Bulk CSV import (field reference and common gotchas)
When you have more than a handful of products to add, or a price list to update every month, you do not want to open each product. Settings → Import / Export loads a CSV file into the store in the background and maps your columns to TitanCart’s fields on screen, so the file does not have to be in any particular layout. This article covers preparing the file, the import step by step, what every product column does, how the job runs and how to stop one, and the handful of behaviours that silently produce wrong data if you do not know about them.
In one sentence: a UTF-8 CSV with a header row, up to 10 MB, mapped column by column; products are matched by Product Code, yes/no columns take
1/0oryes/no, the Import Settings are sent with the file so set them before you choose it, and Perform test import really does stop after five rows.
1. What can be imported
The Entity Type list offers Products, Categories, Orders, Order Items, Product Reviews, Waitlist and Customers. Each is a separate file and a separate run; an order and its lines, for example, are two imports (section 6). The page is reached from Settings, from the Import and Export buttons at the top of Products → Products, and from the Dashboard’s Import Data tile.
Moving from WooCommerce on the same site? Do not export and re-import: Settings → WooCommerce Migration reads WooCommerce’s own tables directly. See Migrating from WooCommerce. A WooCommerce or CS-Cart CSV export can be loaded through this page, but most columns have to be mapped by hand and the category and image formats differ (section 7).
2. Preparing the file
- CSV, UTF-8, header row first, 10 MB at most. A larger file is refused with File exceeds the 10 MB limit — split it. A file saved in another encoding (Windows-1252 from older Excel versions) is not converted: accented characters arrive garbled. In Excel use CSV UTF-8; in Google Sheets or Numbers the default export is fine. A byte-order mark is handled.
- Any delimiter Excel produces. Comma, semicolon or tab is detected from the header line (whichever occurs most). Quoted values with commas or line breaks inside are read correctly.
- Header names are yours. They are matched to TitanCart’s fields on the mapping screen. A header equal to a field’s label (Product Code, Quantity, Track Inventory) or its internal key (
product_code) is pre-mapped; a header of four or more characters that begins a key, or that itself begins with a key (Desc → Description, Price USD → Price), is pre-mapped as a guess; anything else you map by hand. - Start from the sample. Below the file box, Download sample CSV for Products (or whichever type you chose) gives a file whose headers pre-map, with example rows — including four T-shirt rows (red and blue, M and L) that show how a variation group is declared. Its Meta Keywords column has no field and stays on Skip; Color, Size and Brand only pre-map once attributes of those names exist in your store.
- An export re-imports as it is. Since 0.2.78 the Export tab’s column labels match the import fields, so on re-import every products column is pre-mapped (Quantity, Track Inventory and Images included — older exports called them Stock Qty, Inventory Tracking and Image URLs, which you had to map by hand). Only ID is left on Skip — keep it there — and so is Date Created if you included it; the attribute columns in a products export are always empty. Since 0.2.77 the products export writes Categories and Secondary Categories in the importer’s own
Parent///Childformat, so a re-imported export lands in the same tree; an export made with an older version holds flat names, and re-importing it makes the first name in each cell the product’s main category. The orders export is a report, not an importable orders file.
3. The import, step by step
Step 1 — Upload CSV
The Import Settings travel with the file. They are sent the moment you pick the file, so set them first; changing a box after the upload has no effect on that job (start over with ← Import Another File if you forgot one).
- Choose the Entity Type.
- Set the Import Settings (the block is shown for every type and names it — Import all categories from the file, and so on):
- Import Strategy — Import all … from the file (create new, update existing), Update existing … only or Create new … only. For products (matched by Product Code) and categories (matched by name and parent) all three work as written: an unknown code under Update existing only, or a known one under Create new only, is skipped without a message and counts as processed — except that an unknown code with no Name is refused first as a new product without a name (section 4), so a price-only file with a typo in a code shows that row as an error. Customers and orders only distinguish Create new only (an existing e-mail or order number is skipped) from the other two (existing records are updated and new ones are still created) — there is no update-only run for them. Reviews, order items and waitlist entries ignore the setting.
- Perform test import — only the first 5 rows of the file are written; the rest are never read. Run it, check the five products, then import the same file again with the box off (the five rows are simply updated).
- Set quantity of all products to zero — when the file has no Quantity column mapped at all, every product in the file gets stock 0. If Quantity is mapped but a cell is empty, that product’s stock is left alone.
- Delete existing additional images — on every product the file updates, the extra (non-main) images are unlinked before the file’s images are added, whether or not that row has an Images value. Use it when re-importing a file with images to avoid duplicates; leave it off for a price update.
- Images Directory — where bare file names in the Images column are looked for, relative to the WordPress root. Default
wp-content/uploads/titancart/import-images/; upload the pictures there by FTP first.
- Click the box Click to select a CSV file. The upload starts as soon as you pick the file; there is no separate upload button.
Step 2 — Map Columns
Each CSV column is listed with a Map To drop-down (— Skip — or a TitanCart field; required fields carry a *) and the first row’s value as Sample Data. The banner ⚡ Required: names what the type needs. Check every pre-mapped row: a guessed match (section 2) can be wrong, and a column called Size lands on an attribute of that name only if you have one. Map by hand what is wrong and set what you do not want to Skip.
On Pro with the Advanced Products Import extension enabled you can 💾 Save as preset (the mapping plus the Import Settings, keyed by your header names) and pick it from Import Preset or Apply preset next time; Delete removes one.
Start Import → first checks the mapping and the first 500 rows. A required field left on Skip stops you with Map these required fields before starting: Product Code — nothing runs until it is mapped. The check reads the first 500 rows: for products it covers Product Code, Name, Price, List Price, Cost Price, Quantity and Status; for categories the Name; for customers a valid Email; and since 0.2.77 the orders, order items, reviews and waitlist files as well — a valid e-mail, a known order status, numeric money columns, an order number and Product Code that exist, a rating from 1 to 5, a known review status (before 0.2.77 those four types showed their problems only as row errors after the run, and for reviews and order items after the good rows were already appended). Value problems are listed as N row(s) have validation errors with the first five messages, and you are asked Found N validation error(s). Continue anyway? — Cancel to fix the file, OK to run and skip the flagged rows. Since 0.2.78 the run itself re-checks every row, not just the first 500: a row that fails the check — a price with a currency symbol, a non-numeric quantity, an unknown product status, a new product without a name — is not written at all; it is counted as an error and logged (Row N: Price must be numeric.), and the job carries on with the next row. Rows past the first 500 that the dialog never showed you are handled the same way, so read the job log after a Continue anyway run. (On 0.2.77 and earlier only rows the writer itself refused were skipped; a value that merely failed the check was still written, as 0, 0 or active.)
Steps 3 and 4 — Importing and the result
The job runs in the background through WordPress’s scheduled-task system, 100 rows per pass, five seconds between passes, and the page polls it: a progress bar, N% complete (rows read against the row count, so it reaches 100), the running error count and ⏳ Processing in background…. Keep the tab open to watch, but closing it does not stop the job. The result card reads ✅ Import Complete, ⏹ Import Cancelled or ❌ Import Failed with N rows processed · N error(s) and, when there is anything to report, the job log: one line per failed row (Row 12: Product Code is required. — several problems on one row share the line) and one per warning on a row that still succeeded (Row 12 (warning): Image ‘x.jpg’ could not be imported: …, or an unknown tax class). ← Import Another File resets the wizard.
The right-hand Recent Imports card lists your five most recent jobs with a status badge (queued, running, completed, failed, cancelled), the file name and N of M rows processed — processed counts rows written or deliberately skipped, so a clean run reads M of M. A job that wrote anything to its log (row errors, warnings, a cancellation, the time-out note) also carries a Log line you can open there, so a past job’s messages are not lost once the wizard is reset (0.2.77; a clean job has no Log). The list refreshes by itself when a job starts and when one ends. A queued or running job carries a Cancel button: confirm Cancel import products #N? The uploaded file is discarded. and the job turns cancelled, its remaining passes are dropped and the CSV is deleted; rows already written stay. A pass that is in the middle of its 100 rows finishes them first; since 0.2.78 that last pass no longer overwrites the Cancelled by … line or flips the job back to completed (its errors and warnings are appended after it), but the rows it wrote are not added to the N of M rows processed figure, so a job cancelled mid-pass can show fewer rows than it actually wrote. Cancelling from Recent Imports while Step 3 is still on screen ends the wizard with Import Cancelled (0.2.77; earlier versions kept saying Processing in background… until a reload).
The uploaded file is deleted when the job completes, is cancelled, or is cleared by the hourly clean-up. A started job that is still unfinished 24 hours after it started (the scheduler never ran, or PHP died mid-pass) is marked failed by that clean-up, so it cannot block later imports (its Log records Timed out after 24 hours with no progress.). An upload you never started — you chose Cancel on the validation dialog, or left the page — does not appear in Recent Imports at all (0.2.78; earlier versions listed it as queued and then failed), and the same hourly clean-up deletes it with its file once it is a day old. Nothing was written, so there is nothing to do. If the progress stays at 0% for minutes, WordPress’s scheduler is not firing on your host; see Troubleshooting.
4. Products — field reference
Products are found by Product Code and nothing else; there is no matching by name or by ID. A row with an empty code fails. On an update, a column you did not map leaves that field alone, and so does an empty cell — in a number, status or yes/no column, and since 0.2.77 in Description, Short Description, Meta Title and Meta Description too (on 0.2.76 and earlier a mapped but empty text cell cleared the stored text, so on an older store leave those columns unmapped). A new product takes whatever the cell holds, blank included — except the Name, which a new product must have. An existing URL slug is never regenerated (next table). If a file only updates prices, map only the price columns.
| Field (label in Map To) | Values | Notes |
|---|---|---|
Product Code * |
Text | The match key. Two rows with the same code: the second updates what the first wrote. Blank → row error. |
| Name | Text | Required only for a new Product Code (0.2.78): a code that already belongs to a live product may leave Name blank or unmapped, and the row is a normal update. A new code without a Name is listed before the run and skipped during it with Row N: Product name is required for a new product. — no nameless product is created (0.2.77 wrote one). A code that belonged to a deleted product counts as new. A new product’s URL slug is generated from the Name when no URL Slug is given, and from the Product Code when the Name is all symbols. |
| Price | Plain number: 19.99 |
No currency symbol, no thousands separator — $19.99 or 1,299.00 is refused: the row is skipped and logged Price must be numeric. (0.2.78; earlier versions wrote 0). An empty cell on a new product gives 0.00. |
| List Price, Cost Price | Plain number | Checked like Price: a non-numeric value skips the row with List Price must be numeric. / Cost Price must be numeric. (0.2.78; on 0.2.77 a value past the first 500 rows or continued past was written as 0 or 1, and 0.2.76 never checked them). |
| Status | active, disabled, hidden (also A / D / H, enabled, published, draft, 1 / 0, yes / no) |
Anything else skips the row with Status must be active, disabled, or hidden. (0.2.78; earlier versions wrote active). New products default to active. |
| Description / Short Description | Text | Description keeps basic HTML; Short Description and the meta fields are stripped to plain text. |
| Meta Title, Meta Description, URL Slug | Text | A slug you supply is cleaned and written, with -2, -3 added when another live product already has it; a file with just Product Code and URL Slug is enough to change addresses — but unlike the product editor, the importer records no redirect from the old address, so links to it break. Without the column, the slug is generated (from the Name, else the Product Code) only for a product that has none yet — an existing slug is never regenerated, so renaming a product keeps its address. All of this is 0.2.77: earlier versions wrote a supplied slug even when taken, regenerated the slug on every update that carried a Name, and ignored a file that mapped only Product Code and URL Slug. |
| Categories | Parent///Child///Leaf, several separated by | |
The first entry becomes the product’s one main category — on a re-import too, so a different first entry moves the main (0.2.77; earlier versions added a second main beside the old one, and Migration 85 repairs those by keeping the category with the lower id). Each name is looked up under its parent, then anywhere in the tree, and created if not found — a typo makes a new category. Existing links are kept, never removed; new ones are added after them. Secondary Categories uses the same format for extra links and never changes the main. |
| Quantity | Whole number, 0 or more | Negative or decimal skips the row with Quantity must be a whole number. Written even if inventory tracking is off, but only an existing, tracked product counts as a stock movement: dropping into the low-stock band feeds the digest, climbing from 0 fires the waitlist notices (see Inventory and stock control). |
| Track Inventory, Allow Backorders, Free Shipping, Requires Shipping, Downloadable, Featured | 1 / yes / y / true / on or 0 / no / n / false / off (any case) |
An empty cell, or any other word, leaves the flag as it is — on a new product that means off, except Track Inventory and Requires Shipping, which start on. |
| Stock Status | in_stock, out_of_stock, on_backorder |
Anything else → in_stock. |
| Out of Stock Action | none, buy, subscribe |
What the product page offers at zero stock: nothing, buy anyway, or the waitlist form. Anything else → none. |
| Min Qty, Max Qty, Qty Step, Position | Whole numbers | Min and Step are at least 1. |
| Weight, Length, Width, Height, Shipping Freight | Numbers | With Weight Unit (kg, g, lb, oz; else kg) and Dimension Unit (cm, mm, in; else cm). |
| Tax Class | The class’s name as shown under Settings → Tax Rules, or its handle | An unknown name leaves the product’s tax class unchanged and adds a warning line to the job log. |
| Product Type | simple, variable, downloadable, virtual |
Else simple. |
| Visibility | visible, catalog, search, hidden |
Else visible. |
| Search Words | Text, comma-separated | Extra terms the storefront search matches on; the customer never sees them. |
| Images | Full URLs or bare file names, separated by | |
URLs are downloaded into the Media Library; bare names are looked up in the Images Directory. Images are added after the existing ones; the first becomes the main image only if the product had none. A picture that cannot be fetched or found is skipped with a warning in the job log and under Settings → System Logs; the row still succeeds. |
| One field per attribute (named after the attribute) | Value name; several separated by | for multi-value attributes |
Values are matched to the attribute’s existing values by name in your admin language and created when unknown; yes/no attributes take 1 / yes / y / true. Single-value types keep only the first entry. Existing values of that attribute on the product are replaced. |
| Variation Group Code | Text | Give every member of a variation group the same code; the group is created on first use and the members are told apart by their variation attributes (Color, Size…). See Product variations. |
| Variation Set as Default | Y / 1 / yes / true |
Marks that member as the group’s default. |
5. Categories, customers, reviews and waitlist
- Categories — Name is required; Parent Category is matched by name anywhere in the tree (an unknown parent silently makes the row a top-level category); Status takes active, disabled or hidden (or A / D / H, the same shorthands as products); Slug, Description, Meta Title, Meta Description and Position are written. An existing category is matched by name under the same parent and updated with only the cells that are filled — an empty Description or URL Slug leaves the current one alone, so a re-import never churns category URLs; a supplied URL Slug that another live category already has gets
-2(0.2.77). The three strategies apply (section 3): Create new only skips a category that already exists rather than duplicating it. - Customers — Email is required. A new e-mail creates a WordPress login (the e-mail as username, a random password, your site’s default role, no e-mail sent — the customer uses Forgot password?) and the customer record. An e-mail that already belongs to a guest customer record (someone who checked out without an account) is adopted: the login is attached to that record and the order history stays with it. On an existing customer only the filled cells are written (an empty Phone keeps the phone), and the login’s own name is not changed. Company and the address columns fill the customer’s default billing address only when Address 1 is filled; a re-import updates that address rather than adding another, and on that update an empty Company, Address 2, City, State, Postcode or Country cell keeps the saved value (0.2.77; earlier versions blanked it). The customer editor shows this address in its single address card, titled Billing Address for an imported customer since 0.2.78 (earlier versions said Shipping Address whatever the type). A malformed e-mail skips the row (‘x’ is not a valid email address.). Under Create new … only an e-mail that already has a login is skipped (a guest record is still adopted); under the other two strategies new e-mails are still created. A record that cannot be created is a row error (Customer record could not be created for …). See Managing customers.
- Product Reviews — Product Code and Rating (1–5) required, and an unknown Product Code, a rating outside 1–5, an unknown Status or a malformed e-mail is listed before the run (0.2.77) and skips that row during it (0.2.78), so the good rows are appended and the bad ones are not; the reviewer is matched to a customer by e-mail, otherwise the Name / Email given are kept as a guest review; Status is
pending,approved,rejectedorspamin any case (0.2.76 and earlier stored a capitalised Pending as approved), default approved; Date is used when given. Every row is inserted — re-running a file duplicates the reviews. See Product reviews. - Waitlist — Email required and checked (a malformed one skips the row); Product Code (unknown or empty → a general subscriber), Name, Date. One row per e-mail and product; duplicates are ignored.
6. Orders and order items
Orders come in two files. The Orders file needs Order ID (used as the order number — an existing number is matched; a blank one gets a generated number in your store’s own format, TC-… by default) and Email (linked to the customer with that e-mail, otherwise a guest order). Subtotal, Discount, Shipping Cost, Tax Amount and Total are written as given, nothing is recalculated, and a cell that is not a plain number skips the row (Row N: Total must be numeric., 0.2.78) — as does a missing or malformed e-mail; Currency Code is upper-cased and an empty one becomes USD. The billing and shipping columns fill the two addresses; shipping falls back to billing when empty. Customer Notes is kept. There is no date column: imported orders carry the date and time of the import.
Status takes one of TitanCart’s status handles — pending, processing, on_hold, completed, cancelled, refunded, failed, or any handle you added under Sales → Order Statuses — in any case; empty means pending. The single letters older files use are translated (O pending, P processing, C completed, F failed, D and I cancelled, B on hold). Anything else on a row that creates an order (a blank Order ID, or a number not yet in the store) skips the row with Unknown order status ‘…’ — use a status handle such as pending, processing, completed. rather than saving an order you cannot find; the check before the run lists it too, together with a missing or malformed e-mail and non-numeric money columns. On a row that updates an existing order the Status cell is not read at all (0.2.78): an unknown word there is no error and the order keeps its status (0.2.77 flagged those rows needlessly).
When an order already exists, an update touches only its addresses and customer note — totals and status are never changed by a re-import.
The Order Items file then adds the lines: Order ID (the order number, which must exist) and Product Code (which must exist) are required — both are verified before the run (Order ‘TC-…’ does not exist — import the orders file first), as are Quantity (a whole number, 1 or more) and Price (numeric), and since 0.2.78 a row that fails any of those is skipped during the run instead of being appended wrong; Product Name, Quantity and Price fall back to the product’s current name and price. Lines are appended every time the file runs — a second run doubles the lines — and the order’s totals are not recalculated from them.
7. How the import runs, and the gotchas
- 100 rows per pass, five seconds apart. A row that fails is logged and skipped; the next pass starts from the row after it, so a failed row never causes the ones behind it to be read twice. Re-running a corrected file is safe for products, categories and customers (an update; since 0.2.77 an empty cell changes nothing, whatever the column), but not for reviews or order items, which are appended on every run — import only the rows that were never written.
- A job that stops moving can be cancelled from Recent Imports (section 3); one left alone turns failed after 24 hours. Rows already written stay either way.
- Yes/no columns:
1/0oryes/no; a blank cell changes nothing (section 4). - Prices: plain numbers. Price, List Price and Cost Price are checked on every row and a bad one skips the row (0.2.78). The number columns that are not checked — Weight, the dimensions, Shipping Freight, Min / Max Qty, Qty Step, Position — are still cast silently, so
$5in Shipping Freight becomes 0 (and 1 in Min Qty or Qty Step, which are never below 1). - Category names: a single
/inside a name is fine; only///splits levels and only|separates entries, so a name containing|is split in two. - Attribute values are created when they do not exist — Blue and blue may be one value (matching ignores case in most databases) but Blu is a new one. Check Products → Attributes after a big import.
- Images that fail leave a warning line (section 4); read the job log before assuming every picture arrived. Remote URLs are fetched by your server, so the host must allow outbound connections.
- Multi-line cells inflate the row count the job is measured against, so the of M figure in Recent Imports may be higher than your real number of rows; the job still reads to the end and completes.
- HTML is kept in Description only.
- Imports are recorded under Settings → System Logs as Import job started: #N (always) and Import job cancelled: #N (follows the Settings changes toggle); image failures are logged there too (the Products Updated toggle). Exports leave Export link issued: products when you generate one and Export downloaded: products (N rows) when it is opened (0.2.78).
8. Exporting
The Export tab builds a CSV of Products, Categories, Orders, Product Reviews, Waitlist or Customers, filtered by status and date, with a Columns to Include picker; ⬇ Generate CSV puts a ⬇ Download CSV link beside it. A products export writes the main category as a full Parent///Child path in Categories and every other link in Secondary Categories (0.2.77), the format the importer reads; its column labels match the import fields, so the file re-imports with every column pre-mapped (section 2). Files are UTF-8 with a byte-order mark, named products_export_2026-11-18_143000.csv (the time is UTC).
Nothing is stored on the server (0.2.78). Generate CSV does not write a file; it issues a Download link that is valid for 15 minutes, works only for the admin who generated it while signed in, and builds the CSV from live data at the moment it is opened — so a large export takes its time when you click Download, not Generate, and clicking the same link twice within the 15 minutes gives you the current data twice. Anyone else opening the link, or you from another browser or after signing out and back in, gets a short refusal in the browser tab instead of a file (a cookie-check or not-allowed message); after the 15 minutes you see This download link has expired — generate the export again. Generate again when it has expired. On 0.2.77 and earlier the export was a file under your uploads folder that anyone holding the link could open for 24 hours; a store that updates removes those older files by the hourly clean-up once they are a day old.
9. TitanCart Lite and Pro
The importer, the mapping step, Cancel and all seven import types work on both editions. Saved import presets need Pro and the Advanced Products Import extension; without it the preset controls are not shown and the Pro API refuses them. See TitanCart Lite.
Troubleshooting
- The progress bar stays at 0%. WordPress’s scheduled tasks are not running. Visit any page of the storefront in another tab to trigger them, or ask your host to run WP-Cron from a real cron job. The job resumes where it was; if it never does, Cancel it from Recent Imports and upload again once the scheduler works.
- “Map these required fields before starting.” A required column is on Skip in Step 2 — usually Product Code, whose header in your file did not match. Pick it in the Map To drop-down.
- Rows say “Product name is required for a new product” but I am only updating prices. Those codes do not match a live product — a typo in the code, or a product that was deleted. Since 0.2.78 an existing code is never asked for a name, and Continue anyway skips those rows rather than creating anything. Fix the codes, or add a Name column if they really are new.
- Prices came in as 0 or 1. The import ran on 0.2.77 or earlier: a currency symbol or a thousands separator past the first 500 rows, or continued past, was written as 0 or 1 (0.2.76 never checked List Price and Cost Price at all). Since 0.2.78 such a row is skipped and logged instead. Fix the file and re-import with Update existing … only.
- “Cookie check failed” (or another message) when I pick the file. The admin tab had been open too long (its security token had gone stale) or your login session changed, and the upload was refused. Reload the admin page and pick the file again. Since 0.2.78 the message is WordPress’s own; earlier versions said only Upload failed. A network problem shows Upload failed: … with the reason.
- The Download link opens a message instead of a file. It is older than 15 minutes, was generated by a different admin, or you are signed out (section 8). Go back to the Export tab and generate it again.
- My Import Settings were ignored. You changed them after choosing the file; they are sent with the upload. Click ← Import Another File, set them, then pick the file.
- The test import wrote five products and stopped. That is what it does. Untick Perform test import and upload the file again — those five are updated, not duplicated.
- New categories appeared with typos. Categories are created on the fly. Rename or merge them under Products → Categories and fix the file.
- Some images are missing. Open the job log: each failed picture has a Row N (warning): Image … could not be imported line with the reason. Check the URLs are reachable from the server and that bare names are in the Images Directory, then re-import those rows with Delete existing additional images off.
- A product’s tax class did not change. The job log says Tax class ‘…’ not found under Tax Rules: use the class’s exact name from Settings → Tax Rules.
- Reviews or order lines were imported twice. The file was run twice; both types are appended on every run. Delete the duplicates on the product / order and import only the rows that are missing.
- Order rows fail with “Unknown order status”. The value is not a handle from Sales → Order Statuses (or one of the letters O / P / C / F / D / B / I). Fix the column and re-run the failed rows.
- Cancel seemed to do nothing. The job finished between your click and the request; since 0.2.77 the page says so (Only queued or running imports can be cancelled.) and Recent Imports refreshes with the final badge.
- A job turned failed on its own, with no errors. The hourly clean-up timed it out 24 hours after it started (its Log in Recent Imports says so): the scheduler never picked it up, or PHP was cut off mid-pass. Rows written before that are in the store; check Products (or whichever type) and import the remainder as a new job.
- A product shows two main categories, or the wrong one. Before 0.2.77 a re-import whose first Categories entry differed added a second main. Updating runs Migration 85, which keeps one per product (the category with the lower id); re-import the product with the right first entry to choose the other.
- My products’ URL slugs changed after a re-import. 0.2.76 and earlier regenerated the slug from the Name on every update. Since 0.2.77 an existing slug is only changed by a URL Slug column. The importer does not redirect the old addresses (the product editor does); re-import the old slugs, or add redirects under Storefront → SEO → URL Redirects.
- “File exceeds the 10 MB limit.” Split the file; the image URLs do not count, only the CSV itself.
See also
- Migrating from WooCommerce — the table-based migration for a WooCommerce store on the same site.
- Products overview — the fields the import writes, as they appear on the product form.
- Product variations — what a variation group is and how members are told apart.
- Categories, tags and filters — the tree the Categories column builds.
- Inventory and stock control — what a quantity change triggers.
- Managing customers — the records a customers import creates.
- Product reviews — the reviews import in more detail.