How to Convert JSON to CSV in PowerShell (Nested JSON Too)

To convert JSON to CSV in PowerShell, read the file as one string, turn it into objects with ConvertFrom-Json, and save them with Export-Csv:

(Get-Content -Path .\employees.json -Raw | ConvertFrom-Json) | Export-Csv -Path .\employees.csv -NoTypeInformation

That works for flat JSON. Nested objects, dates and records with different fields need a little more, so I tested each case in PowerShell 7 and Windows PowerShell 5.1.

Convert a JSON file to CSV

My sample file has three employees. -Raw reads the whole file as one string, and the parentheses matter in Windows PowerShell 5.1, as the next section shows:

(Get-Content -Path .\employees.json -Raw | ConvertFrom-Json) |
    Select-Object name, city, salary |
    Export-Csv -Path .\employees.csv -NoTypeInformation

Get-Content -Path .\employees.csv

Output:

"name","city","salary"
"Maria Lopez","Austin","98500"
"Tom Baker","Denver","87250"
"Dana Reyes","Seattle","91000"

Select-Object picks and orders the columns. -NoTypeInformation is the default in PowerShell 7 but needed in 5.1, so keep it for scripts that run in both. Reading a JSON file into an array covers the first half in more detail.

Why does my CSV have only one empty row?

In Windows PowerShell 5.1, ConvertFrom-Json sends a JSON array down the pipeline as one object instead of one object per record. Here’s the same pipeline with and without parentheses:

$csv = Get-Content -Path .\employees.json -Raw | ConvertFrom-Json |
    Select-Object name, city | ConvertTo-Csv -NoTypeInformation
"Without parentheses: $($csv.Count - 1) data rows"

$csv = (Get-Content -Path .\employees.json -Raw | ConvertFrom-Json) |
    Select-Object name, city | ConvertTo-Csv -NoTypeInformation
"With parentheses:    $($csv.Count - 1) data rows"

Output:

Without parentheses: 3 data rows
With parentheses:    3 data rows

That’s the PowerShell 7 output, which gives three rows either way. In 5.1 the first version writes one row of empty values. Wrapping the command in parentheses, or saving it in a variable first, fixes it in both.

Why do dates change when converting JSON to CSV?

ConvertFrom-Json in PowerShell 7 turns ISO timestamps into DateTime objects, and Export-Csv then writes them in your date format:

$rows = Get-Content -Path .\employees.json -Raw | ConvertFrom-Json
"hired is a $($rows[0].hired.GetType().Name)"
$rows | Select-Object name, hired | ConvertTo-Csv -NoTypeInformation | Select-Object -First 2

Output:

hired is a DateTime
"name","hired"
"Maria Lopez","10/7/2026 1:15:00 PM"

That’s PowerShell 7, which shows 1:15 PM because the Z means UTC. Windows PowerShell 5.1 leaves the text alone and writes 2026-10-07T13:15:00Z. In PowerShell 7.5 and later, add -DateKind String to keep the original text:

(Get-Content -Path .\employees.json -Raw | ConvertFrom-Json -DateKind String) | Export-Csv -Path .\employees.csv

For other date layouts, formatting dates in PowerShell shows how to write them your way.

Convert nested JSON to CSV

CSV has no place for objects inside objects. Export-Csv calls ToString() on them, which gives unreadable values, so pull the inner values out with calculated properties:

$people = Get-Content -Path .\nested.json -Raw | ConvertFrom-Json

"As-is:"
$people | ConvertTo-Csv -NoTypeInformation
"Flattened:"
$people | Select-Object name,
    @{ n = 'city';   e = { $_.address.city } },
    @{ n = 'state';  e = { $_.address.state } },
    @{ n = 'skills'; e = { $_.skills -join ';' } } |
    ConvertTo-Csv -NoTypeInformation

Output:

As-is:
"name","address","skills"
"Maria Lopez","@{city=Austin; state=TX}","System.Object[]"
"Tom Baker","@{city=Denver; state=CO}","System.Object[]"
Flattened:
"name","city","state","skills"
"Maria Lopez","Austin","TX","PowerShell;Azure"
"Tom Baker","Denver","CO","SQL"

I joined the skills array with semicolons so it stays in one cell. Select-Object calculated properties work the same way for any nested data.

Keep every column when records have different fields

Export-Csv builds the columns from the first record only. Dana’s manager field is the last one, so it silently disappears unless you collect every property name first:

$rows = Get-Content -Path .\employees.json -Raw | ConvertFrom-Json

"Default (columns from the first record):"
($rows | ConvertTo-Csv -NoTypeInformation)[0]

$allColumns = $rows | ForEach-Object { $_.PSObject.Properties.Name } | Select-Object -Unique
"All columns:"
($rows | Select-Object $allColumns | ConvertTo-Csv -NoTypeInformation)[0]

Output:

Default (columns from the first record):
"name","city","salary","hired"
All columns:
"name","city","salary","hired","manager"

Select-Object -Unique keeps the order the names first appear. Records without a manager get an empty cell. The Export-Csv guide has more on how it picks columns.

Convert JSON with a top-level property

APIs often wrap the list in a property, like employees. Converting the top-level object gives one useless row, so point at the array instead:

$data = Get-Content -Path .\wrapped.json -Raw | ConvertFrom-Json

"Top level:"
$data | ConvertTo-Csv -NoTypeInformation
"The employees array:"
$data.employees | ConvertTo-Csv -NoTypeInformation

Output:

Top level:
"company","employees"
"Contoso","System.Object[]"
The employees array:
"name","city"
"Maria Lopez","Austin"
"Tom Baker","Denver"

If you’re pulling JSON from a web API, getting JSON data from a URL shows how to fetch it first.

Convert JSON