Magento 2 Google Shopping Feed

View Demo

Create a feed file with the Google Shopping extension for Magento and get more orders while increasing your revenue.

The Google Shopping is a go-to sales channel for online stores to increase their revenue and customer base. Roughly 1 billion shopping sessions a day are conducted on this platform, according to Google.

Sell directly from the Google Search engine by placing your products on its dedicated Google Shopping tab. Get new audience by advertising your products in Google Ads on hundreds of websites.

  • Generate a fully compliant Magento 2 Google Shopping feed
  • Keep your feed always up to date with an automatic update option
  • Filter out products to be excluded from the Google Shopping schema
  • Easily configure what product attributes to include in the feed
  • See your feed success with its clicks, orders, and revenue statistics
  • Start with a Google-ready preset, copy it for related campaigns, or build a feed from scratch
  • Run pre-flight checks before upload to Google Merchant Center

ℹ️ Important: This sub-module is included in the Advanced Product Feeds extension.

Buying this item gives you access to all feeds including Facebook, Instagram, Google Shopping, and others in one powerful package.

Magento Cloud
Compatible with:
Community:
2.3.* - 2.4.9
Enterprise:
2.3.* - 2.4.9

Business Value

Google Shopping is a proven sales channel to get you more orders. Roughly 46% of product searches begin on Google, and Google Shopping accounts for 36% of product discovery searches. Almost 35% of Google shoppers make a purchase within 5 days of searching for a product.

  • Problem: Presenting your product catalog on Google’s comparison shopping engine means you need to format data on your products within Google’s XML feed specification and taxonomy. Google also expects this info to be automatically updated.
  • Solution: Magento Google Shopping extension easily makes a product feed that is fully compliant with Google’s specification and automatically updates the feed. It tracks the feed business success and serves best for your business goals.

How Google Shopping feed drives your sales

Using the Google Shopping Magento extension to sell your products directly on Google, you can expect rapid sales growth. Here are three main reasons for this:

Match with shopper's interests

Products on Google Shopping automatically match relevant keywords and search queries. Shoppers will see items they are potentially interested in buying.

Reach wider audience

Google is the most popular search engine, serving millions of search queries per day. By selling on Google, you can reach a wider audience in a short time period.

One feed for multiple platforms

A Google Shopping feed can be used for selling on other sales channels. Facebook, Microsoft Bing, and dozens of other sales engines accept this feed.

Fully compliant with Google feed specification

Google Shopping extension generates a product feed that can pass the check by Google Shopping from the start. The shopping feed Google expects may have such information on products:

  • Basic product data
  • Price and availability
  • Product category
  • Product identifiers
  • Detailed product description
  • Shopping campaigns and other configurations
  • Destinations
  • Shipping
  • Tax

Some of the product info, like price and availability or product category, is mandatory. The corresponding attributes are included in the default feed template when you create it in the Magento Google shopping feed module.

To stand out your products from competitors, let your potential shoppers to fully see your products. For that purpose, include optional product data, such as additional images, expiration date, or mobile-optimized landing page. Google shopping extensions' capabilities allow you to present goods in the best way possible.

Two steps to get a feed

To speed setup, the builder opens on a start screen with popular destinations, an A–Z index for quick filtering, and instant search to help you pick the right preset.

Using a prebuilt template for Google Shopping, you can get the feed literally in a minute. Make a product feed in two steps:

  • Provide a feed file name
  • Click the Generate button

Then, simply copy and paste the feed file URL into your Google Merchant account. Alternatively, configure the FTP/SFTP feed file upload to Google if it suits you better.

However, typically, the Google Shopping product feed requires more than two steps as you may want to make it more personal. To stand out from competitors, you can edit the default feed template to change its output. For example, customize the template to add more product information and apply product filters to filter out irrelevant items.

For similar destinations, you can duplicate a preset, and both presets and feeds support helper metadata (logo, guide link, short note) for clearer organization.

Additionally, configure e-mail notifications regarding the feed status and reporting on the user statistics. Attach your UTM tags to track your feed access in Google Analytics.

Product Attributes

Configure your Magento Google product feed easily in a visual manner, no experience of working with XML required.

Simply click the Library of patterns and find a product attribute you want to include in your feed. Use the Preview function to get a view on the output the pattern delivers. Then copy the pattern and paste it in your feed template.

Magento Google product feed

Product Filters

Make your Google Shopping feed more effective by filling it with products that meet your business interests best.

For example, Google requires to exclude out of stock products otherwise it will disapprove the product. Additionally, you may want to promote only products of a specific price range.

Apply the Base Product Filter or Extended Filter or both to your product feed to filter out irrelevant products from your feedt.

Magento Google product feed

Feed Preview

Magento Google product feed

Make sure you get a perfect XML feed before it is uploaded to Google. Use the preview option to correct any misconfiguration that may occur when editing the feed default template.

You will see the output exactly as it will be received by Google Shopping. Check for possibly deleted mandatory attributes, tags have opening and closing elements, etc.

Having such a preview tool helps to save time by avoiding a rejection when the feed is uploaded to Google.

Feed Validation

Catch issues before export and avoid rejections. Validate any feed against your own rules and get a clear report before upload.

Set checks in Content Settings > Validation Rule, generate the feed, and run validation. If a rule is present, it can run automatically; you can also start it manually at any time.

Typical checks include title length, required fields, character encoding, price and currency formats, and dimensions, etc. Reports show line numbers and plain-language messages, so you can fix values, regenerate, and validate again.

Magento Google product feed

How to sell on Google Shopping with a product feed

To start selling on Google Shopping, Google Ads, and other Google services, the first thing you need to do is to create a product feed file. This file contains all information on your products you want to sell via Google services.

Next, create a Google Merchant account and configure the access to the feed file. You may set up Google Merchant to fetch your feed by the URL or FTP/SFTP.

The last step is to finish the setup of your Merchant account by providing such business info as shipping, taxes, branding, returns.

Finally, you can publish your product listings to start selling on Google.

Magento Google product feed
No hidden fees
Lifetime access to source code
Access to free support and updates for 1 year
Updates and support prolongation - $108

Pay today $179 for the first year.

Then $108 for updates and support services per year.

Cancel anytime.

30 days money back guarantee
See it in action!
Pick a quick tutorial to learn about various aspects of this extension
Customer Reviews 0
Earn points for your review about this extension modules. $1 = 10 points
Write Your Own Review

check-circle You submitted your review for moderation.

Manual & Support
Need more help?

Save time by starting your support request online and we'll connect you to an expert.

Changelog
Version 1.15.2Aug 12, 2026
JSONL — a new feed format for data pipelines — pick JSONL as the feed type and every product is exported as one JSON object per line, instead of a CSV row or an XML node. You build it in the same column editor you already use for CSV: each column's header becomes the key, and its value comes from an attribute or a pattern exactly as before. Values are escaped as valid JSON, so a quote or a line break in a description can no longer break the line. A column set to All Attributes emits every attribute the product has, keyed by the attribute's store label. This is the format most warehouses and ingestion tools (BigQuery, Snowflake, marketplace bulk APIs) expect, and unlike a single large JSON file it can be read line by line, so a partner never has to load the whole catalog into memory to process it.
Compress the generated file (gzip) — a Compress output (gzip) switch on the feed's General tab, available for CSV, TXT and JSONL feeds. The finished file is packed and published as, for example, products.csv.gz, which for a typical catalog is several times smaller — faster to upload, cheaper to transfer, and accepted directly by most channels and data pipelines. Compression is streamed, so turning it on does not increase the memory a large export needs. Quality-control validation still checks the real published file: the archive is unpacked before the rules run, including on stores that keep media on remote storage such as AWS S3. An export that produced no products yields a valid empty archive rather than an unreadable zero-byte file. Note: enabling this changes the feed's filename and therefore its URL — update anything that fetches the previous address.
⚠️ The feeds table gains an is_compressed column, added automatically by setup:upgrade.
Version 1.15.1Aug 11, 2026
Real numbers, booleans and nulls in the JSON output of an XML feed — until now every value in the generated .json arrived as a string, a field with one nested element became a single object where two made a list, and an empty value simply disappeared. A partner asking for a price as a number or a stock flag as true/false could not be served by any template. You now say what a value is where you write it, the same way an XML schema does: mark the element xsi:type="xs:decimal", xs:integer or xs:boolean and it is converted, mark it xsi:nil="true" and it comes out as null, mark it xsi:type="array" and it is always a list even when the product has only one of them. Nothing is guessed from the value itself, so an SKU of 0012333 keeps its leading zeros and a text attribute reading "true" stays text. A value that does not match the type you declared is left as a string rather than turned into 0 or false, so a typo in a template cannot silently corrupt a feed. Feeds that declare nothing produce exactly the same JSON as before.
JSON tab in the feed preview — with Generate JSON enabled, the preview now shows the JSON alongside the XML, so the output can be checked while the template is being written instead of after a full export. If the JSON cannot be produced, the preview shows why, on the same screen where the template is fixed.
A template whose output is a series of top-level elements rather than one enclosing element is now converted to JSON as well — previously the .json for such a feed could not be produced at all.
Templates that use prefixed tags such as <g:id> without declaring the prefix on the root — which is what shopping channels expect — no longer produce parser warnings during export.
When the generated XML was broken (a truncated file, unbalanced tags, an empty result), the export finished quietly and left the previous, now stale .json published next to the new .xml. The export now stops with a message naming the tag and the line in your template output that caused it.
A value declared as a decimal ships in its shortest form — a price formatted as 9999.00 becomes 9999 — while a genuine fraction such as 34.50 keeps it, and a number too large for the platform's integer range stays text instead of being replaced by a wrong one.
⚠️ Upgrading from an older version could stop with a fatal error when a stored feed validation schema was not parseable on its own; such records are now skipped and the upgrade completes.
Version 1.15.0Jul 31, 2026
AI-assisted feed optimization — Your catalog was written for your product pages; shopping channels want something else. That gap is why Merchant Center disapproves items, why variants collapse into one listing, and why products miss the searches they should win. Point an AI profile at a field and the extension fills what is missing:
Two new console commands, for stores that automate their feed pipeline:
show the group count on a preview
delete a stored value from the results grid
generate a batch on demand from the AI results grid
working preview, coverage in the grid, and no one-option questions
shipped AI profiles follow Google's title guidance and see the whole product
results grid, approve gate, real preview, spend surface, Validator hand-off (pm[#844]())
category-mapping executor with cross-run phrase cache (pm[#841]())
batched text-rewrite and attribute-enrichment executors (pm[#840]())
AI optimization run model — orchestrator, drip cron, CLI catch-up (pm[#839]())
AI export seam — generation reads stored optimized values (pm[#842]())
merge PROP-0218 branch (pm[#838]() engine) into S7
merge S7 AI profile library admin (pm[#843]())
AI profile library admin (pm[#843]() / PROP-0218 S7)
merge AI engine foundation slice (pm[#838]()) into PROP-0218 branch
AI engine foundation services for feed optimization (pm[#838]())
AI feed optimization data model (S1)
AWS S3 / remote media storage support — Feeds now generate, validate, and display correctly on stores that keep media on a remote backend such as AWS S3. Previously the feed logo was reported as missing, feed generation could fail outright, the public feed URL pointed at a private bucket path that browsers couldn't open, and quality-control validation of the generated file failed. Now logos are detected on the remote, the feed is built locally and only the finished file is stored remotely, the feed URL is a proper browsable media URL, and validation reads the file through the storage backend so it works even on private buckets.
The Advanced Product Feeds admin menu no longer lists every feed you have created. On a store with a few dozen feeds the menu grew long enough to push the section links off the screen; the Feeds grid reaches any one of them in a click, with the search and filters the menu cannot offer.
Button labels across the extension's admin screens now read in Title Case, and every button that creates something is phrased the same way — Add New Feed, Add New Template, Add New Filter, Add New Profile, Add New Attribute, Add New Category Mapping, Add New Variable. Previously these six differed from one another, and labels such as Clear values or Test connection were cased differently again. Field and column labels are unchanged.
Nine categories in the shipped Google product taxonomy had corrupted characters (for example Piñatas instead of Piñatas, Sauté Pans instead of Sauté Pans). They appeared mis-spelled in the Category Mapping autocomplete and could not be selected reliably; the taxonomy file now uses the correct encoding.
The Feed Templates grid now honours the admin role restriction that guards the page it lives on. The grid's own data and export routes were not covered by that check, so an admin role without access to the feed section could still read or export the list by calling them directly.
The Delete button was missing from the Filters form, so a filter could only be deleted from the grid. Its Save button also offered no Save & Close, unlike every other form in the extension — the form now carries the same split Save, both while creating a filter and while editing one.
give the AI profile and filter forms the Save split button they declared
satisfy phpstan, and assert what the two AI acceptance tests can actually observe
say that assigning a profile to a feed rewrites the feed's own attributes
say which ai: tokens a feed can use, and stop hiding half of them
drive Save through the form adapter, which is what actually binds
a plain Save click works again, and a state is just its badge
preview lists every attribute asked about, answered or not
preview's Original column covers every attribute the profile fills
declare skipped in the run-totals return shape
set the prompt placeholder through config, not a settings element
recognise Enter by key as well as keyCode
give a typed attribute name the option fields the list renders
one label per attribute field, and pick a text rewrite's target too
say when a profile is generating for a feed that never reads it
eight defects the QA pass found in AI optimization
bring the unit suite back to green
stop offering category mapping the pacing it ignores
preview costs what the launch it previews would cost
preview an enrichment profile before saving it
an ETA that counts the work a launch does for free
make the coverage bar's two numbers add up
the Tokens column no longer multiplies enrichment spend by its outputs
the Original column shows the attribute the row is about
stop recording "the product does not say" as a failure
an enrichment column no longer emits a different attribute
do not treat the filters placeholder as a filter
stop the validation report's filters leaking between feeds
send a missing profile id back to the library, not to Edit
let the results-grid row menu out of the grid wrapper
sort the numbered attribute groups naturally
refuse a batch size the executor would only clamp
make the AI preview endpoint POST-only
run the assignment checks on mass enable
offer Fix with AI on an XML feed's fields
a wrong field count took feed generation down with a 500
record a fill-mode skip, or the profile never converges
stop the results-grid filters leaking between profiles
save the inline override the results grid actually posts
point the declared State filter at the filters container
give the State filter its options
show a value held for approval in the results grid
keep a results-grid mass action on the grid, and let select-all mean it
gate the AI and template grids at the dataSource, not just the page
cast the serializer result so the json filter keeps its string return
connection() must not native-type its return, or the resource tests TypeError
the AI drip was dead on a stock MySQL, and a retarget generated nothing
the split-button option ids collided with the UI form's own save selectors
make the rewrite reply-shape a constant so the scanner stops asking for __()
satisfy the phpstan and translation-scan gates
point ParamsExportCest at the labels this branch renamed
the AI templates must escape through $block — $escaper does not exist on 2.3
the smaller findings — context leak, GET-triggered bulk delete, escaping, store labels
re-optimize has to blank the stored hash, or it queues nothing
bound a launch by its wall clock, and stop bisecting failures that are not the reply's
one insertOnDuplicate per column set, so a mixed portion cannot lose values
one Save control on the profile form, and one place that resolves scope
Save and Delete on the AI profile form, and show a launch in progress
make the AI screens usable — three groups, real defaults, honest grid
sentence-case admin labels across every screen, not just the AI ones
admin labels use sentence case, per the field-label convention
make the profile chooser actually render, and stop pre-selecting Blank
split AI profile creation into choose-then-form, like a new feed
place AI Optimization in both menus, and give its screens their stylesheets
AI assignment substitution must cover method-backed attributes (pm[#842]())
one-line notices on Is active and Code, none on Apply to feeds
explain Is active, Code and Apply to feeds on the profile form
AI-assisted feed optimization. A library of named AI profiles that rewrite or enrich selected feed fields — product titles, descriptions, Google category mapping, and structured attributes such as colour, size and material — using the AI provider you already configured for Mirasvit Core. Optimization runs in the background and its results are stored, so feed generation reads ready-made values and never waits on an AI call: your exports stay exactly as fast as before. Assign a profile to your feeds and every reference to the target attribute picks up the optimized value with no change to the feed itself, or reference ai:<code> on a single column when you want one field optimized and another left alone. Profiles preview on a handful of real products before you enable them, report the tokens they have used and the tokens still needed, and can hold generated values for your approval before any of them reaches a live feed. Where a profile is switched off, or has no value for a product yet, the feed emits the original value exactly as it does today.
A Fix with AI action on the feed validation report, which turns a flagged missing or invalid field into a prefilled optimization profile in one step.
A console command, mirasvit:feed:ai:optimize, for covering a large catalog on demand instead of waiting for the background schedule — useful before a campaign launch.
Button labels across the extension's admin screens now read in Title Case, and every button that creates something is phrased the same way — Add New Feed, Add New Template, Add New Filter, Add New Profile, Add New Attribute, Add New Category Mapping, Add New Variable. Previously these six differed from one another, and labels such as Clear values or Test connection were cased differently again. Field and column labels are unchanged.
Saving an AI profile that no feed uses yet no longer shows a warning about spending tokens without changing feed output. Generating ahead of wiring a feed up is an ordinary step, and the profile screen already states that it covers every store view until a feed narrows it, and shows the tokens spent.
Nine categories in the shipped Google product taxonomy had corrupted characters (for example Piñatas instead of Piñatas, Sauté Pans instead of Sauté Pans). They appeared mis-spelled in the Category Mapping autocomplete and could not be selected reliably; the taxonomy file now uses the correct encoding.
The Delete button was missing from the Filters form, so a filter could only be deleted from the grid. The Save button on the Filters and AI profile forms also offered no Save & Close, unlike every other form in the extension — both forms now carry the same split Save, and it is available while creating a record as well as while editing one.
Version 1.14.9Jul 24, 2026
The Test Connection button no longer uploads anything to your FTP/SFTP destination. It used to place a deliverer.txt file there — actually a copy of one of the module's own PHP files — which exposed source code and, on stores whose remote folder is watched by a feed consumer (such as OpenAI Ads Manager), was picked up and ingested as a bogus feed. The test now only verifies that the server is reachable, the credentials work, and the destination path exists, leaving the folder untouched. Failed tests also report the reason more clearly, distinguishing a wrong SFTP login from an invalid SFTP path.
CSV and tab-delimited feeds no longer lose the last row's empty trailing columns. If the final product in the feed had no value for the last few columns, those columns were trimmed away, so the closing row had fewer columns than the header — and marketplaces that read values by column position misread or rejected that product. Every row now has the full set of columns.
Feed generation no longer fails for products with an empty list field. When a template used the first or last filter on something a product had no values for (for example a product with no gallery images), the export aborted with a fatal error; such fields now simply come out empty and the feed generates to the end.
Test Connection for FTP/SFTP no longer leaves any file on your destination folder. The button used to upload the module's own source code as a deliverer.txt file to the remote directory — on stores whose remote folder is watched by a feed consumer, that stray file could be ingested as a bogus feed, and it exposed PHP source. Test Connection now only checks that it can reach, sign in to, and open the configured folder, without writing anything to it.
Fixed the Indonesian "Last %1 days" date-range label, which showed a literal %1 instead of the number of days because of a stray space in the translation.
AWS S3 / remote media storage support — Feeds now generate, validate, and display correctly on stores that keep media on a remote backend such as AWS S3. Previously the feed logo was reported as missing, feed generation could fail outright, the public feed URL pointed at a private bucket path that browsers couldn't open, and quality-control validation of the generated file failed. Now logos are detected on the remote, the feed is built locally and only the finished file is stored remotely, the feed URL is a proper browsable media URL, and validation reads the file through the storage backend so it works even on private buckets.
Version 1.14.8Jul 23, 2026
Testing an FTP/SFTP feed connection no longer leaves a leftover file behind. The Test Connection button used to upload a deliverer.txt file to the destination folder and never remove it; it now writes a temporary file only to confirm write access and deletes it once the test passes, so the target directory stays clean.
Reasons to choose Mirasvit
Client focusing and satisfaction

These are our primary. A major portion of our new clients come from referrals from our existing clients. Our professional team of developers, marketers and support staff have invested the best knowledge and experience in the field into our work, so you know you can come back to us again and again.

Remarkable support

One year free and high quality support. We go to great lengths to provide maximum satisfaction with every module you have purchased in our store. By helping you with installation, configuration, answering your every question, we do all our best to eliminate any possible problems.

Risk-free Investment

30-days money back guarantee. If you are not satisfied with our extension performance for any reason, we provide a full refund.

Constant improvements and upgrades

We constantly add new features to all our modules, and are always interested in hearing your opinion and implementing your suggested features in our future developments.

Comprehensive Documentation

We provide an expanded user guide for every aspect of our extension, so you can find answers for all your burning questions.

Unencrypted source code of our products

You can customize extension according to your needs and requirements.

Usability and Performance

The Module is easy to install and upgrade, just follow our step-by-step user guide.

Ready for Magento Cloud

No core modifications. The extension has been tested in a Magento Cloud environment and is fully compatible with it.

Loading...