How to Create an Expense Report in Looker with AI
Building an expense report in Looker allows you to move beyond static spreadsheets and get a real-time view of your company's spending. This article will show you how to set up your data, build a report the traditional way, and then explore how new AI-powered features can dramatically simplify the entire process.
Why Use Looker for Expense Reporting?
Traditionally, expense reports live in spreadsheets. A finance manager exports data from QuickBooks, a credit card statement, or an expense app like Expensify. They then spend hours cleaning it up, creating pivot tables, and formatting charts before emailing a static PDF report to stakeholders. The process is manual, slow, and prone to errors. By the time the report is ready, the data is already outdated.
Moving this process into a business intelligence tool like Looker solves these core problems. Here are the main advantages:
Centralized Data: Looker connects directly to your data sources, whether that's an accounting platform, a SaaS tool, or a cloud data warehouse where all your financial data lives. This creates a single source of truth, eliminating confusion from multiple spreadsheet versions.
Real-Time Insights: Because Looker queries your data directly, your reports are always live. If a new expense is logged, it can appear on your dashboard instantly. This empowers managers to make decisions based on what's happening now, not what happened last week.
Automation and Scalability: Once you build an expense dashboard, it runs on its own. You don’t have to rebuild it every week or month. It automatically updates, saving your finance team countless hours that can be reallocated to more strategic analysis.
Interactive Exploration: Unlike a static spreadsheet, a Looker dashboard is interactive. Stakeholders can click into charts, apply their own filters, and drill down to see the specific transactions behind a high-level number. This encourages self-service analytics and frees up the data team from responding to endless follow-up questions.
Getting Your Data Ready for Looker
Before you can visualize anything, you need to ensure your data is clean, structured, and available for Looker to access. For an expense report, you'll typically pull data from sources like QuickBooks, Stripe, Expensify, or even simple Google Sheets where teams log their spending.
A well-structured expense dataset should include fields (columns) like:
Transaction ID: A unique identifier for each expense.
Date: The date the expense occurred.
Vendor: Who was paid (e.g., "Amazon Web Services," "Delta Airlines").
Amount: The total cost of the transaction.
Currency: The currency used for the transaction.
Category: Your internal classification for the expense (e.g., "Software," "Travel," "Office Supplies").
Employee Name: The person who incurred the expense.
Department: The team associated with the expense (e.g., "Marketing," "Sales," "Engineering").
Description: A brief note about the expense, if available.
Traditionally, connecting this data to Looker requires defining a data model using LookML, Looker's proprietary modeling language. This is where things can get technical. A data analyst or engineer writes LookML code to tell Looker how tables relate to each other and how to calculate metrics (e.g., defining "Total Spend" as the sum of the "Amount" column). This model is what gives business users the drag-and-drop "Explore" interface to build reports without writing SQL. Getting this foundation right is critical, but it often represents a significant technical hurdle for teams without dedicated data resources.
The Traditional Way: Building a Report Manually in Looker
Once your data model is set up, anyone with permission can build a report using Looker's "Explore" feature. If you wanted to create a simple table showing spend by category for the last 90 days, the manual process would look something like this:
Navigate to an Explore: From the Looker menu, you'd select the relevant "Explore" that your data team created for financial data (e.g., "Expenses").
Select Dimensions and Measures: In the left-hand panel, you'll see a list of available fields. You'd click on the "Dimensions" you want to group by (like Category and Vendor) and the "Measures" you want to calculate (like Total Spend).
Apply Filters: To limit the report to a specific timeframe, you would use the filter section at the top. You’d select the Date field, choose a condition like "is in the last 90 days," and click Run.
Choose a Visualization: Looker will default to a data table. You could then click on the visualization options to change it to a pie chart, a bar chart, or another format that best tells the story. For an expense report, a simple table is often the clearest option.
Run and Save: After each change, you click the "Run" button to see the results. Once you're happy with your report, you can save it as a "Look" for future reference or add it directly to a new or existing dashboard.
This process is powerful, but it comes with a learning curve. You need to know which Explore to use, understand what each Dimension and Measure represents, and be comfortable navigating the interface of filters and visualization settings. For non-technical users, it can feel like trying to find the right combination of buttons and knobs to get the answer they need.
Using AI to Instantly Create Your Looker Expense Report
The biggest shift in data analysis is the move away from clicking menus and toward asking questions in natural language. Instead of manually building a report piece by piece, you can now describe what you want, and an AI will build it for you.
Imagine you have a chat interface connected to your Looker data. Instead of following the five steps above, you could simply type a question (or "prompt") and get the same result in seconds.
Here are a few examples of prompts you could use:
"Show me total expenses by category for last quarter as a table."
"What were our top 10 vendors by spending this year? Make it a bar chart."
"Compare monthly software spending for the engineering vs. marketing departments over the last 6 months."
"Give me a list of all travel expenses over $1,000 for the sales team in July."
When you ask a question like this, the AI parses your request, identifies the key concepts ("total expenses," "category," "last quarter"), maps them to the correct fields in your Looker data model, and generates the final visualization. It handles selecting the dimensions, measures, filters, and chart type for you.
This approach completely changes the user experience. You no longer need to learn the intricacies of the BI tool. If you can ask a clear question, you can get a clear answer. It empowers team members in finance, marketing, or operations to get the data they need without having to wait for a data analyst to build a report for them.
Best Practices for AI-Powered Reporting
Getting the most out of an AI data analyst requires a slightly different way of thinking. Here’s how to approach it:
1. Start Broad, Then Go Deep
Think of it as a conversation. Start with a high-level question and use the results to inform your next one. For example:
Initial Prompt: "What was our total company spend last month?"
The AI returns a single number: $85,400.
Follow-up Prompt: "Break that down by department."
The AI generates a bar chart showing spending for Sales, Marketing, and Engineering. You notice Marketing is unusually high.
Deeper Dive: "Show me the marketing department's spending by category for last month."
You see that the 'Paid Advertising' sub-category was responsible for most of the cost.
This iterative process helps you quickly uncover the "why" behind a number and not just the "what."
2. Use Clear, Everyday Language
You don't need to know the official technical name for a field in your database. The best AI tools are trained to understand business vocabulary and context. You can say "customer conversions" instead of "sum_of_converted_leads" or "travel costs" instead of "T&E_spend_amount." Write your prompt as if you were asking a colleague for the information.
3. Turn Q&A into a Lasting Asset
The goal isn't just to answer one-off questions, it's to automate the entire reporting process. Once you have used the AI to create the set of charts and tables for your ideal expense report, ask it to combine them into a single view.
For example: "Create a new dashboard called 'Monthly Expense Overview' with the chart showing spend by department and the table of top vendors."
The AI can then build a full dashboard that will automatically refresh with live data. You’ve now gone from a simple question to a permanent, automated report in just a few minutes.
Final Thoughts
Using a tool like Looker for expense reporting provides a powerful, real-time alternative to manual spreadsheets. By centralizing your data, you create a system that is transparent, automated, and ready for deeper analysis. The addition of AI tools on top of this foundation makes generating those reports easier than ever before, turning a complex technical task into a simple conversation.
At Graphed , we’ve built the entire experience around this conversational approach. Our platform connects directly to your data sources — like QuickBooks, Shopify, Google Ads, and more — and allows you to create dashboards and reports by simply describing what you want in plain English. We want to help you skip the technical hurdles of setting up a data model so you can go straight from asking questions to getting answers, turning hours of reporting work into quick conversations.