FREE TOOL · JIRA CLOUD
Jira CSV import validator
Paste or open the CSV before you import it. Every row is checked against the rules of Jira’s CSV importer: Summary, parents and sub-tasks, dates, estimates, labels, users and attachments. You get the row, the reason and the fix, and a corrected file. The file stays in your browser.
Result
6 errors, 2 warnings in 4 work items
- Error, row 2: Its Parent "1" comes later in the file. Parents must come before their children: move row 3 up.
- Warning, row 2: "payments, web" holds several values in one cell. Put each one in a separate Labels column (the fixed file below does it).
- Error, row 2: Original Estimate "4h" must be a number of seconds: 14400.
- Warning, row 2: Assignee "Maria Lopez": Jira matches users by email address or Atlassian account ID.
- Error, row 4: Original Estimate "2h" must be a number of seconds: 7200.
- Error, row 5: Summary is empty. Every work item needs one.
- Error, row 5: Parent "9" matches no Work item ID in the file and is not a work item key such as PAY-12.
- Error, row 4: Date "03/12/2026" is not in the format of the other dates (yyyy-MM-dd). The importer takes one date format per file.
- Note: Work type values to map: Story, Epic, Sub-task. Values Jira does not have can be created while mapping.
Fixed file
- 2 estimates turned into seconds (8-hour day, 5-day week).
- Values in one cell split into repeated columns (1 field).
- Rows reordered so every parent comes before its children.
Work item ID,Work type,Summary,Parent,Labels,Labels,Original Estimate,Due Date,Assignee 1,Epic,Checkout v2,,payments,,,2026-12-01,[email protected] 2,Story,Pay by card,1,payments,web,14400,2026-12-03,Maria Lopez 3,Sub-task,Card form,2,,,7200,03/12/2026, 4,Story,,9,,,,,
The fix moves parents above their children, puts each label, component and version in its own column and turns estimates like 2h into seconds. Missing summaries, unknown parents and mixed dates need you: they stay as they are.
What Jira’s importer requires, and what is checked
All quotes are from Atlassian’s Import data from a CSV file unless marked.
| Rule | Atlassian says | The validator |
|---|---|---|
| Summary | “Each CSV file must possess a heading row with a Summary column”; it is “the only required field” | Error when the column or a row’s summary is missing |
| Quotes | Content with commas and new lines “is enclosed in quotes”; a quote inside is doubled: ""Add"" | Parses quoted values; a row with more values than columns points to an unquoted comma |
| Several values | “specify the same column name for each value you wish to import to the same field” | Warns on “a, b” in one Labels, Components or version cell; the fix splits it |
| Parents | Columns “Work item ID”, “Work type” and “Parent”; “the parent work items must appear before their subtasks or child work items” | Error for a parent that comes later, a Parent that matches no ID or key, a duplicate ID |
| Dates | The date format uses “syntax that complies with the Java SimpleDateFormat” | Detects the format, flags rows in another one or not matching yours |
| Estimates | Original and remaining estimate and time spent: “specified as number of seconds” | Error for 2h or 1d, with the seconds; the fix converts them |
| Users | Supported: “email address” and “Atlassian Account ID” | Warns on a display name such as Maria Lopez |
| Attachments | URLs “support the HTTP and HTTPS protocols” | Error for a file path |
| Several spaces | Columns “Space name” and “Space key”, on every row | Error when one of the two is missing |
| Status | “Can only be mapped to existing workflow statuses” | Lists the status values to check before mapping |
| Size | “Recommended number of work items per file: 1500” | Warns above 1,500 rows |
A template to start from
An epic, a story under it and two sub-tasks under the story, with two labels, estimates in seconds, due dates in yyyy-MM-dd, a comment in Atlassian’s “date;user;text” form and a quote inside a summary. Map Work item ID, Work type and Parent in the importer, and enter yyyy-MM-dd as the date format.
Work item ID,Work type,Summary,Description,Parent,Priority,Labels,Labels,Original Estimate,Due Date,Assignee,Comment Body 1,Epic,Checkout v2,"New checkout, card and wallet",,High,payments,web,,2026-12-18,[email protected], 2,Story,Pay by card,"As a buyer, I pay by card",1,High,payments,,28800,2026-12-11,[email protected],"2026-12-01 09:00;[email protected];Spec agreed" 3,Sub-task,Card form,,2,Medium,web,,7200,2026-12-04,, 4,Sub-task,"Error message for ""card declined""",,2,Low,web,,3600,,,
Why Jira’s own Validate is not enough
Jira checks your mapping against your site, which no page outside Jira can do, so use its Validate too. But Atlassian’s guide to the new import experience warns that your import can fail if errors are present after 30 rows: it checks the first 30 only. The admin importer has no Validate button at all (JRACLOUD-64607, “Include validation button on Admin CSV importer”), and the non-admin importer’s Validate has reported a false error for parent and sub-task files (MIG-395). This page reads all rows before you start.
Date formats
System dates such as Created, Updated and Due Date use “yyyy-MM-dd HH:mm:ss” internally, and custom date fields must match the “Date format in input file”, by default yyyyMMddHHmmss (common CSV questions). Whatever you choose, use one pattern for the whole file:
| In your file | Pattern to enter |
|---|---|
| 2026-12-01 | yyyy-MM-dd |
| 2026-12-01 14:30 | yyyy-MM-dd HH:mm |
| 01/12/2026 (1 December) | dd/MM/yyyy |
| 12/01/2026 (1 December, US) | MM/dd/yyyy |
| 01.12.2026 | dd.MM.yyyy |
| 01/Dec/26 2:30 PM | dd/MMM/yy h:mm a |
MM is the month and mm the minutes; HH is 0–23 and hh1–12 with a for AM/PM. A day/month file where no day is above 12 cannot be told from a month/day one: the validator says so instead of guessing.
Updating work items that already exist
“To update existing work items, your CSV file needs to contain a column that maps to Work Item Key. If a work item exists for a given key, it will be updated.” Leave a cell empty to keep the value; Atlassian’s example removes labels with the marker <<!clear!>>. The validator notes when a key column is present, so you know the import will change work items rather than add them.
After the import: the due dates on one board
Imported a plan with due dates? Deadline Tracker shows every work item across your spaces by due date, from overdue to two weeks out, lets you drag a card to reschedule it and sends a daily reminder of what is due. Free for up to 10 users. Counting working days for those dates: the business days calculator. Writing queries for what you imported: the JQL builder.
Frequently asked questions
What date format does the Jira CSV import use?
The one you type in the importer, written as a Java SimpleDateFormat pattern such as yyyy-MM-dd or dd/MM/yyyy HH:mm. Every date in the file, Due Date and Created included, must follow it. The validator detects the format your file uses and flags the rows that differ.
How do I import sub-tasks or child work items from CSV?
Add three columns, Work item ID, Work type and Parent. Give every row an ID, put the parent’s ID in Parent, and list parents before their children. Parent can also hold the key of a work item that already exists, such as PAY-12.
How do I import several labels or components?
Repeat the column: Labels, Labels, Labels, one value in each. A cell with "payments, web" is one label with a comma in it, not two labels. The fixed file the validator offers splits such cells for you.
How do I update existing work items with a CSV import?
Add a column mapped to Work item key. Rows whose key exists in the space are updated instead of created. Atlassian’s example clears a field with the marker <<!clear!>>.
In what unit are estimates and time spent?
Seconds. 2h is 7200, a day of 8 hours is 28800. The validator converts 2h, 1d 4h or 30m for you in the fixed file.
Why not just use Jira’s Validate button?
Use it too: it checks your field mapping against your site. But the new import experience checks only the first 30 rows, and the admin importer has no Validate button. This page reads every row before you start, and says what to change.
Is my file sent anywhere?
No. The file is read by your browser and checked there. The page has no analytics, and its Content-Security-Policy (connect-src none, form-action none) forbids it from sending any request.
Sources
- Atlassian Support, Import data from a CSV file, read 21 November 2026.
- Atlassian Support, Common CSV file questions and known issues, read 21 November 2026.
- Atlassian Support, Import data to Jira using a CSV file (new import experience), read 21 November 2026.
- Jira issue tracker: JRACLOUD-64607, MIG-395.