1. Docs
  2. Build tests
  3. Automate
  4. Variables and test data

Variables and test data

Steps rarely need fixed values. This page explains how a step reads a {{variable}}, which steps produce one, where data sources and datasets come in, and how a pipeline runs the same test once per data row.

6 min read Updated CustomersAdmins

In short

On this page you can

  • Insert a {{variable}} instead of a fixed value in any step.
  • Store a value from the screen, an API or a query for later steps.
  • Read test data from data sources, global parameters and datasets.
  • Keep one value per environment with NAME_DEV, NAME_TEST, NAME_PRP.
  • Run a test case once per dataset row in a pipeline.

How to use it

  1. Select the step and click the data picker icon next to a value.
  2. Pick a data source key, a global parameter or a dataset column.
  3. To pass a value on, give the producing step a key such as orderNo.
  4. Use {{orderNo}} in a later step, save and dry run.
  5. In the pipeline, set the execute count to iterate over the dataset.

How variables are written#

A variable is a name in double curly braces. You can use it as the whole value or inside text: Order {{orderNo}} was created. When the step runs, the variable is replaced by its value.

WriteMeansExample
{{name}}A value from this run, or a stored global parameter{{orderNo}}
{{tc:name}}A value this test case produced{{tc:orderNo}}
{{tc:1484.name}}A value test case 1484 produced earlier in the same pipeline run{{tc:1484.customerId}}
{{ds:name}}A key from a data source{{ds:adminUser}}
{{ds:name}}_columnOne column of a data source record{{ds:customers}}_email
{{excel:name}}_columnOne column of the current row of a dataset{{excel:logins}}_password

The prefixes make the source explicit and are what the data picker inserts. A name without a prefix still works, with limits (resolution order).

Insert a variable with the data picker#

  1. Select the step

    Click the step in the list. Its values open in Parameters List.

  2. Open the picker

    Click the picker icon next to the value (“Data Sources - Global Parameters - Datasets”). It is offered on typing, selection, navigation, hover, visibility, assertion and global parameter steps, on all mobile steps, and inside code, AI prompts, mail and Kafka messages.

  3. Choose the source

    The dialog Data Sources / Global Parameters / Datasets has three tabs:

  4. Pick and save

    Click the entry. The variable is inserted where your cursor was. Click Save.

    The value shows the variable in colour, for example {{ds:adminUser}}.

The picker warns you about order mistakes: “This parameter is produced by this step, so it has no value inside the step itself” and “This parameter belongs to a later step and may not have a value yet”.

Data sources and datasets are set up in the Data Source module of the main menu.

Steps that produce a variable#

A step produces a variable when you give its output a name (a key). Later steps read it as {{key}}.

StepProduces
Get (web), Get Text / Get Attribute / Store Variable (mobile, desktop)The text or attribute of a control; on a Get step, transforms can clean it first
APIValues extracted from the response by JSON path or XPath, plus @status and @header:Name
DatabaseOne variable per selected column
MailThe OTP code or link found in the mail
NetworkValues extracted from a network call
Python, JS, AI stepThe returned value, when the output is set to VALUE
Global parameterThe keys and values you type into it
  • Keys can only contain letters, digits and underscore (“Parameter key can only contain letters, digits and underscore”).
  • Each key that creates a variable must be unique in the test case. Save stops with “Names that produce a global parameter must be unique within the test case. Conflicts: …” otherwise.
  • Code and AI steps have three output modes: VALUE creates a variable; STATUS turns the returned value into pass or fail and creates no variable; NA produces nothing. An AI step that performs an action (AI Do) has no output.
  • A value longer than 8,000 characters is not stored.

Kafka steps produce no variable: the step dialog has no output fields. The exact parameters of each producing step: step reference.

Where a variable gets its value#

Resolving {{name}}When the step runs
{{tc:name}}   1. values produced earlier in this test case (shared run data)
              2. values of earlier test cases in the pipeline
              3. the project's stored global parameter of that name
{{ds:name}}   the data source key          {{excel:key}}_column → the dataset row
{{name}}      1. values produced earlier in this test case
              2. only when the reference is the whole value: the stored global
                 parameter, i.e. this test case's variable, else a data source key or dataset
              (inside longer text, step 1 only)
environment variant first: NAME_<ENV> if it exists, else NAME
not found → the text {{name}} is used literally

Example: step 3 is a Get with key orderNo and reads A-1029. Step 7 types {{orderNo}} into a search field: it types A-1029. If step 3 is deactivated, step 7 looks for a stored global parameter orderNo: values that steps produce are also stored, so it may type the order number from an earlier run. If there is none, it types the literal text {{orderNo}}.

An assertion treats a variable it cannot find as empty and logs “{{x}} not found, treated as empty”.

A wrong value is worse than no value

Because unresolved variables are typed as written and stale values can come from earlier runs, check the dry-run log of the step that should produce the value. For an API extraction, for example, the log shows the stored value as orderNo = A-1029.

Scope#

  • Run data belongs to one test case run. A test case does not see another case’s values by plain name. Step-level details: How long a variable lives.
  • In a pipeline, a later test case reads an earlier one’s value with {{tc:<caseId>.name}}.
  • A template runs inside its caller: it reads the caller’s variables and the caller reads what it produces. The template step’s Reads and Produces list them.

One value per environment#

A user name or a customer number often differs between DEV, TEST and PRP. Keep it as separate data source keys or datasets named NAME_DEV, NAME_TEST and NAME_PRP, and use only {{NAME}} in steps. A run on TEST uses NAME_TEST if it exists, otherwise NAME. The suffix must be upper case and exact. A name you write with a suffix is used as written.

Example · data source keys
adminUser_DEV   dev.admin@example.com
adminUser_TEST  test.admin@example.com
adminUser       fallback.admin@example.com

Step value: {{ds:adminUser}}   →   on TEST: test.admin@example.com

Run a test case once per data row#

A test case that reads {{excel:logins}}_user and {{excel:logins}}_password can run once for every row of the logins dataset. You set this in the pipeline, as the test case’s execute count, in Schedule:

Execute countRuns
{{excel:logins}}_#Every row, in order
{{excel:logins}}_#5The first 5 rows
{{excel:logins}}_?Every row, in random order
{{excel:logins}}_?55 random rows

Each row is a full run of the test case with a fresh browser. If you ask for more runs than there are rows, the rows repeat from the start.

Dry run uses the first row

A dry run does not iterate. It uses the first row of each dataset and the values stored in the steps; pipeline parameter overrides do not apply.

Test case values and pipeline values#

Each pipeline keeps its own copy of the parameters of the test cases it runs, so the same test case can run with different values in different pipelines. When you change a test case’s parameters, Save offers to sync them into its pipelines; see Test Case Parameters Changed.

Secrets and personal data#

  • Values typed into a step are stored in the test case and visible to everyone in the project. Put passwords and tokens in a data source and reference them with {{ds:…}}, so they live in one place.
  • Values typed into fields whose name looks like password, secret, token or PIN are shown as •••••• in run logs.
  • The runner receives the values it needs to run the test, wherever it runs: the server or your agent.
  • Use test users and test data. Never put real customer data into datasets.

If something goes wrong#

Didn’t find your answer?Write to support with your project name and what you see on screen.

info@smartestqa.ai
© 2026 Selenity Technology · SmartestQAsmartestqa.aiThis page as MarkdownChecked against smarttest-fe predev 2a8c79cc, smarttest-be 633c2a03, smarttest-testsuite runner, research/automate-steps-and-runs.md