Checking an Import Before You Commit It

Bulk import starts from the Bulk Import button on a list screen. The window asks you to download the template, fill the required columns marked with an asterisk, and upload the completed file as .xlsx, .xls or .csv. Two buttons then sit side by side: Check File and Import File. Check File is offered for the eight templates that import rows — vendors, items, customers, assets, asset custodians, warehouses, locations and the chart of accounts. Existing records are updated where a stable code, email or barcode matches, so the same file can add new records and change existing ones.

Check File does not use a separate set of rules. It runs the real import, row by row, inside a transaction that is always rolled back, so the result cannot drift from what Import will do. Each row runs on its own, which means one bad row does not hide the rest: every bad row is reported, not just the first. Later rows see earlier rows, so a file listing the same code twice shows one create and one update, exactly as the import would. While it runs, queued work and email are switched off, so a dry run never sends a low-stock alert or a welcome message.

A dry run checks up to 2,000 rows. A bigger file is refused with a message suggesting you split it or import it directly, and your plan’s rows-per-file limit applies to a check as well as to an import. In practice: an implementation lead migrating a wholesaler loads 1,400 items. Instead of importing and hoping, she clicks Check File and within a minute has a list of 37 rows that would fail and 112 with warnings, without a single record created in the live workspace.

The Import Dry Run page opens with a verdict: This file is ready to import, or This file would not import, with the number of rows to fix. Beside it are the counts — rows, new, updated, already up to date, skipped, with warnings and with errors. The table below has one line per row, numbered as in your spreadsheet, with the record’s name, a Result of New, Update, No change, Skipped or Error, and Details. Rows with errors and warnings are listed first; up to 200 clean rows follow, and the rest are summarised as more rows with no problems.

An Update row lists only the fields it would actually change, such as Changes: Price, Status. On an existing record a blank cell leaves that field as it is, so a short file that only fills Price changes only prices. An error is something the import refuses, written against the row in plain words: a value outside the list the screens offer, a missing required column, or a row that matches an existing record when you may add records but not edit them. One error stops the whole import, which is why the check lists all of them at once.

Warnings are the more valuable half, because they are things the import accepts silently: a Warehouse Code or Location Code that matches nothing, a Custodian Code that matches no custodian so the asset will have none, an Inventory Account that matches no account in your chart, the same record twice in one file so the later row overwrites the earlier, or Stock on a service item, which carries no stock and is ignored. In practice: a warning that Warehouse Code MSA-01 matches nothing turns out to be a typo for MSA01 across 300 rows; fixed before importing, those items land in Mombasa instead of in no warehouse at all.

When no row has an error, the button at the top reads Import with the number of rows, for example Import 1,400 rows. It imports the exact file you checked, held for you in your signed-in session, so there is no second upload to mix up. Rows with warnings are imported as they are, so read the warnings first. While any row has an error the button is disabled: fix the spreadsheet and run Check File again. The import itself is still all-or-nothing, so if something changed in between and a row fails at import time, nothing is saved.

A small file imports at once and the page reports how many records were created, updated and skipped. A file over 256 KB is handed to the background, and you are notified when it completes. The checked file can be imported once: after that, or if your session ends, you are told the dry run has expired and asked to upload the file again. Checked files nobody imported are deleted after a day, so an abandoned dry run does not leave your data lying on a server.

In practice: a finance officer updating supplier contacts runs Check File on 600 vendor rows and sees 598 updates and two errors — two rows are missing the phone number, which vendors require. She adds the numbers, checks again, reads that 41 rows will change email address and confirms that is intended, then clicks Import 600 rows. Because the dry run listed the changes field by field, she can tell her manager exactly what moved before it moved.