How to Create a SaaS Dashboard in Looker with AI

A great SaaS dashboard is the command center for your business, giving you a clear, at-a-glance view of essential metrics like MRR, churn, and customer lifetime value. This article walks you through how to build a SaaS dashboard in Looker (now part of Google Cloud) and explains how new AI features can dramatically speed up the process.

What Exactly is a SaaS Dashboard (and an Even Better Question...Why do You Need It?)

Think of a SaaS dashboard as a single screen that tells you the story of your business's health in real-time. Instead of logging into Stripe, then HubSpot, then your product database to piece things together, a dashboard consolidates your most important Key Performance Indicators (KPIs) into one place. This unified view helps you spot trends, identify problems before they spiral, and make smart, data-driven decisions about where to focus your resources.

Without a centralized dashboard, you're flying blind. You might feel like things are going well, but you won't know for sure until you've manually pulled three different reports, stitched them together in a spreadsheet, and spent half a day just getting the numbers. By then, the opportunity to act has often passed.

Key Metrics Every SaaS Dashboard Should Include

While dashboards are customizable, nearly all top-performing SaaS businesses track the following core metrics. This isn't an exhaustive list, but it's the foundation of a dashboard that truly informs business strategy.

• Monthly Recurring Revenue (MRR): The predictable revenue your business earns from active subscriptions each month. You should break this down further into:

  • New MRR: Revenue from new customers.

  • Expansion MRR: Additional revenue from existing customers (upgrades, cross-sells).

  • Churn MRR: Lost revenue from customers who cancel or downgrade.

  • Net New MRR: (New MRR + Expansion MRR) - Churn MRR. This is your true growth metric.

• Customer Lifetime Value (LTV): The total revenue you can expect to earn from a single customer account over their entire lifetime with your product.

• Customer Acquisition Cost (CAC): The total cost of sales and marketing to acquire a new customer.

• LTV:CAC Ratio: This critical ratio tells you your return on investment for customer acquisition. A healthy ratio is typically considered 3:1 or higher - for every dollar you spend getting a customer, you get three dollars back.

• Customer Churn Rate: The percentage of customers who cancel their subscriptions in a given period. It's crucial to also track Revenue Churn, which is the percentage of revenue lost from those cancellations.

• Average Revenue Per Account (ARPA): The average MRR across all of your active customers. Tracking this helps you understand the impact of pricing changes and customer 'fit' over time.

Preparing Your Data for Your Looker Dashboard

Before you can build anything in Looker, you need to get your data ready. For most SaaS companies, this is the most challenging step because sensitive information lives in different applications that don't talk to each other. Your payment data is in Stripe, your customer interactions are in HubSpot or Salesforce, and your product usage data is in a production database.

Looker doesn't store your data, it queries it wherever it lives. This means you need a centralized, analysis-ready location for Looker to connect to. This is typically a cloud data warehouse.

Step 1: Get Everything Into a Data Warehouse

A data warehouse is a central repository designed for fast and complex business analysis. Popular choices include Google BigQuery, Snowflake, and Amazon Redshift. To get your data from sources like Stripe, Salesforce, or Facebook Ads into your warehouse, you'll need an ETL (Extract, Transfer, Load) or ELT tool like Fivetran or Stitch. These tools act as pipelines, automatically pulling data from your SaaS apps and loading it into your data warehouse on a schedule.

Step 2: Understand LookML, Looker's Secret Sauce

Once Looker is connected to your data warehouse, you need to define your business logic using Looker's modeling language, LookML. This is a powerful feature but also introduces a significant learning curve.

LookML is where data analysts define dimensions (like user_signup_date or customer_plan_type), measures (like sum_of_revenue or count_of_active_users), and the relationships between your data tables. By creating a robust LookML model, you create a trusted, 'single source of truth' that allows non-technical team members to explore data and build reports without writing a single line of SQL. The tradeoff is that setting it up requires dedicated developer time and ongoing maintenance.

How to Build Your SaaS Dashboard in Looker (The Old-School Way)

Assuming your LookML model is in a good place, building a dashboard involves several manual steps - pulling data, creating visualizations, and arranging them logically.

1. Start an "Explore": The Explore interface is where you build queries. Your data team will have prepared a set of data models (Explores) based on the LookML you can use. You'd pick one, like "Subscriptions," to get started.

2. Select Dimensions & Measures: In the left-hand panel, you’ll see all available fields defined in your LookML. To see MRR over time, you would select a date dimension (e.g., "Subscription Created Month") and your MRR measure (e.g., "Total MRR").

3. Run the Query: Click "Run." Looker will automatically write the SQL in the background, query your data warehouse, and return the results in a table.

4. Choose a Visualization: Above the results table, select a visualization type. For MRR over time, a line chart is a good choice. You can customize colors, labels, and axes. A chart or visualization in Looker is called a "Look."

5. Save your Look to a Dashboard: Once you're happy with your visualization, you can save it directly to a new or existing dashboard.

6. Repeat and Arrange: Repeat this process for every metric - churn rate, LTV, CAC, etc. - until you have all the necessary components. Then, you can drag and drop the tiles on your dashboard and resize them to create a clean, logical layout. You’d probably put your most important KPIs as large numbers at the top.

7. Add Filters: Finally, add dashboard-level filters. A date range filter is essential, but you might also want to filter by subscription plan, customer segment, or acquisition channel to enable deeper analysis.

This process works, but it can be time-consuming. Building a comprehensive dashboard from scratch often requires dozens of individual queries and a fair amount of clicking, configuring, and tweaking.

Introducing Looker's New AI Co-Pilot, Duet AI

The traditional method for building dashboards places a heavy burden on the user to know exactly what they want and how to get it through the UI. Realizing this, Google has embedded its Duet AI directly into Looker to serve as an intelligent data assistant, dramatically simplifying the report-building process.

Instead of manually clicking through menus and dragging fields, you can use conversational language to get what you need. This makes data analysis accessible to a much wider range of team members who aren't Looker power users.

How to Use AI To Build and Explore Looker Dashboards

With Duet AI, your starting point is no longer a blank "Explore" menu - it's a chat box. Here’s what the new, AI-powered workflow looks like:

1. Ask for a Visualization in Plain English: You can start with a simple prompt. Instead of hunting for the MRR measure and date dimension, you just ask:

   show me total monthly recurring revenue as a line chart by month

   Duet AI analyzes your prompt, identifies the relevant dimensions and measures from your LookML model, and automatically generates the Explore for you, complete with data and the correct visualization.

2. Follow-up and Refine Your Questions: This is where the real power comes in. Analysis is never a one-step process, it's a conversation. As you see your initial chart, new questions always pop up immediately. With AI, you can keep the conversation going and drill down deeper into your data using natural language questions as prompts, for example:

   now segment this by plan type

   Duet AI remembers the context of your previous query and adds the "Plan Type" dimension, automatically creating a stacked bar chart or a multi-line chart to show the comparison.

   what were the top 5 countries by LTV last quarter?

   The system builds the query, filters it by last quarter, ranks the output, and returns either a bar chart or a table answering your question in seconds.

3. Go From Question to Dashboard in Minutes: The entire workflow is conversational. You can go from a blank slate to a fully fleshed-out visualization for your SaaS dashboard in under a minute. Just like before, once a chat generates a chart you're happy with, you can click to add it to your dashboard. This not only saves a considerable amount of time but also frees team members up to think about what the data is actually trying to show them rather than focusing on how to configure dropdown menus and create a specific visualization manually.

Limitations to Keep in Mind

While incredible, Duet AI still relies completely on the quality of the underlying LookML model. If the data team hasn't precisely defined a metric like "Revenue Churn" or properly joined your subscriptions table with your customers table in LookML, the AI won't be able to answer questions about it. The AI is a brilliant interface for a well-structured data model, not a replacement for one.

Final Thoughts

Building a powerful SaaS dashboard is essential for monitoring the health and growth of your business. While tools like Looker offer incredible analytical depth, the process has traditionally been gated by the technical complexity of setting up data pipelines and defining a robust LookML model, and dashboards usually must be built manually by someone with those technical skillsets.

As we've seen, building dashboards and reports manually is time-consuming and tedious, and businesses have had to handle business intelligence this way across a lot of different systems. So, Graphed was created to reduce the burden of building custom reports and dashboards completely. Connecting directly to data sources like Google Analytics, HubSpot, Salesforce, or Stripe in minutes, making them "AI ready" instantly, we've designed Graphed to allow for the seamless and effortless process of building real-time, interactive dashboards using natural language prompts, without ever writing a line of code or setting up a complex data model or warehouse. AI functionality like in Duet is powerful, but they only offer a quicker and better user experience within an extremely technical ecosystem, creating a new tool just as, if not more, complicated than every solution that has preceded it. At Graphed, you're the master. No need for data analysts. All that is required of you to start creating dashboards is to simply ask, the rest is created instantly using AI. That's the way it should work.