Product feed specifications
What are product feed specifications?
Product feed specifications are standardized data file formats that advertising, search, and shopping platforms like Google Shopping, Meta, TikTok, ChatGPT, and Perplexity require to list and advertise products. They dictate the precise file types, structures, attributes, and logic a channel needs to display products accurately when a user searches for them.
What do product feed specifications actually include?
In general, a product feed specification encompasses the following key elements:
- Required attributes: id, title, description, price, and availability
- Accepted field values: Required values can differ by channel, such as in_stock on Google Shopping versus in stock on Meta Commerce Manager for the same availability status
- Supported file formats: CSV, TSV, TXT, XML, or JSON
- Encoding and structure rules: UTF-8 encoding, a single header row, and one product or variant per row
- Validation logic: Character limits, image resolution, prohibited content, and multimedia attributes
Explore all 2500+ Productsup integrations 👉
Why product feed specifications matter for sellers and channels
83% of shoppers said they would leave a product page that doesn’t give them the information they want. Similarly, 80% agreed that inaccurate content damages their perception of the brand behind it. Therefore, adhering to feed specifications does more than reduce technical errors.
It reduces disapproval, keeps price and availability synchronized, and gives AI-led channels such as Gemini the product facts they rely on when citing or comparing items. Here are the key benefits of feed specifications:
For sellers: Better visibility and feed performance
- Reduce algorithmic rejection to near-zero
- Improve visibility across AI shopping assistants and recommendation engines
- Increase conversion rates and customer trust
For commerce platforms: Better data quality and shopping experiences
- Accurate feeds give ranking algorithms reliable signals to work with, while poor-quality data can lead to less accurate predictions.
- Stock mismatches and price discrepancies impact user experience and platform credibility
- Unoptimized data at scale causes irrelevant search results, reducing shopper engagement
- Lower user satisfaction drives the advertiser budget away, creating a downward spiral
Ultimately, enforcing quality checks through mandatory feed parameters protects both algorithmic integrity and the entire commerce ecosystem. Without consistent feed standards, sellers may see lower ROAS, platforms risk losing advertiser confidence, and shoppers are more likely to encounter inaccurate or irrelevant product results.
Core attributes every product feed specification requires
The following core attributes help you, the seller, and the brands to form the product’s commercial identity. Therefore, maintain them centrally and treat them as the authoritative source of product truth before product content syndication begins and is mapped into channels like Google Merchant Center or Amazon.
| Category | Required attributes |
|---|---|
| Identity | [id], [sku], [brand], [gtin], [mpn], [manufacturer], [model_number] |
| Content | [title], [description], [key_features], [specifications] |
| Commerce | [price], [currency], [availability], [condition] |
| Destination and media | [link], [canonical_url], [image_link], [additional_image_link] |
| Classification | [product_type], [category] |
| Variants | [item_group_id], [variant], [color], [size] |
| Physical product data | [material], [dimensions], [weight], [package_dimensions], [shipping_weight] |
| Market and fulfilment | [country], [language], [quantity], [shipping_cost], [shipping_time], [return_policy] |
| Compliance | [country_of_origin], [compliance/safety_information] |
*Disclaimer: The precise attribute requirements for product feeds vary by channel and product category. Always refer to the channel's official requirements.
*
Master Google Merchant Center product feeds
Learn how to structure, optimize, and troubleshoot your product data for Google Merchant Center with actionable rulebox examples and best practices.
File formats and technical requirements for product feed specifications
Most marketing and advertising channels accept product feeds in these four formats. Each format serves a different purpose and platform combination.
| Format | Accepted by |
|---|---|
| CSV | Meta Commerce Manager, TikTok Shop |
| TSV | Google Merchant Center, Meta Commerce Manager |
| XML | Google Merchant Center, Meta Commerce Manager, Criteo |
| JSON | OpenAI Commerce feeds |
In terms of technical requirements, most marketplaces and advertising platforms now prefer files in UTF-8 encoding (8-bit Unicode Transformation Format).
Product feed formats are changing
While traditional channels like Google Merchant Center don't natively prioritize standard JSON, modern endpoints and platforms, such as OpenAI's commerce feed, use JSON Lines or compressed JSON formats for bulk data transfers.
Specifically, OpenAI prefers Parquet, a columnar format optimized for large datasets, for its native commerce feed. It also supports gzip-compressed JSON Lines (.jsonl.gz), CSV (.csv.gz), and TSV (.tsv.gz) for file uploads via SFTP (Secure File Transfer Protocol).
Fixes for common feed specification errors
Most disapprovals across Google Merchant Center, Meta Commerce Manager, Amazon, and OpenAI fall into a small set of repeatable patterns. The fix is usually a single rule in your feed mapping layer or a one-time catalog cleanup.
| Channel | Common error | Fix |
|---|---|---|
| Google Merchant Center | Image below 500 x 500 pixels | Re-render primary and secondary images at 1,500 x 1,500 or above before the January 31, 2027 enforcement date. |
| Price or availability mismatch with landing page | Align feed, landing page, and structured data, then resubmit. | |
| GTIN not related to brand | Use a valid GTIN, or set identifier_exists to no when no GTIN or MPN exists. |
|
| Meta Commerce Manager | One invalid field rejects the whole row | Remove unused deep-link columns or repair values, then verify ingested count matches source. |
| Invalid price format | Send 89.99 USD with a space and ISO code, not 89.99 or USD89.99. |
|
| Image too small or policy-violating | Replace with compliant images and strip health or efficacy claims from copy. | |
| Amazon marketplaces | Legacy pre-2026 inventory template | Regenerate the template from Add Products via Upload and migrate data. |
| Modified first row breaks parsing. | Never edit row one; download a fresh template if corrupted. | |
| Excel strips leading zeros from UPCs | Format the product ID column as text before entering UPCs. |
|
| OpenAI / ChatGPT commerce | Inconsistent eligibility flags | Set is_eligible_search to true before enabling checkout, and supply seller_privacy_policy and seller_tos URLs. |
| JSON or XML instead of delimited format | Switch to Parquet, or gzip-compressed .jsonl, .csv, or .tsv, encoded in UTF-8. |
The future of feed specifications is AI-ready product data
The way shoppers discover, compare, and purchase products is changing rapidly. More people are relying on conversational AI platforms like ChatGPT, Gemini, and Perplexity to buy their next product. AI-driven retail traffic to US websites increased 138 percent year-over-year during the first half of 2026. AI-referred visitors also converted at a 31 percent higher rate than visitors from other sources.
However, this does not replace traditional feed requirements; instead, it expands them and makes an AI-ready product feed specification a non-negotiable.
Think of it this way:
- A standard feed tells a channel what a product is.
- An AI-ready feed helps an assistant explain why it fits the shopper’s request.
For example, “Trail Shoe Waterproof” gives an assistant too little to work with. “Men’s trail running shoe with a waterproof membrane, rock plate, 8 mm drop, and traction designed for technical terrain” supports a far more useful response to a shopper asking about waterproof shoes for rocky trails.
Productsup AI Enrich helps teams add that missing context across enterprise catalogs. We’re processing 2 trillion-plus SKUs every month and enriching them with agentic attributes such as:
- Product highlights
- Use cases
- Q&A pair
- Occasion tags
- Competitive differentiators
The goal is to create semantic content for every SKU that gives AI systems reliable information they can interpret when product discovery starts with a question.
Book a demo to discover how seamless it is to maintain product feed specifications for traditional and agentic AI shopping platforms. You only need to drag and drop; Productsup manages the rest!
You may also be interested in learning more about:
FAQs
Yes. Google Merchant Center, Meta Commerce Manager, and ChatGPT each enforce their own attribute names, value formats, and file requirements, so one feed rarely works across all three without remapping.
Products get disapproved, dropped from ads and shopping surfaces, or marked ineligible for AI recommendations and checkout, which reduces visibility and costs revenue until the feed is corrected.
Update product feeds daily or in real time, especially for pricing and stock updates, to avoid selling items you no longer have. You can use feed management tools like Productsup, which update product feeds automatically based on changes.