How to Use What-If Analysis with ChatGPT

Thinking about the future of your business often means asking a lot of "what if" questions. What if we increased our ad spend? What if we hired two more salespeople? What if we discounted our best-selling product? Answering these questions is the core of what-if analysis, a powerful technique for modeling different scenarios to make smarter decisions. This article will show you how to use a tool you probably already have open - ChatGPT - to run these analyses quickly, without needing complicated spreadsheets.

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 What-If Analysis, Anyway?

At its core, what-if analysis is the simple process of changing variables in a model to see how those changes affect the outcome. It’s something you likely do in your daily life without thinking about it. For example, if you're planning a vacation budget, you might ask:

• "What if we fly on a Tuesday instead of a Friday? How does that change our total cost?"
• "What if we stay in an Airbnb instead of a hotel? How much more money does that leave us for activities?"

In a business context, this process helps you anticipate challenges, spot opportunities, and build more resilient strategies. Instead of just reacting to changes, you can proactively plan for them. It’s the difference between guessing what will happen and modeling a range of potential futures.

Why Use ChatGPT for Your Scenarios?

Traditionally, this type of analysis lived inside spreadsheets using tools like Excel's Scenario Manager or Goal Seek. While powerful, these tools can be clunky and require a bit of a learning curve. ChatGPT offers a faster, more conversational, and often more creative way to explore possibilities.

Here’s why it works so well:

• It's Fast and Conversational: You can model complex scenarios by simply describing them in plain English. There are no formulas or cells to manage. This turns a potentially tedious task into a quick brainstorming session.
• It Requires Zero Setup: You don’t need special software or templates. If you can write a prompt, you can run a what-if analysis.
• It Uncovers New Angles: ChatGPT can sometimes introduce factors or potential outcomes you hadn't considered, adding a new layer to your strategy sessions.

One important thing to remember: ChatGPT is a powerful reasoning engine, but it's not a foolproof calculator. For purely mathematical or mission-critical financial modeling, always double-check its math. Think of it as a brilliant brainstorming partner who sometimes makes small slip-ups.

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

A Step-by-Step Guide to What-If Analysis with ChatGPT

Running a successful analysis with ChatGPT depends entirely on the quality of your prompts. Follow these five steps to ensure you’re getting clear, valuable responses.

Step 1: Set the Stage with Good Context

Vague prompts lead to vague answers. Before you ask your "what if" question, give ChatGPT all the necessary background. The goal is to create a small, self-contained model for it to work with. Your context should include a clear starting point with defined metrics.

Pro Tip: Start by assigning a role. This puts ChatGPT in the right frame of mind and improves the quality of its response.

Here’s an example prompt for setting up a marketing budget scenario:

Act as a marketing strategist. I'm planning my ad budget for next quarter and need your help with some what-if analysis. Here's our baseline situation: First, calculate my total number of conversions and total revenue based on this current setup.

Step 2: Ask Your Main What-If Question

With the context established, you can now introduce your change. Be specific about the variable you want to modify. Instead of saying "what if I spend more on Facebook?" say "what if I move $2,000 from the Google Ads budget to the Facebook Ads budget?"

Using the scenario above, your prompt could be:

Great. Now, run this "what-if" scenario: What if I change the spend allocation to be $5,000 for Google Ads and $5,000 for Facebook Ads? Keep all other metrics (CPC, conversion rate, revenue per conversion) the same. Calculate the new total number of conversions and the new total revenue. How does this compare to our baseline?

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 3: Add Layers by Chaining Your Questions

This is where the real power of conversational AI comes in. The initial setup is good, but in reality, things are more complicated. One change will likely affect another element. In our example, moving more budget onto Facebook will likely increase the platform's CPC.

You can model this complexity by asking follow-up questions that build on the previous response.

A good follow-up prompt:

That's interesting. However, it's likely that increasing our Facebook spend so much will raise our CPC. Let's run a new scenario. We'll still use the 50/50 budget split ($5,000 to each platform). But this time, assume the Facebook Ads CPC increases by 20% due to the higher spend. Re-calculate the revenue for this scenario.

Step 4: Explore Multiple Scenarios at Once

Don’t limit yourself to changing one thing at a time. The real world has many moving parts. You can ask ChatGPT to model several scenarios simultaneously to get a broader view of your options.

An example of a multi-scenario prompt:

Okay, this is helpful. Now I want you to model and compare three different scenarios to the baseline. Scenario 1: We move $2,000 from Google to Facebook (as you calculated earlier). Scenario 2: We keep our budget allocation the same ($7k/$3k), but a new landing page improves our Google Ads conversion rate to 6%. Scenario 3: We increase our total budget to $12,000 and allocate the extra $2,000 to Facebook Ads. Calculate the projected revenue for each scenario.

Step 5: Ask for a Summary Table

After running multiple scenarios, the conversation can get cluttered. The best way to make sense of everything is to have ChatGPT consolidate the results into a table. This is easy to read, simple to copy, and great for sharing with your team.

A simple final prompt:

Please summarize the baseline and all three scenarios in a Markdown table. The columns should be: "Scenario Name", "Google Spend", "Facebook Spend", "Total Conversions", and "Total Revenue".

This will give you a clean, organized output comparing every potential outcome, making your final decision much clearer.

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

Real-World Examples of What-If Analysis

You can apply this method to almost any area of your business. Here are a few practical examples to get you started.

Sales Team Headcount Planning

Scenario: You're a sales manager deciding if you should hire more reps next quarter. You need to understand the potential impact on revenue.

• Context: "I have a sales team of 4 reps. Each rep receives 80 marketing qualified leads (MQLs) per month. The average MQL-to-close rate is 15%, and the average deal size is $3,000. Start by calculating our current expected monthly revenue."
• What-If Questions: "What happens to our total revenue if we hire 2 new reps, assuming the number of leads per rep drops to 65 due to splitting the existing leads?" and "Now, what if we keep the team at 4 reps, but a new training program boosts our close rate to 18%?"

E-commerce Pricing and Discount Strategy

Scenario: You run an online store and are considering a flash sale for your top-selling product.

• Context: "My flagship product sells for $80. The cost to produce it (COGS) is $30. On a typical month, I sell 1,000 units. Please calculate my current monthly profit for this product."
• What-If Questions: "What happens to my total profit if I run a 20% discount and sales volume increases by 60% as a result? Compare that to a scenario where I only offer a 10% discount and sales volume rises by 20%."

Product Launch Decision Making

Scenario: You're a founder trying to estimate the profitability of a new SaaS feature tier.

• Context: "I'm launching a new 'Pro' plan for $49/month. Our fixed costs (salaries, software) are $25,000 per month. The variable cost per customer is $5/month for server usage. I need to figure out my break-even point."
• What-If Questions: "What is our total monthly profit if we sign up 800 customers in the first three months? Now, what happens to that profit figure if our variable costs are actually $7 per month per customer? Finally, how many customers do we need to reach break-even if costs are $7 per month per customer?"

Final Thoughts

Using ChatGPT for what-if analysis allows you to quickly explore options, test assumptions, and build a better feel for the dynamic forces driving your business. By starting with clear context, building with well-defined prompts, and asking for summary tables of outcomes, you can turn it into a powerful planning tool that doesn’t require spreadsheet skills.

While ChatGPT is incredible for modeling scenarios based on manual input, its real power unlocks when you connect it to your actual business data. Instead of copying and pasting data, integrate your tools like Google Analytics or Salesforce, and instantly ask "what if" questions using natural language, such as "What would happen if we increased our Facebook ad campaign?" or "Show a dashboard forecasting revenue if we launch a new service package." We believe your time is too valuable to be spent copying and pasting numbers - instead, you should be making decisions based on live data.