Debug
When a step fails, the results page carries everything needed to work out why: the error, the screenshots either side of the failure, the logs, and what else depends on the element. This is the order to work through it in.
Open the failing step
Section titled “Open the failing step”-
Go to Run Results and click the test plan.
-
Click the test case you want to investigate.
-
Select the failed step from the step list on the left.

The list is headed by the step count and the number of failed steps, and each step carries an icon showing whether it passed, failed, or was not executed. Drag the divider between the list and the details pane to resize either side.
The step header shows the step status, start time, duration, step number, and step action. Its icons open the Logs, Step Settings, and Metadata panels, plus Locators on a healed step. See Step details panels.
Read the error and the verdict
Section titled “Read the error and the verdict”A failed step shows 2 sections below the step header: ERROR and ANALYZER VERDICT. Testsigma writes a verdict when a step cannot find its element (NO_SUCH_ELEMENT), and when a verification step reads a value that does not match (an ASSERT_ERROR code). For verification steps, see Read the verdict for a verification step.

ERROR shows the error code and the error message, with Read more for the full text. When the page has iframes or shadow DOM hosts that were not searched, the message says so. If the element sits inside one of them, add a step that switches into it before the failed step.
Testsigma analyzes the failure as soon as you open the step. While it runs, the section reads “Reading the evidence for this step. This usually takes under a minute.” The ANALYZER VERDICT then carries the result:
| Part | What it shows |
|---|---|
| Category | The type of failure as a badge, such as Wrong locator or Flow changed. Badges after it describe the failure in more detail |
| Summary | One line naming the failed step and the cause, then a short explanation |
| REASON | Why the step failed, from the page source, the screenshots, and the earlier steps, compared against the passing baseline where one exists. Read more expands it |
| Recommendation | What to change, and how to confirm the fix |
| Evidence | What the screenshots show for this run and the baseline |
| Fix | One or more named fixes, each with a one-line description. When Testsigma can make the change, it lists options and an Apply fix button |
To fix the step from here:
-
Check the fix option selected in the ANALYZER VERDICT. Where it lists several, select the one you want.
-
Click Apply fix. The button reads Applying… while Testsigma updates the test case, then Fix applied.
-
Click Re-run beside it. Testsigma starts a new run of the test case.

The verdict never changes the test case on its own. Where it names a fix without an Apply fix button, make the change in the test case.
Two similar failures can get different categories, because each one is analyzed separately. Read the summary and REASON for the cause in each run.
For suggestions you can apply, follow-up questions, or a bug report, click Analyze with Agent at the bottom of the page. See Analyzer.
Steps after the failed step do not run when the step is set to stop the test case on failure. They show the not executed icon.
When the failure started at an earlier step
Section titled “When the failure started at an earlier step”A step can fail because the run left the recorded path before it. An earlier step opened a different page from the one recorded, did not finish loading it, or left a pop-up covering it, so the element the failed step needs never appeared. The verdict category says which of these happened.
| Category | What happened | Fix |
|---|---|---|
| Flow changed | The run reached a different page or state from the one recorded, so the element the step targets is not on the current page | Re-record the steps from where the run left the recorded path, add the missing navigation before the failed step, or make a step conditional when the screen it handles appears only on some runs |
| Prerequisite not completed | An earlier step did not finish as recorded, such as a page that was still loading or a pop-up that was not closed, so the next step could not find its element | Select the fix option in the verdict, such as Add a wait for the page to settle, and click Apply fix |

-
Select the failed step and read the ANALYZER VERDICT. The summary names what was on the page instead, such as a product page where a search page was expected. REASON explains how the run left the recorded path, and whether earlier failures caused it. The recommendation names the step to re-record from, or what to add before the failed step.
-
Select the steps before the failed step, and check their screenshots in THIS RUN to find where the run left the recorded path.
-
Fix the test case. For Flow changed, re-record the steps from where the run diverged, or add the missing navigation step. For Prerequisite not completed, select the fix option and click Apply fix.
-
Rerun the test to confirm the fix.
Not every verdict has an Apply fix button. When the fix means re-recording steps, the verdict names and describes it, and you make the change in the test case.
Where a screen such as a sign-in pop-up appears only on some runs, use a conditional step, for example Click on element if visible.
In a Copilot session, the same analysis appears in the chat. See Diagnose a failed step.
Read the verdict for a verification step
Section titled “Read the verdict for a verification step”A verification step, such as Verify that the current page displays text or Verify that the element displays text, fails when the value it reads does not match the expected value. ERROR names how the 2 values differ. ANALYZER VERDICT says why, and whether to change the test, the application, or the moment the check runs. Verification steps are not auto-healed, because the element was found.

The first badge is the category:
| Category | What it means | What to do |
|---|---|---|
| Expectation drift | The test is behind the application, or the value changes by design | Update the expected value, or replace it with a test data parameter |
| Preconditions not met | The step checked the wrong thing, or the element could not be checked fairly, such as a form field value read as text, a locator that matched several elements, or an element covered by another | Change what the step checks, narrow the locator, or add a step that clears what covers the element |
| Prerequisite | An action the step depends on did not happen first | Add the missing step before the verification step |
| Flow changed | The run was on a different page or state from the one recorded | See When the failure started at an earlier step |
| Application defect | The application is wrong, and the test caught it | Report the defect. The test is correct |
| Page still loading | The step read the value before the data arrived | Add a wait before the verification step |
| Unidentified | No cause had enough evidence | Review the evidence and investigate the step |
Update the expected value
Section titled “Update the expected value”When the only difference is cosmetic, such as letter case, spaces, or encoding, the verdict offers Update the expected value to what the page shows, with the change, such as "LOGIN PAGE" → "Login Page".
-
Check that the difference is not what the step exists to catch.
-
Click Apply fix. Testsigma writes exactly the value shown into the step’s test data and changes nothing else.
-
Click Re-run. The step passes with the new value.
When the values differ in substance, the option reads Review the step’s expected value, with “The new value is worked out when you apply this fix.” Check that the page is right before you click Apply fix. Testsigma then shows “Test data updated. Re-run the test case to check it held.”
Read the error code
Section titled “Read the error code”ERROR names how the values differ:
ASSERT_ERROR_TEXT_CASE: the text matches except for letter caseASSERT_ERROR_TEXT_WHITESPACE: the text matches except for spaces or line breaksASSERT_ERROR_TEXT_ENCODING: the text matches except for how characters are encoded, such as a curly quote in place of a straight oneASSERT_ERROR_TEXT_SUBSTRING: one value contains the otherASSERT_ERROR_TEXT_NOT_RENDERED: the text is in the page source but is not shownASSERT_ERROR_TEXT_IN_CSS: the text comes from a CSS rule, not the page contentASSERT_ERROR_TEXT_IN_FORM_FIELD: the text is a form field’s value, not text shown on the pageASSERT_ERROR_ELEMENT_ZERO_SIZE: the element has no width or heightASSERT_ERROR_ELEMENT_DISABLED_BY_ANCESTOR: a parent element disables itASSERT_ERROR_ELEMENT_DISABLED_BY_ARIA: anaria-disabledattribute disables itASSERT_ERROR_ELEMENT_MULTIPLE: the locator matched more than one elementASSERT_ERROR_ELEMENT_ATTRIBUTE_ABSENT: the attribute the step checks is not on the elementASSERT_ERROR_ELEMENT_OBSCURED: another element covers the elementASSERT_ERROR_PAGE_URL_PARTIAL: the URL differs in one part, such as the pathASSERT_ERRORandASSERT_ERROR_PAGE: the values differ, and the step cannot say how
The analyzer does not write a verdict for conditional steps (If, Else If, Else, While), steps with Ignore this step result in Test Case Result selected, or Wait until steps that time out. A verification step that cannot find its element gets the element-not-found verdict instead.

Check the logs
Section titled “Check the logs”-
Click the Logs icon in the step header. The Logs panel opens on the right.
-
Click Console or Network.
-
Select This step from the scope list to see only the logs captured during the step.
-
Click the download icon to save the logs.

The Network tab shows the number of failed requests, the time window the requests started in, and a table of requests with their status, method, URL, size, and time. The total request count sits below the table.
Some logs have no step markers. The log then shows every entry for the run, with the message Not scoped to this step: this log has no step markers, so the whole run is shown.
Network logs are captured only when they are switched on for the test machine, and they arrive once the upload from the test machine finishes. Until then the Network tab reads No network log was recorded for this run.
Check whether the step was healed
Section titled “Check whether the step was healed”When auto-healing resolves a locator mid-run, the step’s status reads Healed and the step shows the auto-heal icon in the step list. A banner above the list reports how many steps were healed.
Select the healed step to see the Healed this step card, with the locator change and Approve as Primary. See Healer.
Compare against an earlier run
Section titled “Compare against an earlier run”Comparing a failing step against a run that worked shows what changed. Two entry points:
- Compare Runs in the page header compares the whole test case across 2 runs, step by step
- Compare Steps in the THIS RUN section compares the current step alone
See Results.
Check what else the element affects
Section titled “Check what else the element affects”When an element causes a failure, check which other tests depend on it before you fix it.
-
Click Affected tests in the step header.
-
Review each tab in the Affected Instances panel: Test Cases, Step Groups, Test Suites, and Test Plans, each with a count.

Where nothing else uses the element, the panel reads There are no Test Cases associated with the element.
Act on the failure
Section titled “Act on the failure”All from the results page:
- Apply fix in the ANALYZER VERDICT, where the verdict lists fix options, including a new expected value for a verification step. See Update the expected value. When the run left the recorded path, re-record the affected steps instead
- Analyze with Agent in the action bar returns the error type, the root cause, and suggestions you can apply as a step update
- Run using Copilot in the action bar reruns the test case interactively
- Update element corrects a locator yourself
- Relearn step re-captures the step
Where auto-healing fired, approve the new locator from the Healed this step card.
Common causes of failure
Section titled “Common causes of failure”- The element or UI identifier changed in the application
- The application does not load
- The local internet connection fails
- The element is unchanged, but navigation to it changed
- Addon code does not perform as expected
- The application under test contains a defect
- The result is a false positive or false negative
- An API is down
- The element locator is incorrect
Was this page helpful?
Thanks for the feedback.