# Create a Salesforce test case

> Build Salesforce test cases with smart NLPs and Copilot, work with auto-learned elements, and recover when the org changes.

Salesforce applications are hard to automate through the UI alone, because the workflows are long, the DOM is complex, and the org changes often. Three things work around that. Metadata learns your org's components, so elements exist before you write a step. Smart NLPs collapse long UI flows into single steps. Copilot records against the live org.

Before you start, you need a Salesforce project with a metadata connection, and the [Testsigma Recorder extension](https://testsigma.com/docs/v2/get-started/installation/recorder-extensions/) installed. See [Set up Salesforce testing](https://testsigma.com/docs/v2/application-types/salesforce/setup/).

## Create the test case

Go to **Create Tests > Test Cases**, create a test case, and build its steps with smart NLPs, with Copilot, or with both. Then click **Run**.

## Write steps with smart NLPs

A smart NLP does through the API what would otherwise take a page of UI interactions. Creating a lead and editing it takes 4 steps rather than dozens.

1. `Login to Salesforce application using Salesforce Connection connection`. Select a connection, or click **Add Connection** to create one.

2. `Switch to Application`, replacing the application from the dropdown.

3. `Create record in Salesforce Object Form using Salesforce Connection and store the record id in variable test data`. This is an API step:
   - Click **Salesforce Object Form** to open **Create record using API**, and select or search for the object.
   - Fill in the form that appears for that object and click **Save**. Only the fields you filled are sent in the request.
   - Click **Salesforce Connection** and select the connection.
   - Store the record ID in a variable, such as `Lead Records`. That variable is available to later steps and to other test cases.

4. `Open the edit Salesforce Object form where record is Record ID`, using the **$ Runtime** test data type to pass the variable from step 3.

![A Salesforce test case whose first step logs in through a connection](https://s3.amazonaws.com/website-static-docs.testsigma.com/new_images/projects/Updated_Doc_Images/update_sf_tc_2.png)

### Parameterize the login connection

Rather than hardcoding a connection, click **Salesforce Connections** on the login step, type `/`, and select a test data type such as **Environment**. The connection then comes from the environment the test runs in.

## Record steps with Copilot

1. Click **Copilot** on the Test Case Details page.

2. Confirm Copilot is ready in the **Debug & Record** overlay, and click **Launch**.

3. Click **Rec** and perform the actions you want as test steps.

4. Click **Stop Recording**, then **Exit Copilot**.

5. Click **Stop Session** in the **Stop & Exit Session** dialog.

6. Refresh the Test Case Details page to see the recorded steps.

Copilot and smart NLPs mix freely in one test case. A common shape is API steps to set up the data, then Copilot to record the UI flow that acts on it.

## Elements

Salesforce elements are learned automatically once metadata synchronizes, so you do not create them by hand.

Their names follow a convention. A field name is `ObjectName_FieldName`, so `Account_CreatedDate` is the `CreatedDate` field on the `Account` object. A screen name is the object name as it appears on screen, such as `Account`.

Auto-learned elements cannot be edited, and element names have to be unique to avoid conflicts.

To browse them, open a test case, add a step, hover over the element, and click **Select Element**. The **Elements** overlay lists everything learned.

To confirm an element was pre-learned, click it, select **View/Edit Element** in the Testsigma Debugger and Recorder, and the overlay tells you it is a standard Salesforce element.

### Create an element during recording

An element the metadata did not cover can be captured live.

1. Click **Copilot** on the test case.

2. Confirm Copilot is live in the **Debug & Record** overlay, select a connection under **Salesforce Metadata connection**, and click **Launch**.

3. Click the element in the NLP and select **Create Element**.

4. Click the UI element you want. The recorder captures its details.

5. Check the details and click **Create**.

![Create Element in the Debug & Record overlay, capturing an element from the live org](https://s3.amazonaws.com/website-static-docs.testsigma.com/new_images/projects/Updated_Doc_Images/Elem_Details_SF.png)

## When the org changes

A metadata refresh downloads your org's current structure and identifies fields that no longer exist, so a removed or renamed field surfaces as a warning rather than as a failed run later.

### Find the affected test cases

Impacted test cases are flagged with a warning on the **Test Cases** page. Click **View All** to filter to the deprecated ones, or use the highlighted filter to show only test cases containing errors.

### Fix them

1. Open the deprecated test case.

2. Hover over the highlighted step to read the error.

   ![A test case with an error count, the deprecated element highlighted, and Change Element offered](https://s3.amazonaws.com/website-static-docs.testsigma.com/new_images/projects/Updated_Doc_Images/Change_Deprecated_Element.png)

3. Hover over the element and click **View/Edit element** to open **Element details**, which names the deprecated element.

4. Replace it with a new element, or disable or remove the step.

The same warnings and filters work at step group level.

For a deprecated element specifically, disable the step that uses it, create a new element in the same context, and swap it in with **Change Element**.

## Frequently asked questions

### Why does authorization fail?

The client ID, the secret, the environment, or the callback URL is wrong. Check the client ID and secret for typos and regenerate them if needed, confirm the environment matches the org you are connecting to, and check the callback URL against the one in Salesforce. Also confirm the app's permissions and scopes are set, and that no firewall blocks the authorization endpoint.

### Why does the login step fail when metadata syncs fine?

The access and refresh tokens have expired, the same app is being used by several users at once, or the app's configuration changed in the org. Re-authenticate the connection to get fresh tokens, then check the app's settings in Salesforce.

### Why does an MFA challenge appear during metadata sync?

The org enforces MFA on direct UI logins. Turn off **Require multi-factor authentication (MFA) for all direct UI logins to your Salesforce org**, and check user-level MFA under **Account > Settings > Advanced User Settings > User Details > Profile**.

Rather than weakening MFA generally, enable **Waive Multi-Factor Authentication for Exempt Users** and create a permission set for automation users.

### Why does an MFA dialog appear during execution?

Salesforce treats a session from a lab or an agent as untrusted and challenges it. Go to **Setup > Quick Finder > Session Settings > Session Security Levels**, remove the MFA requirement, and confirm passwordless login and username and password login are both enabled.

### Why does execution fail on labs but pass locally?

The session was generated on one machine and executed on another, so Salesforce treats it as invalid.

1. Confirm the executing user has the same privileges as the System Administrator, under **Setup > Quick Finder > Users**.
2. Run the test on a local agent. If it passes, the problem is lab session handling.
3. Go to **Setup > Quick Finder > Session Settings** and disable **Lock sessions to the IP address from which they originated**, then enable **Lock sessions to the domain in which they were first used** and **Force relogin after Login-As-User**.