Power Platform Flow Expressions and Dynamic Content

Every flow moves data from one step to the next. Dynamic content picks values from earlier steps, and expressions reshape those values into exactly what you need.

Dynamic Content

When you click inside a field, a panel lists values from the trigger and previous actions. Pick Subject from an email trigger and the flow inserts that subject at run time. The panel saves typing and prevents spelling mistakes.

The Expression Editor

Next to the dynamic content tab sits the Expression tab. It accepts functions that calculate, format, or test values. Flow expressions use the Workflow Definition Language, which looks like spreadsheet formulas.

Field: Email Subject
  +---------------------------------------------+
  | concat('Order ', triggerBody()?['OrderId']) |
  +---------------------------------------------+
Result at run time:  Order 1045

Text Functions

Function	Job	Example
concat	Joins text	concat('Hello ', 'Ravi')
toLower / toUpper	Changes case	toLower('ABC')
substring	Cuts a piece	substring('Dataverse', 0, 4)
replace	Swaps text	replace('a-b', '-', ' ')
split	Breaks into an array	split('a,b,c', ',')
trim	Removes extra spaces	trim(' text ')

Date and Time Functions

utcNow()                                  current time
addDays(utcNow(), 7)                      one week from now
formatDateTime(utcNow(), 'dd MMM yyyy')   28 Oct 2026
convertTimeZone(utcNow(), 'UTC', 'India Standard Time')

Date functions drive due dates, reminders, and report names that include the current date.

Logic Functions

if(greater(variables('Amount'), 5000), 'Manager', 'Auto')
equals(triggerBody()?['Status'], 'Open')
and(equals(a, 1), not(empty(b)))
coalesce(triggerBody()?['Phone'], 'Not given')

Coalesce returns the first value that is not empty, which makes it ideal for default values.

Reading Data From Earlier Steps

• triggerBody() reads the output of the trigger.
• body('Action_Name') reads the full output of an action.
• outputs('Action_Name') reads the raw output including headers.
• items('Apply_to_each') reads the current row inside a loop.
• variables('VarName') reads a variable you created.

The question mark in ?['Field'] protects the flow when a field is missing. The flow returns null instead of failing.

Diagram: Data Travelling Through a Flow

[Trigger: new row]  Name = "asha", Amount = 7200
        |
        v
[Compose]  toUpper(triggerBody()?['Name'])  --> "ASHA"
        |
        v
[Condition] Amount greater than 5000 ?
     | yes                        | no
     v                            v
[Send approval]            [Auto approve]
Subject = concat('Review for ', outputs('Compose'))

Layman Example: A Kitchen Blender

Dynamic content gives you the raw fruit. Expressions work like the blender blades that cut, mix, and shape it. A smoothie recipe may need only the pulp, a splash of milk, and a pinch of sugar. An expression selects, combines, and formats the data until the next step receives a clean mix.

Debugging Expressions

Use Compose Actions

Place a Compose action after a tricky step and read its output in the run history. You see the exact value the flow produced.

Check Data Types

Many errors come from mixing text and numbers. Convert values with int(), float(), or string() before comparing them.

Watch Quotation Marks

Text inside an expression uses single quotes. Double quotes cause an error.

Practical Mini Examples

A file name that includes a date uses concat('Report_', formatDateTime(utcNow(), 'yyyyMMdd'), '.xlsx'). A greeting that adapts to an empty name uses coalesce(triggerBody()?['Name'], 'Customer'). A deadline seven days away uses addDays(utcNow(), 7). Build a small flow that sends yourself an email with each of these values and read the results to see how each function behaves.

Working With Arrays and Objects

Many connectors return lists. The first() function reads the first item, length() counts items, and join() combines items into one text value with a separator. The Select action reshapes a list into new fields, and the Filter array action keeps only the rows that match a condition. Parse JSON turns raw text into named fields that appear in the dynamic content panel. These tools let a flow clean a list of fifty items before sending a single tidy email.

Key Takeaways

Dynamic content fetches values and expressions transform them. A handful of text, date, and logic functions covers most real flows.