How to Optimize Performance in Power BI

Nothing kills the momentum of a data deep-dive faster than a perpetually loading Power BI report. You click a slicer, and the little dots start swirling for what feels like an eternity. This guide cuts through the complexity and gives you a clear path to speeding up your dashboards. We'll cover the most effective techniques, starting from the foundation of your data model and moving all the way to writing faster DAX.

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

It All Starts with the Data Model

If your Power BI report is slow, the most likely culprit is a bloated or poorly designed data model. A messy model forces the DAX engine to work much harder than necessary to answer your questions. Getting the model right is the single most important step you can take for better performance.

Embrace the Star Schema

Many people starting with Power BI import a single, massive, flat table that looks like an Excel spreadsheet, with dozens or even hundreds of columns. This is the least efficient way to build a report. The industry-standard best practice is to structure your data in a star schema.

A star schema consists of two types of tables:

• Fact Tables: These contain the numbers and events you want to measure, like sales amounts, transactions, website sessions, or ad clicks. They can have millions of rows but should have very few columns - typically just numeric values and key columns to link to dimension tables.
• Dimension Tables: These tables describe the "who, what, where, when, and why" behind the numbers in your fact table. Examples include a Customers table, a Products table, or a Calendar table. They have many descriptive columns (like customer name, product category, or date) but far fewer rows than a fact table.

Think of it like this: your Sales fact table might contain SaleAmount, DateKey, ProductKey, and CustomerKey. The Products dimension table connects via the ProductKey and contains all the product details like ProductName, Category, Brand, and Color. This structure is incredibly compact and allows the Power BI engine to slice and dice data with minimal effort.

Reduce Cardinality and Column Count

Beyond the star schema, here are a few other data modeling habits that pay huge performance dividends:

• Remove columns you don't need. Before you load data, be ruthless. Each column you add consumes memory and slows down processing. If you're not going to use it in a visual or a calculation, remove it in Power Query before it even enters your model.
• Watch out for high cardinality. Cardinality just refers to the number of unique values in a column. Columns with millions of unique values (like a primary key on a massive fact table, a transaction ID, or an email address column in a multi-million-row sales table) can dramatically increase your file size and reduce performance. Always remove high-cardinality columns from fact tables if they aren't used for relationships.
• Split date/time columns. If you have a column with both date and time (e.g., 2024-10-26 10:30:15), but you only need to analyze trends by date, split them into a Date column and a Time column in Power Query. The date column will have much lower cardinality, improving performance.

Master Power Query for Efficiency

The Power Query Editor is where you clean and transform data before loading it into your model. Doing the heavy lifting here is critical for a snappy report.

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

Filter and Shape Your Data Early

The order of your "Applied Steps" in Power Query matters a lot. Always try to filter data or remove columns as your very first steps. This is key for taking advantage of a feature called query folding.

Query folding is Power BI’s ability to translate your transformation steps (like filtering or removing columns) into the native language of your data source, like SQL. When this happens, your database or API does all the hard work before sending the data to Power BI. The result is significantly less data being downloaded, processed, and loaded into your model.

You can check if a step is folding by right-clicking on it in the Applied Steps list. If the "View Native Query" option is available (not greyed out), then folding is happening. Perform your filtering and column removals first to give query folding the best chance to work.

Disable Unnecessary Background Refresh Options

As you're developing your report, Power BI is constantly running queries in the background to provide data previews in the Power Query editor. On very large datasets, this can make the development experience itself laggy. You can control this behavior.

Navigate to File > Options and settings > Options. Under the "Data Load" section for the Current File, consider unchecking "Allow data preview to download in the background." This stops Power Query from constantly running previews, giving you back computer resources and making the editor more responsive when working with complex transformations or huge datasets.

Write Smarter, Not Harder DAX

Once you have a lean data model, poorly written DAX (Data Analysis Expressions) can still be a major bottleneck. A single inefficient formula can bring your interactive report to a standstill.

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

Variables are Your Best Friend

Using variables with VAR is more than just good practice for readability - it's a massive performance booster. Variables let you calculate a value once, store it, and reuse it multiple times within the same formula. This prevents the DAX engine from having to run the same complex calculation over and over.

Inefficient DAX (calculating last year's sales twice):

YoY Growth % =
DIVIDE (
    ( [Total Sales] - CALCULATE ( [Total Sales], SAMEPERIODLASTYEAR ( 'Date'[Date] ) ) ),
    CALCULATE ( [Total Sales], SAMEPERIODLASTYEAR ( 'Date'[Date] ) )
)

Optimized DAX with a variable:

YoY Growth % =
VAR SalesLastYear = CALCULATE ( [Total Sales], SAMEPERIODLASTYEAR ( 'Date'[Date] ) )
RETURN
    DIVIDE ( ( [Total Sales] - SalesLastYear ), SalesLastYear )

The second version is far more efficient because it only calculates the value for SalesLastYear one time.

Avoid Full Table Scans with Iterators

DAX functions ending in "X" (like SUMX or COUNTX) and the FILTER function are called iterators. They work by iterating through a table row by row to perform a calculation. While necessary sometimes, they can be extremely slow on tables with millions of rows.

Whenever possible, substitute these iterator functions with a measure inside of CALCULATE. The CALCULATE function uses the engine's highly optimized filtering context, which is almost always faster than a row-by-row scan.

For example, imagine you want to calculate sales from a specific country, say, the USA.

Slow approach using FILTER:

USA Sales (Slow) =
CALCULATE (
    SUM ( Sales[SalesAmount] ),
    FILTER (
        ALL ( Customers ),
        Customers[Country] = "USA"
    )
)

Fast approach using a simple filter predicate:

USA Sales (Fast) =
CALCULATE (
    [Total Sales],
    Customers[Country] = "USA"
)

The second formula is far superior. It doesn't scan the entire Customers table, it uses the optimized engine to apply a direct filter and get the result much more quickly.

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

Fine-Tuning Your Report Page

Even with a perfect model and slick DAX, an overloaded report page can feel sluggish. Every visual on your canvas fires at least one query to the data model. More visuals mean more queries.

• Fewer visuals, more focus: A common mistake is to cram as many charts and KPIs onto a single page as possible. This overwhelms your users and generates a storm of queries every time the page loads or a filter is changed. Stick to the most important visuals to answer a specific business question.
• Edit visual interactions: By default, clicking a data point in one visual filters every other visual on the page. On a dense report, this can set off a long chain reaction of queries. Go to the Format tab and click Edit Interactions. You can then select a visual and decide which other visuals it should (and shouldn't) filter. Turn off interactions that aren't necessary.
• Use custom visuals cautiously: Not all visuals from the marketplace are optimized for performance. Stick with the core visuals or certified custom visuals whenever you can. If you must use a third-party visual, be sure to test its performance.

Find the Culprit with Performance Analyzer

If you're still facing performance issues, Power BI has a built-in tool to help you diagnose exactly where the problem is.

In Power BI Desktop, navigate to the View tab and click on Performance Analyzer. A pane will open on the right.

1. Click Start recording.
2. Now, interact with your report. Click a slicer, sort a visual, or switch pages to trigger the actions that feel slow.
3. The Performance Analyzer will show you the exact load time for every element on your page, broken down by:

This is your diagnostic roadmap. If you see a chart with a very long "DAX Query" time, you know you need to go optimize that visual’s underlying measures. If the "Visual Display" time is high, perhaps the visual is too complex or is a poorly optimized custom visual.

Final Thoughts

Optimizing Power BI is about making smart choices at every step, from designing a lean star-schema data model and writing efficient Power Query steps to crafting performant DAX and keeping your report canvas clean. By applying these best practices and using the Performance Analyzer to hunt down bottlenecks, you can turn slow, frustrating dashboards into lightning-fast tools for insight.

While these tips are powerful for building fast Power BI dashboards, we know that many marketing and sales teams spend too much time just getting their data ready for tools like this. Manually pulling data from Salesforce, Google Analytics, Shopify, and various ad platforms can take hours before you even start building. To address this, we built Graphed. It connects to all your mission-critical data sources in a single click and allows you to create real-time dashboards using simple, natural language. Instead of wrangling with complex tools, you can just ask questions and get instant answers, giving you back the time to focus on strategy instead of report-building.