--- title: "Run a What-If Analysis" description: "Change a Policy's rules, run it against a baseline simulation, and see exactly what moved." ---

TL;DR

A What-If Analysis lets you change a Policy's rules and see what those changes would have done, without touching the Policy itself. Run a normal Simulation first to serve as the baseline, then click Run → What If Analysis on the Policy, pick that baseline, edit the rules, and run. The results come back as a side-by-side comparison against the baseline. If you like what you see, you can copy the changes back into a real Policy.

## What is it? A **What-If Analysis** answers questions like *"what happens to our approval rate if we drop the minimum FICO from 660 to 640?"* before you commit to the change. When you start one, the platform copies your Policy into a behind-the-scenes policy that belongs to the job. You edit the rules on that copy and run it. While it stays a what-if, it cannot affect the live Policy: it never appears in the Policy Registry, and it cannot be approved, deployed, or scheduled. That copy is not discarded when you are done. If the results are good, you can write the changes into a real Policy from the job itself (see [Apply the changes](#apply-the-changes-optional) below), which converts the copy into a normal draft Policy. If you abandon the analysis instead, the copy is cleaned up. The result is always a comparison. A what-if job is created already paired with its baseline, so when it finishes you get the baseline and your modified version side by side, with every changed block, segment, and rule marked. What-If Analysis is available on **Policies** only. ## Before you start You need a **completed Simulation on the Policy** to act as the baseline. This is the one hard requirement, and there is no way around it: a what-if measures change, so it needs something to measure against. The baseline must be a plain [Simulation](/user-guide/how-tos/run-a-simulation/) that finished successfully on the version of the Policy you are working from. A previous what-if job cannot be used as a baseline. The Policy itself can be in any state. Draft, approved, or live all work, though the Policy's state does change how you apply your changes at the end. ## How to Run a What-If Analysis

#### Start the analysis

Open the Policy from the **Policy Registry**, click **Run** in the top right, and choose **What If Analysis**. You are asked **Do you have a baseline scenario?** - Click **Yes** to pick your baseline. Choose the completed Simulation you want to start from and click **Start What-If Analysis**. - Click **No** if you have not run one yet. The platform offers to **Run Baseline Scenario**, which takes you to the normal Simulation form. Run it, wait for it to finish, then come back and start the what-if again.

#### Change the rules

You land on the job form with an **Update Rules** section showing the Policy's decision workflow. Click any block to open it and edit its segments, rules, and thresholds. You can change the logic inside blocks, but not the shape of the workflow itself. Blocks cannot be added, removed, renamed, or rewired, so the what-if stays comparable to its baseline. :::caution[Leave before running and you lose the work] The rule changes only exist inside this pending job. If you navigate away, you are asked to **Discard unsaved progress?**; discarding deletes the what-if copy and your edits with it. Once the job has run, the analysis is preserved and this no longer applies. :::

#### Run the job

Fill in the rest of the job form as you would for any [Simulation](/user-guide/how-tos/run-a-simulation/), then click **Run Job**. Sampling, Reports, and Data Sources all behave the same way. Scheduling is not offered. A what-if is a one-off analysis, so it cannot be set to run later or on a recurrence.

#### Read the comparison

When the job completes, open it. Alongside the usual job output and any reports you selected, a **What-If Scenario** section shows the baseline and your modified version side by side, with changes marked:

Badge	What it means
Added	Present in your what-if, not in the baseline.
Deleted	Present in the baseline, removed in your what-if.
Modified	Present in both, but the logic changed.
Reordered	Unchanged logic, different position in the sequence.

Badges appear at block, segment, and rule level, so you can trace a shift in approval rate down to the specific rule that caused it. Because the comparison is what makes a what-if meaningful, it is fixed. You cannot swap the baseline, point it at a different Policy, or stop the comparison. To try a different set of changes, start a new what-if from the same baseline.

#### Apply the changes (optional)

If the results are what you wanted, use the job's actions menu to write the changes into a real Policy. Which options you get depends on the state of the Policy's latest version:

Action	What it does
Copy as New Policy	Creates a standalone draft Policy from your what-if changes, leaving the original untouched. Always available.
Update Latest Policy Version	Overwrites the latest version in place. Offered when that version is still editable (a draft). You are asked to confirm, since the current contents are replaced.
Create New Version	Creates a new version of the original Policy carrying your changes. Offered when the latest version is approved or live, and so cannot be edited.

Each of these produces a **draft**. The what-if itself is never promoted; it only hands its changes to a normal Policy, which then goes through [approval](/user-guide/how-tos/approve-an-object/) like any other.

## What's Next - [Approve an Object](/user-guide/how-tos/approve-an-object/) to take the resulting draft Policy through the normal review path. - [Run a Simulation](/user-guide/how-tos/run-a-simulation/) for the full job form reference, including Sampling, Reports, and Data Sources. - [Understand Job Output](/user-guide/how-tos/understand-job-output/) to read the row-level results behind the comparison.