Import records from a file

Import records lives in the ⚙ (Actions) menu that sits next to the view title, above Export All. Unlike export, it is not restricted to a user group: any internal user who can create the records can import them. That is worth knowing before you hand the feature to somebody — an import writes as many records as the file has rows.

The entry is absent, with no explanation, in four cases:

Choosing Import records opens a full screen called Import a File, with the model you came from in the breadcrumb. Until a file is loaded there are only two buttons — Upload Data File and Cancel — and the body of the screen offers the download links for any templates that exist for this model, plus a link to the Import FAQ.

The file picker accepts .csv, .xls, .xlsx, .xlsm and .ods. Spreadsheet formats are the safer choice: dates, numbers and encodings are carried in the file itself, so the whole Formatting section below never has to be touched. A CSV is plain text and has to be described to the importer before it can be read.

A file that is not one of those formats is refused with "Unsupported file format …, import only supports CSV, ODS, XLS and XLSX", and an empty or damaged one with "Import file has no content or is corrupt".

A template is a spreadsheet with the column headers already written the way the importer expects, so the mapping is guessed correctly on the first try. Templates exist only for the models somebody wrote one for; on every other model the Help panel simply has no download link, and that is normal.

Models that ship a template today include contacts, products, price lists, vendor price lists, chart of accounts, journal entries, journal items, leads and opportunities, employees, timesheets, mailing contacts, bills of materials, project tasks, purchase orders and sales orders.

When there is no template, the reliable substitute is to export a handful of the same records first — Export All in the same ⚙ (Actions) menu, with I want to update data (import-compatible export) ticked. The file that comes back has exactly the column names this importer recognises.

Once a file is loaded, the left panel describes it and the right panel maps it. The top block, Data to import, names the file and offers the settings that decide how it is read at all.

Formatting — CSV files only

This block is absent for .xlsx, .xls, .xlsm and .ods files, because those formats carry their own encoding and cell types. For a CSV the importer guesses these settings from the first rows and shows you what it guessed; changing any of them and pressing Reimport re-reads the file with the new settings.

Batch Import — only above 100 rows

A file of more than 100 rows adds a Batch Import block and the notice "The file will be imported by batches". The import then runs in slices, with a progress bar and a control to stop it part-way, instead of as one long request that a worker may abandon.

Files to import — images and attachments

This block appears only once a column has been mapped to a binary field such as Image. Put the file names in that column, then click Upload your files and select the actual image files; the count of selected files appears beside the button. During the import each name is matched to an uploaded file. Names that match nothing are not an error — the record is created without the picture.

Images can also be given as a URL or as base64 text in the cell instead of as uploaded files. Importing by URL is restricted to administrators, because it makes the server fetch arbitrary addresses; everybody else is stopped with "You can not import file via URL, check with your administrator or support for the reason." A cell that is neither a URL, a known file name nor valid base64 is refused with "Found invalid image data, images should be imported as either URLs or base64-encoded data."

Advanced — developer mode only

With developer mode on, two more blocks appear. Neither is needed for an ordinary import, and both are easy to get wrong, which is why they are hidden.

The right-hand side is one row per column of your file, under three headings.

The importer guesses each mapping from the column header and from the shape of the first ten rows — a column of nothing but numbers is only offered numeric fields, for example. It also remembers: once you have mapped a column named, say, CUST_REF on this model, the same header is mapped the same way the next time anybody imports into it. That is what makes a recurring import from another system a one-click job after the first run.

A column mapped to nothing is skipped in silence. That is deliberate — it is how you import a file that carries extra columns you do not want — but it also means a column you meant to map and forgot produces no warning at all. If every column is unmapped the import refuses to start: "You must configure at least one field to import".

Two notes appear in Comments on their own, before any error:

• "To import multiple values, separate them by a comma." — on a column mapped to a tags-style field, where one cell holds several values.
• "This column will be concatenated in field …" — when two file columns are mapped to the same field, so their contents are joined rather than one overwriting the other.

The most common failure is not a broken file — it is a value that has no counterpart in the database: a country spelled differently, a category that does not exist yet, a status word nobody uses. For a column mapped to a link, a status or a yes/no field, the Comments cell then offers a chooser headed When a value cannot be matched:, and what it offers depends on the kind of field.

Choose one per problem column, then press Test again. The choice applies to that column for this import only.

Everything above is easier when the header row already uses the names WindoorERP knows. Start from a template or from an import-compatible export, then:

• Add, remove and reorder columns freely to fit your data. Order does not matter; names do.
• Keep the External ID column. It is the handle the system uses to recognise a record next time, and it is the difference between a second import correcting your data and a second import doubling it.
• Give every row a unique value in that column. Typing the first one and dragging it down the column is enough — the spreadsheet numbers the rest.
• Prefix external IDs that came from another system with the name of that system's table, so two tables cannot collide: company_1 and person_1 rather than 1 twice.

If a column stubbornly refuses to map, do not fight the guesser: export the same records, read the header the export produces for that field, and use that name in your file.

A field that points at another record can be given three ways, and you must pick exactly one per column. Mapping two of them onto the same field is what produces contradictory data.

When two records genuinely share a name — two product categories both called Sellable under different parents, say — matching by name links every row to whichever the importer finds first, which is almost never what you wanted. Either rename one, or switch that column to /External ID.

Two shapes need particular care:

• Tags and other many-to-many fields. Put every value in one cell, separated by commas, with no space after the comma: Manufacturer,Retailer.
• Lines inside a record, such as the lines of a sales order. Give each line its own row. The first line shares the row that carries the order's own fields; every further line gets a row of its own with those order columns left empty.

Three conversions cause almost all the rejected cells.

If the file carries an External ID or Database ID column, every row whose identifier already exists is updated; only the rest are created. That makes the round trip the standard way to correct hundreds of records at once:

1. In the list, tick the records to change — or none of them, to take the whole list.
2. ⚙ (Actions) › Export All, then tick I want to update data (import-compatible export). That adds the external identifier to the file and narrows the field list to fields that can actually be imported back.
3. Edit the exported file in a spreadsheet, leaving the identifier column untouched.
4. Import it back through the same screen.

The identifier column is the whole mechanism. Edit it, re-sort it away from its row, or delete it, and the second import stops recognising the records and creates a second copy of every one of them.

Test runs the entire import and throws the result away. It is not a sample and not a guess: everything that would fail, fails now, on the real data, with row numbers. It costs the same time as the import itself and saves the clean-up, so there is no good reason to skip it.

A clean test reports "Everything seems valid." A failed one puts a red banner at the top of the table — "The file contains blocking errors (see below)" — and fills the Comments cell of each guilty column.

Then press Import. Above 100 rows a progress bar replaces the screen and can be stopped part-way. Whatever was already written stays written — stopping is not an undo. When an import stops early, whether you stopped it or an error did, the panel says "Click 'Resume' to proceed with the import, resuming at line <n>." and "You can test or reload your file before resuming the import.", Start at line is set for you, and the Import button becomes Resume.

Load Data File, beside it, swaps the file without leaving the screen — the way to re-upload the same file after correcting it in the spreadsheet, keeping the mapping you have already fixed.

There is no Import records entry. Check the view first: it appears on list and kanban views only, and never where records cannot be created. On a phone-sized screen it is hidden outright.

The whole file shows as one column. Wrong separator. The importer says so — "A single column was found in the file, this often means the file separator is incorrect." — so change Separator: and press Reimport. If your file uses tabs and your spreadsheet will not write them, re-save it from the spreadsheet with the separator options exposed rather than fighting it here.

The preview will not load at all. "Import preview failed due to: … For CSV files, you may need to select the correct separator." Same cause, one step earlier; the encoding is the other suspect.

"Error while importing records: all rows should be of the same size, but the title row has X entries while the first row has Y." The file has been cut into columns wrongly — usually an unquoted separator inside a value, such as a comma in an address in a comma-separated file. Set Text Delimiter: or re-save the file with values quoted.

Arabic or accented text arrives as symbols. Wrong Encoding:. Choose utf-8 and reimport; if the file came out of an older Windows tool, try windows-1252 or windows-1251.

Dates land a month out. A day-month ambiguity the guesser resolved the other way. Set Date Format: explicitly and reimport.

The import fails with no useful message. "An unknown issue occurred during import (possibly lost connection, data limit exceeded or memory limits exceeded). Please retry in case the issue is transient. If the issue still occurs, try to split the file rather than import it at once." Lower Batch limit before splitting the file — it is often enough on a heavy model.

An image column is refused for size. "File size exceeds configured maximum (… bytes)". Resize the pictures before importing; a product thumbnail does not need a 5 MB photograph.

The import ran and you need to find what it touched. There is no undo, but there is a filter. Group or filter the list on Created on or Last Updated on for the minutes the import ran, and you have exactly the records it wrote.