TL;DR: Send complex data back from a flow as one text output containing JSON via Respond to a Power App or flow, then rebuild it in Power Apps with ParseJSON. You get nested records, arrays and typed values from a single field, and it stays standard/non-premium because the flow runs in the Power Apps app context.

🧰 Free tool: Paste a JSON sample and get ready-to-use Power Fx — field access with casts, a Type() definition and a ClearCollect/ForAll for arrays (with a custom collection name): ParseJSON in Power Apps converter.

Returning structured data from Power Automate to a canvas app used to mean either the old Split string trick or defining a long list of individual return outputs. Neither scales for nested objects or arrays. The modern way: return a single JSON string, and let ParseJSON turn it back into a Dynamic value (or a typed object) inside Power Apps. One output, any depth.

💡 Challenge

Your flow produces a complex result, for example a customer record with a list of orders, and you want it in your canvas app.

  • Respond to a Power App or flow only offers flat, typed outputs (Text, Number, Boolean, …). Modelling a nested/array structure means dozens of fields, or it is simply not possible.
  • Many makers reach for the Response (HTTP) action instead, which shows a premium diamond. That action belongs to the Request/Response (HTTP) connector and is premium.
  • The result: over-engineered flows or an accidental premium dependency for a scenario that does not need one.

✅ Solution

Serialize your result as JSON text, return it in one Respond to a Power App or flow output, and parse it in Power Apps with ParseJSON. Respond to a Power App or flow (from the Power Apps connector) is standard. As long as the rest of the flow uses no premium connector, the flow runs in app context and needs no premium license, only a Microsoft 365 license. That is the “shown as premium but it isn’t” confusion: the premium diamond you saw is on the HTTP Response action, not on this one.

🔧 How it’s done

1. Trigger the flow from Power Apps.

🔸 Use the Power Apps (V2) trigger so inputs are typed and the flow stays in app context.

2. Build the JSON in the flow.

🔸 Shape your data with Select / Compose, or wrap objects and arrays and run them through json(). Example payload:

{
  "customer": { "name": "Jane Doe", "vip": true, "since": "2022-09-01" },
  "orders": [
    { "id": 1001, "total": 149.90 },
    { "id": 1002, "total": 32.50 }
  ]
}

3. Return it as one text output.

🔸 Add Respond to a Power App or flow → add a Text output, e.g. jsonResult → set its value to the JSON string (the Compose/json() output). Only this one field leaves the flow.

4. Parse it in Power Apps (dynamic).

🔸 On your button’s OnSelect, capture the response and parse it:

Set( gblResp, ParseJSON( ReturnFlowData.Run(TextInput1.Text).jsonresult ) );

// Access nested fields with dot notation + explicit casting
Set( custName, Text( gblResp.customer.name ) );
Set( isVip,    Boolean( gblResp.customer.vip ) );
Set( since,    DateValue( gblResp.customer.since ) );

🔸 Turn the array into a typed table with Table() + ForAll():

ClearCollect(
    colOrders,
    ForAll(
        Table( gblResp.orders ),
        { id: Value( ThisRecord.Value.id ), total: Value( ThisRecord.Value.total ) }
    )
)

5. (Optional) Skip the casting with a typed ParseJSON.

🔸 Pass a Type as the second argument so the result is directly usable, no per-field conversion:

Set(
    gblTyped,
    ParseJSON(
        ReturnFlowData.Run(TextInput1.Text).jsonresult,
        Type(
            {
                customer: { name: Text, vip: Boolean, since: Date },
                orders: Table( { id: Number, total: Number } )
            }
        )
    )
);

// gblTyped.customer.name and gblTyped.orders are ready to use, no Text()/Value() needed

🎉 Result

Any structure, nested records and arrays included, travels back to Power Apps through a single standard output. No exploding return-field lists, no old Split hacks, and no premium license for a scenario that shouldn’t need one.

🌟 Key Advantages

🔸 One output, any shape: nested objects and arrays in a single JSON string instead of dozens of flat return fields.

🔸 Standard, not premium: Respond to a Power App or flow is a Power Apps connector action; it runs in app context on a Microsoft 365 license. The premium diamond lives on the separate HTTP Response action.

🔸 Typed on demand: use the Type argument of ParseJSON to get ready-to-use records and tables without manual casting.

🔸 Future-proof: replaces the legacy Split string method for returning arrays.

🛠️ FAQ

Q1: Is the Respond to a Power App or flow action premium?

No. It is part of the Power Apps connector and is standard. The premium diamond you may have seen is on the Response action of the Request/Response (HTTP) connector, which is a different action. As long as the flow uses no premium connector, it runs in app context and needs only a Microsoft 365 license.

Q2: Why does ParseJSON need Text(), Value() or Boolean() around fields?

Without the Type argument, ParseJSON returns a Dynamic value. JSON has no dedicated date/GUID types, and numbers/booleans still need an explicit cast for safe use in formulas. Use Text(), Value(), Boolean(), DateValue() etc., or pass a Type as the second argument to skip casting.

Q3: How do I handle a JSON array?

Wrap it with Table( gblResp.arrayField ) to get a single-column table of Dynamic values, then use ForAll() with ThisRecord.Value.<field> to project it into a typed table (see step 4).

Q4: My dates come back wrong. Why?

JSON only stores dates as strings. Return them in ISO 8601 (e.g. 2022-09-01) and convert with DateValue(); for other formats, wrap in Text() first. DateValue/DateTimeValue use the current user’s language settings.

Q5: Is there a size limit?

Keep payloads reasonable. Very large results should be paged or stored (e.g. in a SharePoint/Dataverse source the app reads directly) rather than pushed through a single flow response.

🔗 Related Tips

Marcel Lehmann

Marcel Lehmann

Microsoft MVP Microsoft MVP

BizzApps MVP from Switzerland 🇨🇭 - PowerPlatform Expert & Evangelist & MVP - Turning passion into expertise

MVP since 2023 • Power Platform Expert since 2017
Follow on LinkedIn About Marcel Request Workshop