How to Use What-If Analysis in Power BI with AI

Moving a report from a static summary to an interactive forecasting tool is one of the most powerful things you can do with your data. Instead of just looking at what happened last month, you can start building dynamic models to predict what might happen next. This article shows you exactly how to do that using Power BI's What-If analysis and how to make your models even smarter with built-in AI features.

What Exactly is What-If Analysis?

In simple terms, "what-if" analysis is a technique for exploring the potential outcomes of a decision by changing certain variables. It’s the process of asking questions like:

"What if we increase our marketing budget by 15%? How would that impact projected revenue?"
"What if we lower our product price by $5? How many more units would we need to sell to hit our profit target?"
"What if our website conversion rate improves by 0.5%? What would that mean for new customer acquisition?"

Instead of being assumptions buried in a spreadsheet, Power BI lets you build interactive sliders and controls directly into your dashboards. This allows you, your team, and your stakeholders to dynamically adjust variables and see the impact on key metrics in real-time. It transforms a backward-looking report into a forward-looking planning tool.

Start with a Solid Foundation: Your Data Model

Before you start building scenarios, you need a reliable data model. What-If analysis is only as good as the data it’s built on. While we won’t do a deep dive here, there are two golden rules:

Clean Your Data: Make sure your source data is accurate, complete, and free of errors. Garbage in, garbage out.
Establish Relationships: Your data tables should be properly related. For example, your 'Sales' table should be connected to your 'Products' and 'Date' tables. This allows Power BI to correctly slice, dice, and aggregate your data when you introduce new variables.

With a clean model in place, you're ready to create your first interactive parameter.

A Step-by-Step Guide to Creating a What-If Parameter

The core of what-if analysis in Power BI is the "What-if Parameter." This feature creates a simple, independent table that generates a series of numbers, which you can then control with an on-screen slider. Here’s how to build one using a common scenario: modeling the impact of an increase in ad spend.

1. Create the Parameter

First, navigate to the Modeling tab on the Power BI ribbon. In the "Parameters" section, click on New parameter and select an option. For traditional sliders, we are using What-If Parameter, which until recently was listed as the Numeric range option in new updates.

A dialog box will appear. Let's fill it out for our ad spend model:

Name: Ad Spend Increase (%). This will be the name of the table and the slider.
Data type: Whole number. (You could use Decimal for more granular control).
Minimum: 0. We want the baseline scenario to be a 0% increase.
Maximum: 100. Let’s cap our test scenario at a 100% (doubling) increase.
Increment: 5. This sets the step for the slider, so it will move in 5% increments (0, 5, 10, etc.).
Default: 10. This is the value a visual will display by default.

Finally, make sure the "Add slicer to this page" box is checked. This will automatically add a slider for your new parameter to your report canvas.

2. Understand What Power BI Just Did

When you click "OK," Power BI does two things in the background:

It creates a new table named Ad Spend Increase (%) with a calculated column. This column contains a series of numbers from your minimum to your maximum (0, 5, 10, ... 100).
It creates a DAX measure inside that new table named Ad Spend Increase (%) Value. The DAX expression is simple: SELECTEDVALUE('Ad Spend Increase (%)'[Ad Spend Increase (%)], 10). This measure will return whatever value is currently selected on the slider.

At this point, the slider is on your canvas, but moving it does nothing. It's not connected to your main dataset yet. That’s the next, and most important, step.

Bringing the Analysis to Life with DAX Measures

The What-If parameter is just a number generator. To make it useful, you need to use it in another measure to calculate a simulated outcome. Let's create a [Projected Revenue] measure that uses the value from our slider.

First, let's assume you already have a basic measure that calculates your existing revenue, something like:

Total Revenue = SUM(Sales[Revenue])

Now, let's create a new measure that models the projected revenue based on the slider value. Right-click on your main Sales table, select "New measure," and enter the following DAX formula:

Projected Revenue = [Total Revenue] * (1 + 'Ad Spend Increase (%)'[Ad Spend Increase (%) Value] / 100)

Breaking Down the Formula:

[Total Revenue]: This is our baseline, an existing measure.
'Ad Spend Increase (%)'[Ad Spend Increase (%) Value]: This pulls the current single numeric value from the slider (e.g., 25).
/ 100: We divide by 100 to convert the whole number from the slider into a percentage (e.g., 25 becomes 0.25).
1 + ...: This converts the percentage into a multiplier. For example, a 25% increase becomes a multiplier of 1.25.

3. Visualize the Result

Now you can use your new [Projected Revenue] measure in any visual. A Card visual is a great starting point. Drag [Projected Revenue] and [Total Revenue] onto cards and place them next to your slider.

As you move the Ad Spend Increase (%) slider, you will see the "Projected Revenue" card update instantly. You’ve successfully built an interactive what-if model!

AI-Powered Enhancements for What-If Scenarios

Creating a manual what-if scenario is powerful, but this is where Power BI's AI features can take your analysis to another level. They can help you figure out which variables matter most and explain the impact of your changes.

Find What to Test with the Key Influencers Visual

Sometimes the hardest part of what-if analysis is deciding what variables to even model. Do you test the impact of a price change, a marketing spend increase, or a new promotion? Guessing can be inefficient.

The Key Influencers AI visual can guide you. You can use it to analyze a historical outcome (like "Customer was High-Value") and it will identify the factors that were most correlated with that outcome. For example, it might analyze your sales data and report:

"When the Discount % is greater than 15%, the likelihood of a sale being over $500 increases by 2.1x."

Discovering this insight tells you that "Discount %" is a powerful lever in your business. This is a perfect candidate for creating a what-if parameter, allowing you to model exactly how different discount levels might impact overall revenue and profitability.

See the Impact with the Decomposition Tree

Once you’ve built your [Projected Revenue] measure, the Decomposition Tree visual is a fantastic AI tool for digging into the impact. This visual lets you break down a metric by multiple dimensions in any order you want.

You can set [Projected Revenue] as your metric and then explore how that projected revenue breaks down. For example, as you increase your 'Ad Spend' slider, you can use the Decomposition Tree to simultaneously see how it impacts revenue across different dimensions:

Show me the projected revenue by Country.
Now break down the United States' projection by Product Category.
From there, split the "Electronics" category revenue by Sales Rep territory.

This allows you to see not just the top-level impact but the detailed, granular changes your scenario might produce.

Explain Your Scenarios with Smart Narratives

Finally, your what-if model is only useful if people can understand its output. The Smart Narrative visual automatically generates a plain-English summary of the data on your report page.

When you combine it with a what-if parameter, the narrative becomes dynamic. As you move the slider, the text updates in real-time. It can automatically generate a summary like:

"With an ad spend increase of 25%, Projected Revenue is forecasted to be $1,250,000, which represents a $250,000 lift over current Total Revenue."

This is incredibly useful for instantly communicating the results of your analysis to anyone, technical or not, without them needing to perform mental math on their end.

Final Thoughts

What-If analysis transforms your Power BI reports from static historical views into dynamic, forward-looking tools for decision-making. By pairing interactive parameters with Power BI's built-in AI visuals, you can build smarter models that not only predict outcomes but also help you understand the key drivers behind them.

The core of this process is about making data interactive so you can get answers faster. We designed Graphed around this exact idea. Instead of spending time in model setup and DAX, we let you connect all your marketing and sales data sources in one place and simply ask questions in plain English – like "create a dashboard showing ROAS by campaign" or "what's our projected sales velocity for this quarter?" Graphed builds these real-time dashboards for you in seconds, letting you explore scenarios and collaborate with your team without the steep learning curve of traditional BI tools.