth, the final dashboard is approved and scheduled for weekly delivery.

This process is reliable but can be slow. If you have an urgent follow-up question that wasn’t built into the original report, you're back in the queue waiting for an analyst to help - a common bottleneck that stifles curiosity and quick decision-making.

Step-by-Step: Creating and Scheduling a Weekly Report in Looker

If you have the appropriate permissions and familiarity with Looker, you can set up a weekly report yourself. Here is a simplified walkthrough of the process.

Step 1: Start with an "Explore"

Think of an Explore as your starting point for asking questions. It’s a pre-built view of your data created from your company’s LookML model. For example, you might have an "Orders" Explore or a "Website Traffic" Explore.

Navigate to the "Explore" section and select the data source relevant to your report, for example, Salesforce Opportunities or Google Analytics 4 Data.

Step 2: Build Your First "Look"

A "Look" is a single visualization, like a table or a chart. Let's create one showing weekly revenue.

  1. Select Dimensions and Measures: In the left-hand pane of the Explore interface, you’ll see a list of fields. Dimensions are your attributes (like date, campaign name, or country). Measures are your quantifiable numbers (like total revenue, user count, or conversion rate).

  2. Let’s say you want to see revenue by week. You would select a date dimension (e.g., "Order Created Week") and a revenue measure (e.g., "Sum of Order Total"). Looker will automatically generate a data table.

  3. Add a Filter for Timeframe: To make it a weekly report on recent data, add a filter. Find your date field in the "Filters" section and set it to something like "is in the last 12 weeks."

Step 3: Choose Your Visualization

Once you have your data table, click on the "Visualization" tab. Looker offers many chart types. For weekly trends, a Line Chart is usually best. Customize the colors and labels as needed to make it easy to read.

Step 4: Save the Look to a Dashboard

Happy with your chart? It’s time to save it to a dashboard.

  • Click the gear icon in the top right and select "Save."

  • You can "Save as a New Look" and then choose "to a new dashboard." Give your new dashboard a descriptive name, like "Weekly Marketing Performance."

Repeat steps 2-4 for all the metrics you want in your report. You might create Looks for ad spend, website sessions, conversion rates, and new leads, adding each one to your dashboard.

Step 5: Schedule Your Report for Delivery

This is where you make it an automated weekly report.

  1. On your completed dashboard, click the gear icon in the upper right corner and select "Schedule delivery."

  2. In the delivery options, you can configure the following:

    • Recurrence: Set this to "Weekly" and choose the day and time you want it sent (e.g., every Monday at 9 AM).

    • Destination: Choose where you want it sent. Common options include Email, Slack, or webhook.

    • Format: You can send it as a PDF or a visualization attached to the email.

  3. Click "Save" and you're done! Your report will now arrive automatically every week.

Where AI Really Changes the Game

The process above works perfectly - until something changes or you have a spontaneous question. On Monday morning, your report arrives, and you see that traffic from the US is down 20%. Why?

With the traditional report, your investigation stops. You have to email your data analyst, who then has to go back, modify the query, create a new chart, and send it back to you. By the time you get the answer, it might be Tuesday afternoon. Half the week is wasted.

This is where an AI-powered approach fundamentally differs. It's about turning reporting from a one-way broadcast into an interactive conversation.

AI in Looker vs. AI-Native Tools

Google is integrating AI capabilities like Gemini and the Looker AI Assistant directly into the Looker platform. These features aim to help users create calculations, generate reports from prompts, and generally lower the technical barrier.

While helpful, these are AI features added to a traditional BI tool that a company's data engineers already set up. You still rely on the underlying, manually-created LookML models, and the core workflow remains largely the same.

An alternative is an AI-native tool that is built from the ground up for natural language interaction. Instead of having to learn a software's interface of clicks and menus, you can simply ask for what you want in plain English. The AI understands the underlying structure of your connected data sources automatically, so you don't need a manually built semantic layer like LookML.

This is the difference between an AI assistant inside a traditional tool and a tool that is, at its core, an AI analyst itself. With an AI-native approach, the workflow for getting that weekly report and answering your follow-up can be accomplished almost instantaneously.

  • The Prompt: Instead of clicking through explores and views, you simply type: "Create a weekly dashboard showing revenue from Shopify, website sessions from Google Analytics, and ROAS from Facebook Ads for the last 90 days."

  • The Instant Dashboard: The AI instantly generates the dashboard with all three charts, pulling live data from the connected platforms. No building individual Looks is required.

When you see the traffic drop, your next step isn't to file a ticket. You just ask a follow-up question right there:

“Why did US traffic drop last week? Break it down by device."

A new chart appears in seconds showing that mobile traffic dropped significantly while desktop traffic remained stable. This leads to your next question: "Compare US mobile traffic from our top 5 referrer sites this week vs. last week." And so on. You can go from high-level observation to root cause in minutes, not days.

Final Thoughts

Building and scheduling a weekly report in Looker is a fantastic way to keep your team informed and aligned. By following the steps to build and automate your dashboard, you establish a reliable rhythm for monitoring your business. However, this traditional approach often stops at just delivering data, leaving the deeper insights hidden behind a wall of technical complexity and time delays.

At Graphed , we’ve shifted this entire paradigm. We built an AI data analyst that allows you to connect all of your marketing and sales data sources in seconds and create live, interactive dashboards using simple, conversational language. Instead of spending hours clicking around or waiting on your data team, you can get answers and go deeper in seconds, making your weekly reporting process not just a check-in, but the starting point for immediate, data-driven action that was previously impossible.