How to Do Sensitivity Analysis in Tableau

Building a dashboard is a great way to track what’s already happened, but what about what could happen? If your conversion rate increased by 10%, how much more revenue would you generate? What if your average ad CPC went up by $0.50? Modeling these kinds of ‘what-if’ scenarios is called a sensitivity analysis, and it's essential for smart forecasting and strategic planning. This article will walk you through exactly how to build a dynamic sensitivity analysis in Tableau so you can stop guessing and start modeling.

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

What is Sensitivity Analysis, Anyway?

In simple terms, sensitivity analysis is a financial model that determines how a target variable is affected by changes in other independent variables. It's a way to explore the impact of uncertainty. You have a handful of key assumptions that drive your business outcomes - things like conversion rate, average order value, customer churn, or ad spend. A sensitivity analysis lets you build a small, interactive model where you can tweak those assumptions up or down to immediately see the potential impact on your bottom line.

For example, a marketing manager might build a model to see how changing website traffic and conversion rates affect the total number of new leads. By creating sliders to adjust both variables, they can answer questions like:

• "If our next campaign drives 20% more traffic, but our conversion rate stays the same, how many leads will we get?"
• "What if traffic only goes up 5%, but we manage to improve the landing page conversion rate by half a percent?"

This transforms your dashboard from a static, historical report into an interactive forecasting tool, allowing you to prepare for different outcomes and make better strategic decisions.

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

The Two Core Components You'll Need in Tableau

To pull this off in Tableau, you need to get familiar with two key features: Parameters and Calculated Fields. They work together to create the interactive experience.

1. Parameters

Think of parameters as the steering wheel of your analysis. They are workbook variables that you, the user, can control. Instead of being tied to a specific field in your data source, a parameter allows you to input a custom value. In a sensitivity analysis, a parameter becomes the slider or user input box for one of your assumptions, like "Percent Change in Ad Spend" or "Increase in Average Order Value." When you move the slider, you’re not changing your original data, you’re just changing the value of the parameter that will be used in a calculation.

2. Calculated Fields

Calculated fields are the engine. They perform the math. You’ll write formulas in calculated fields that reference your parameters. For instance, you could create a "Projected Revenue" calculated field with a formula like [Current Revenue] * (1 + [Percent Change in Ad Spend]). This formula takes the parameter's current value (e.g., a 15% increase) and applies it to your existing data to calculate the hypothetical outcome. Every time you adjust the parameter, Tableau automatically re-runs the calculation and updates your visualization.

Step-by-Step Guide: Building a Revenue Projection Model

Let's build a practical example. Imagine you're an e-commerce manager and you have data on your website traffic, average conversion rate, and average order value (AOV). You want to see how changes to your conversion rate and AOV could impact total projected revenue.

Let's assume your baseline metrics for the last month were:

• Total Sessions: 100,000
• Conversion Rate (CVR): 2.0%
• Average Order Value (AOV): $75.00

Based on this, your current revenue would be 100,000 * 0.02 * $75 = $150,000. Now let's build an interactive model around that.

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

Step 1: Create Your "What-If" Parameters

First, we need to create the controls that will let us adjust our assumptions. We'll create two parameters: one for the conversion rate and one for the AOV.

Parameter for Conversion Rate Change:

1. On the Data pane (left side), click the small dropdown arrow at the top and select "Create Parameter..."
2. Name it CVR Percentage Change.
3. Set the Data type to Float. (This allows decimal values).
4. Set the Display format to Percentage.
5. Under Allowable values, select Range.
6. Set the Minimum to -0.5 (a 50% decrease), the Maximum to 0.5 (a 50% increase), and the Step size to 0.01 (for 1% increments).
7. Click OK.

Parameter for AOV Change:

1. Create another parameter and name it AOV Change.
2. Set the Data type to Float or Integer.
3. Set the Display format to Currency (Standard).
4. Under Allowable values, select Range.
5. Set the Minimum to -25 (a $25 decrease) and the Maximum to 25 (a $25 increase). Set the Step size to 1.
6. Click OK.

You've now created the interactive controls. To see them, right-click on each parameter in the Parameters list and select "Show Parameter." Sliders for CVR Percentage Change and AOV Change will now appear on your worksheet, usually on the right side.

Step 2: Create the Calculated Fields

Next, we need to write the formulas that use our parameters to project a new revenue figure. We'll build this in a few steps for clarity.

1. Projected Conversion Rate

This field calculates the new conversion rate based on the parameter's input.

• Click the dropdown on the Data pane and select "Create Calculated Field..."
• Name it Projected CVR.
• Enter the following formula. This takes our baseline 2% conversion rate and modifies it based on the parameter:

0.02 * (1 + [CVR Percentage Change])

2. Projected AOV

This field calculates the new average order value.

• Create another calculated field named Projected AOV.
• Enter this formula, which takes our baseline $75 AOV and adds or subtracts the value from the parameter:

75 + [AOV Change]

3. Projected Revenue

This is our final output, tying everything together.

• Create a final calculated field named Projected Revenue.
• This formula uses our baseline traffic and the two new calculated fields we just made:

100000 * [Projected CVR] * [Projected AOV]

Note: For a real-world scenario, you wouldn't "hard-code" the numbers like 100000, 0.02, and 75. Instead, you'd reference the actual data fields from your source, like SUM([Sessions]), a calculated field for your Original CVR, and AVG([OrderValue]). But hard-coding is fine for learning the concept.

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

Step 3: Build the Visualization

Now we just need to display the result and see our model in action.

1. Drag the Projected Revenue field onto the Text mark on the Marks card.
2. You'll see a number appear on the screen. Try moving the CVR Percentage Change and AOV Change sliders you showed earlier. You will see the revenue number update instantly!
3. To make it look cleaner, format the Projected Revenue number as currency. Right-click the field in the Data pane, go to Default Properties > Number Format..., and select Currency (Standard).

You can make this even easier to interpret by showing the projected value alongside the original baseline revenue.

• Create one more calculated field named Baseline Revenue with the formula 100000 * 0.02 * 75.
• Drag Measure Names onto the Columns shelf.
• Drag Measure Values onto the Rows shelf.
• In the Measure Values card that pops up, keep only Baseline Revenue and Projected Revenue.

Now you have a simple bar chart that instantly compares your baseline with your hypothetical scenario, providing immediate context on the potential gain or loss.

Practical Tips for Effective Sensitivity Analysis

Building the model is only half the battle. Using it effectively is what provides real value. Here are a few tips:

• Start with High-Impact Drivers: Don't try to model every variable in your business. Focus on the 2-3 key metrics that have the biggest impact on your goals. For e-commerce, that's often traffic, conversion rate, and average order value. For SaaS, it might be new trials, trial-to-paid conversion rate, and churn rate.
• Set Realistic Ranges: When setting up your parameters, use ranges that are both optimistic and pessimistic, but still realistic. A 500% lift in conversion rate overnight probably isn't going to happen. Keeping your ranges grounded makes the analysis more credible and useful.
• Test One Variable at a Time: To truly understand the sensitivity of a single driver, keep your other variables at zero while adjusting the one you want to test. This helps you isolate the impact of, for example, a 10% increase in conversion rate without it being influenced by a simultaneous change in traffic.
• Integrate It Into Your Dashboards: Don't keep your sensitivity model in a separate, obscure workbook. Build it into your primary marketing or sales dashboard so it becomes part of your regular review and planning sessions.

Final Thoughts

That's how you can turn Tableau from a tool that just reports the past into an interactive model that helps you plan for the future. By combining user-controlled parameters with dynamic calculated fields, you can instantly answer critical "what-if" questions, explore potential outcomes, and make much more informed, data-driven decisions for your business.

While modeling in tools like Tableau is incredibly powerful, setting up all the parameters, calculated fields, and visualizations for every question can still be time-consuming. That's one of the frictions we aim to eliminate with Graphed. We connect directly to your live data sources, but instead of requiring you to build these models manually, you can simply ask questions in plain English, like "Show me a chart forecasting my revenue if our conversion rate increases by 15% next quarter." Our AI analyst builds the visualization for you in seconds, saving you from navigating menus and writing formulas so you can get right to the insight.