Come importare prodotti in Magento 2 in 2026: La guida operativa per 2.4.8 - IT

Adobe ha spedito Magento 2.4.8 ad aprile 8, 2025, e il modo in cui i grandi cataloghi si spostano attraverso la piattaforma è stato finalmente aggiornato 2026. Gli endpoint di importazione REST nativi (/V1/import/csv e /V1/import/json) ora accetta payload con codifica base64 e compressione gzip. That kills the old “split your CSV into 500 rows and pray” routine. The admin CSV importer most stores still lean on? Same tool, soprattutto. But the ways it breaks have moved, and so have the fixes. This guide covers both paths, and the mistakes that eat a weekend when you miss them.

Risposta rapida: For catalogs under 5,000 SKU, use the admin CSV importer under System > Trasferimento dati > Importare. For anything larger or automated, hit the REST /V1/import/csv endpoint added in 2.4.8. Keep files UTF-8 without BOM, drop images into var/import/images, and always run “Controlla i dati” prima di impegnarsi.

Ultima revisione: aprile 2026. Endpoints, version numbers, and import behaviors verified against Adobe Commerce 2.4.8 release notes and the Experience League documentation.

Prima di iniziare: What Changed in 2.4.8

If you’ve imported products on Magento 2.4.5 or earlier, a few defaults caught up with reality. Nuovo 2.4.8 installs ship with indexers in Update by Schedule modalità, so bulk imports no longer grind the frontend to a halt during reindex. The release closed 497 core bugs (several on the importer itself) and added the REST import endpoints that sidestep the 2MB admin upload ceiling.

Here’s what you need in place before touching the importer:

• Magento 2.4.8 o più tardi. Corsa 2.4.5 or older? Upgrade first. The older importer still works, but bug fixes aren’t being backported.
• PHP 8.3 minimo, 8.4 consigliato. 2.4.8 ships with native PHP 8.4 supporto. Stores on PHP 8.1 will import, but error handling is noisier.
• MySQL 8.0 o MariaDB 10.6/11.4. 2.4.9-beta1 drops MySQL 8.0 e MariaDB 10.6, so if you’re setting up fresh, andare 11.4.
• A writable var/import/images directory. permessi: 775 per le directory, 664 per i file, owned by your web user.
• Sufficient PHP limits. For catalogs over 10,000 SKU, memory_limit 2G e tempo_max_esecuzione 1800 in php.ini. Lower values will silently kill imports partway through.

A quick heads-up: Adobe Commerce (the paid tier) and Magento Open Source share the importer code. Cloud customers don’t need to touch php.ini because the stack comes pre-tuned, but the CSV rules and behaviors below apply identically.

What You Can Actually Import

Magento 2 has six product types, and the CSV rules shift depending on which one you’re moving. The short version:

• Simple products. One SKU, one variation. The template every tutorial uses. Import these first if you’re learning.
• Configurable products. A parent SKU with child simples attached via configurable_variations. Size and color t-shirts are the textbook example. Children must exist before the parent import row, or the link fails.
• Virtual products. Servizi, abbonamenti, warranty coverage. No shipping calculation, no weight column needed.
• Grouped products. A wrapper that shows several simples on one page (think “kitchen knife set, buy individually”). Usi associated_skus in the CSV.
• Bundled products. Build-your-own combos where the customer picks options. Format is fiddly and uses bundle_values with pipe-separated fields.
• Downloadable products. File, licenze software, ebooks. Needs downloadable_links configured.

Import order matters. Simples first, then configurables that reference them, then grouped and bundled last. Importing a configurable whose child SKUs don’t exist yet is the most common self-inflicted error, and the validator won’t always catch it before the import runs.

CSV Import: The Admin Walkthrough

The admin CSV importer handles catalogs up to about 5,000 SKUs without complaint. Worth knowing even if you plan to move to the API later, because the column rules that follow apply to both paths.

Passo 1: Export a Sample File (Don’t Skip This)

Vai a Sistema > Trasferimento dati > Importare. Impostato Tipo di entità per Prodotti, quindi fare clic Scarica il file di esempio. The sample matches your exact Magento install, including any custom attributes your store has added. Using a generic CSV template from a blog post is how column-mismatch errors start.

Save the sample, open it in Notepad++ or VS Code. Don’t use Microsoft Excel. Excel adds a Byte Order Mark to UTF-8 files, and the importer rejects BOM-prefixed files without a clear error message. If Excel is your only option, export as “CSV UTF-8” and run the file through a BOM remover, or switch to Google Sheets (File > Scarica > CSV, which saves clean UTF-8).

Passo 2: Fill In Your Product Rows

Required columns for a simple product: sku, store_view_code, attribute_set_code, product_type, categorie, product_websites, nome, descrizione, prezzo, qty, peso, product_online, tax_class_name, visibilità.

Keep descriptions under ~255 characters per row if you’re on the admin importer. Longer HTML descriptions work, but they inflate file size fast and drag the 2MB default limit closer. Categories use slashes to build hierarchy: Default Category/Apparel/T-Shirts. The separator is a forward slash, not a backslash. One relief: Magento auto-creates category paths that don’t exist yet, so you don’t need to pre-build every category before import.

For configurables, il configurable_variations column carries a pipe-delimited string like:
sku=SHIRT-RED-S,size=S,color=Red|sku=SHIRT-RED-M,size=M,color=Red

Fiddly, sì. Save yourself pain by exporting one hand-built configurable first, then using that row as your template for the next 5,000.

Passo 3: Configure the Import Behavior

What happens when a SKU in your CSV already exists in the catalog? Three behavior options decide, and picking the wrong one can erase data:

• Add/Update. The default and the safe pick. Adds new SKUs, updates matching ones, leaves unlisted products alone.
• Sostituire. Deletes matching SKUs first, then recreates them from the CSV. Wipes URLs, recensioni, and stock movement history on those products. Use rarely.
• Delete Entities. Removes matching SKUs from the catalog. Irreversible without a database restore.

Also set Field separator to comma, Multiple value separator to comma, and check Conteggio errori consentiti. Default is 10. Crank it to a higher number if you expect some rows to fail and want the importer to keep going rather than abort.

Passo 4: Validate Before You Import

Upload your CSV, clic Controlla i dati, e aspetta. This runs the same validation the real import does, but without writing to the database. The output lists every row error by number with the specific column that failed. Fix the CSV, re-upload, re-check. Rinse until green.

If validation says “File is valid” but some warnings appear, read them anyway. Warnings are often missing category paths or store views that Magento will silently skip, creating products with incomplete category trees.

Passo 5: Run the Import

Clic Importare. The progress bar is honest for small files and optimistic for large ones. Don’t close the browser tab. Don’t refresh. The importer runs a single long PHP process, and on slower hosting, closing the tab mid-write orphans the process and leaves partial data behind. Watch var/log/import.log over SSH if you want live progress, o controlla Sistema > Import History once it finishes.

Immagini e media

Images confuse more imports than any other field. The rule is simple once you’ve seen it once.

FTP or SFTP your image files into <Magento root>/var/import/images. Subfolders are allowed: var/import/images/2026-spring lavori. In the CSV, reference each image by its path relative to var/import/images:

• base_image colonna: main product image (shown on product page)
• small_image colonna: thumbnail for category pages
• thumbnail_image colonna: cart and order thumbnails
• additional_images colonna: gallery images, comma-separated

A row importing one product with four gallery shots might look like: /2026-spring/tshirt-red-main.jpg in base_image, poi /2026-spring/tshirt-red-back.jpg,/2026-spring/tshirt-red-detail.jpg,/2026-spring/tshirt-red-fabric.jpg in additional_images.

Watch the file naming. Magento treats Shirt.jpg e shirt.jpg as different files on Linux hosts, and the importer doesn’t auto-correct case. Compress to WebP before upload if you can; hero shots at 300-400KB JPEG will tank your Core Web Vitals the moment real traffic hits. I nostri Magento speed playbook covers WebP and AVIF conversion for existing catalogs.

REST API Import: Il 2.4.8 Path for Large Catalogs

The admin importer caps out fast. Once you’re past 10,000 SKUs or you need automated syncs from an ERP, the REST endpoints Adobe shipped in 2.4.8 are the right tool. Two endpoints matter:

• POST /rest/<store_code>/V1/import/csv for CSV payloads
• POST /rest/<store_code>/V1/import/json for JSON payloads

Both accept a base64-encoded string in the request body. For CSV, the rule is: UTF-8 encode your file, gzip-compress it (optional but smart for anything over 1MB), then base64 the result. Response returns a count of entities imported. Async bulk via /V1/bulk/ works too if you want fire-and-forget.

The flow, in order:

• Authenticate with an admin token (POST to /V1/integration/admin/token)
• Gzip your products.csv, base64-encode it, stuff it into a JSON payload with fonte, entity: “catalog_product”, e comportamento: “append”
• POST to the import endpoint with the token in the Authorization header
• Poll /V1/bulk/{bulkUuid}/stato if you went async

Adobe’s developer docs carry the exact payload shape and update per release, so pull from there rather than copying old StackOverflow threads. Il 2.4.8 payoff: the async path doesn’t hold a PHP request open for up to 30 minutes the way the admin importer does. Your 50,000-row load runs in the background, and the storefront stays responsive to actual shoppers.

Common Import Errors and What Actually Fixes Them

These are the failures that burn the most time. Each one has a specific root cause that isn’t always obvious from the error text.

“Allowed memory size of X bytes exhausted”

PHP ran out of memory mid-import. Raise memory_limit in php.ini to 2G, restart PHP-FPM, retry. If you can’t touch php.ini (hosting condiviso), split the CSV into files of 1,000-5,000 rows each and import sequentially. Third-party extensions like Firebear and Amasty add an “Import without Memory Limit” toggle, but that’s an extension feature, not native Magento.

“Maximum execution time of 30 seconds exceeded”

Your PHP tempo_max_esecuzione è troppo basso. Vai a 1800 secondi. On Nginx, you’ll also need to raise fastcgi_read_timeout to match, otherwise Nginx kills the upstream before PHP finishes.

“Wrong columns sequence” o “Header has invalid columns”

Byte Order Mark. Your CSV was saved from Excel, which prepends a BOM to UTF-8 files that the importer reads as a garbled first column. Re-export from Google Sheets or re-save through Notepad++ as UTF-8 without BOM. If you only check one thing before hitting import, check this.

“Invalid attribute set” o “Invalid attribute value for X”

Your CSV references an attribute set or dropdown value that doesn’t exist in this Magento install. Controllo Stores > Attributes. Dropdown values are case-sensitive and whitespace-sensitive. “Red ” with a trailing space is not “Red”.

Importare “completes” but products don’t appear on frontend

Reindex didn’t run, or cache is stale. SSH in and run bin/magento indexer:reindex seguito da bin/magento cache:flush. Su 2.4.8 with Update by Schedule indexers this is rarer, but still happens with product-URL and stock indexers.

“URL key for specified store already exists”

You’re importing a product whose url_key collides with an existing one. Either change the URL key in your CSV or set behavior to “Add/Update” rather than a fresh “Append Complex Data” correre.

Hosting Requirements for Big Imports

Shared hosting kills more Magento imports than bad CSVs do. If you’re running a production 2.4.8 memorizzare, the honest floor is 4 GB di RAM e 2-4 CPU virtuali; real catalogs (10k+ SKUs, frequent reimports) Bisogno 8-16 GB. Disk should be NVMe SSD, not spinning drives or old SATA SSDs, because reindex and image-resize ops hammer random I/O.

Specific things a Magento-ready host should let you do:

• Edit php.ini (or per-domain equivalent) to raise memory_limit and execution time
• Install and run cron (Magento’s async bulk import uses the consumers queue)
• SSH access for bin/magento CLI commands
• Enough disk headroom that var/import/images can hold your largest batch twice over (Magento copies, not moves)

If your current host can’t deliver those, the importer isn’t your real problem. I nostri Magento VPS hosting roundup breaks down which providers ship a 2.4.8-ready stack without surcharges, e il strumento di ricerca dell'hosting filters by PHP version, RAM, and control panel if you want to compare plans side by side. For merchants running WooCommerce alongside Magento (it happens more than you’d think), il Guida all'hosting e-commerce WordPress covers the crossover stack.

Domande frequenti

What’s the maximum CSV file size I can import in Magento 2.4.8?

The admin importer caps uploads at 2MB by default, controlled by PHP’s upload_max_filesize e post_max_size. Raise both to 64M in php.ini for larger files. For anything past 50MB compressed, switch to the REST /V1/import/csv endpoint, which handles gzip’d payloads and doesn’t bottleneck on browser upload.

Can I import products in Magento 2 without using CSV?

sì. Magento 2.4.8’s REST API accepts both CSV and JSON via /V1/import/json. Third-party extensions (Firebear, Amasty, Magmi for older installs) support XML, Google Sheets sync, and direct ERP connectors. Native admin still requires CSV, so pick your path based on automation needs.

How long does a 10,000-product import take?

On a tuned 2.4.8 pila (8 GB di RAM, NVMe, PHP 8.4, MariaDB 11.4), aspettarsi 15-25 minuti for a 10k-row import with images. Shared hosting pushes this to 60+ minutes if it completes at all. The async REST bulk path runs in the background, so shopper-facing pages stay responsive during the import.

Why does my import succeed but show wrong prices or stock?

Two usual causes. Primo, your CSV has trailing whitespace or European comma-decimal formatting (1,99 invece di 1.99) in the price column; Magento parses these as invalid and keeps the old value. Secondo, multi-source inventory (MSI) in 2.4.8 stores qty per source, not per product globally, so you need the source_code column or the import updates the default source only.

Do I need to disable Magento flat catalog before importing?

Flat catalog was deprecated in 2.3 and removed from default workflows by 2.4.5. If your store somehow still has it enabled, disable it under Stores > Configurazione > Catalogare > Storefront before large imports. It was always a performance liability and 2.4.8 makes this explicit in the admin.

Can I automate product imports from an ERP or PIM?

That’s exactly what the 2.4.8 REST endpoints are for. Tools like Akeneo, Pimcore, and Salsify have native Magento 2 connectors. For custom builds, cron a script that pulls your ERP feed, transforms to CSV, gzips, and POSTs to /V1/import/csv. Schedule for off-peak hours if the import is heavy.

Final Notes

Two rules decide whether your next import works. Uno: save the CSV as UTF-8 without BOM. Sempre. Two: raise PHP memory_limit e tempo_max_esecuzione prima che inizi, not after the first crash. Everything else is detail.

Path-wise, the call is simple. Sotto 5,000 SKUs and running imports occasionally? Use the admin CSV tool. Past 10,000 SKUs or syncing from an ERP nightly? Use the REST /V1/import/csv endpoint added in 2.4.8 and stop babysitting the process. Back up the database before any Replace or Delete run. Never close the browser tab mid-import. And if the stack underneath can’t hold up, no importer tactic rescues it. Fix the host first.

Per letture correlate: il Magento theme roundup covers frontend performance after your catalog is loaded, the speed playbook linked above handles the caching stack, and the hosting guides cover what to run it all on. The boring infrastructure choices compound faster than any single feature flag.