Guide
Product feed optimisation, explained
Product feed optimisation closes the gap between what your store exports and what Google Shopping or Meta can use: complete identifiers, clear titles, accurate prices and availability, useful categories and working images.
The short answer
Product feed optimisation is the work of closing the gap between what your store platform exports and what a sales channel can read and trust: complete identifiers, titles that survive truncation, honest availability and price, accurate categories, and clean images. Done well it means fewer disapprovals and stronger product data; done badly the gaps are hard to spot. SKULayer returns the first audit in minutes. The time to fix it depends on what the feed contains and which changes need source data.
24
checks the audit runs on every item, every run, against the disapproval triggers and quality drags that cost impressions.
Minutes
from connecting a feed to a completed first audit, with what each problem costs and how to fix it.
£0
for the Free plan: one feed, 1,000 products, audited, fixed, and served daily, free forever.
What is a product feed?
A product feed is a file that lists everything you sell in a format a sales channel can read. One row per item, one column per attribute: id, title, description, link, image, price, availability, brand, GTIN, category. Google Shopping takes it as XML, Meta takes it as CSV, and most other channels want a variation on the same idea.
Your store platform can usually produce one. The catch is that it exports what it happens to store, in the shape it happens to store it. Channels have opinions about that shape, and they enforce them.
What is product feed optimisation?
It is the work of closing the gap between what your platform exports and what the channel wants. Some of that is correctness: an item with no GTIN or a broken image link will be rejected, and no amount of bidding fixes it. The rest is competitiveness: a title that truncates before the attribute someone searched for, or a category mapped too broadly, does not get rejected. It just competes with weaker data, which is harder to notice and more expensive over a year.
Feed quality affects whether products are eligible, how accurately they match and what shoppers see, which is why this unglamorous job pays.
What goes wrong in a product feed, and what it costs
Run enough audits and the same families of problem show up in almost every catalogue.
Identifiers
GTIN, MPN, and brand missing or malformed on items that need them.
Google matches your item to a known product using identifiers. Without them it cannot tell what you are selling, so the item loses the comparison it would otherwise win. Items are flagged for limited performance long before they are disapproved outright.
Titles
Titles that truncate, lead with the brand nobody searches, or omit the attribute the buyer typed.
Google accepts titles up to 150 characters, but shoppers typically see only the first 70 or so. Everything after the cut still matches, but the shopper never sees it. A title that opens with a SKU code or an internal product name spends its best characters on nothing.
Images
Placeholder images, watermarks and promotional overlays, or a URL that 404s or times out by the time Google fetches it.
Image problems are one of the fastest routes to a hard disapproval, and a fetch failure at the wrong moment takes the item out mid-campaign.
Availability
Availability that does not match the landing page, or values outside the accepted set.
This is the one Google polices hardest, because it is a promise to the shopper. A mismatch between feed and page can escalate from item-level disapproval to an account warning.
Price and sale price
Price mismatches against the landing page, a sale price above the regular price, a missing currency, or a sale window whose end date lands before its start.
Another promise-to-the-shopper check, and another account-level risk. Broken sale windows are the quiet one: the data is malformed rather than absent, so nothing looks wrong until the promotion fails to run.
Categories and product types
No Google product category, or one guessed at the top level for a whole catalogue.
The category drives which queries you are eligible for and which comparisons you appear in. A catalogue mapped to a single broad category competes everywhere and wins nowhere.
There is a longer walkthrough of the disapproval side in why product feeds fail Google's checks.
What does feed optimisation software do?
Four jobs, in order, on a schedule.
- 1 · Ingest
Read the feed you already have, whatever produces it. A URL from Shopify, WooCommerce, Magento, BigCommerce, or an export from another tool. CSV, TSV, JSON, or XML.
- 2 · Audit
Check every item against the rules that get products disapproved or held back, and report what each problem costs rather than just naming it.
- 3 · Transform
Fix the problems with rules: fill a missing brand, shorten a long title on a word boundary, exclude items out of stock under a price floor, map product types to Google categories.
- 4 · Deliver
Compile a channel-ready output and serve it at a stable URL that the channel fetches on a schedule. Google Shopping wants XML, Meta wants CSV, and both want the current version.
The part that separates tools is the third job. Everyone can fetch a file and write one out. What differs is how changes are expressed, whether you can see the effect before it ships, and how much of the setup you repeat for every channel you add.
How do you choose feed management software?
Six questions worth asking any vendor in this category, including us.
Can you see what a change does before it ships?
A rule that looks right and empties a required field across 4,000 items is a bad afternoon. Look for a preview against your real data, with the count of items affected, not a syntax check.
Does the audit tell you what to do, or just what is wrong?
A list of error codes is a to-do list you still have to translate. A finding that says what it costs and how to fix it is the actual work done.
Is the pricing published?
Quote-on-request pricing means the number depends on how the call goes. That may be fine for you. It is worth knowing which model you are buying into before you invest a week in setup.
How much of your setup is duplicated per channel?
Some tools map every field again for every channel and country. That is fine on day one and expensive on the day you change your title format across nine channels.
Can you undo it?
Rules get changed by people who leave, on catalogues that shift underneath them. Full version history, and the ability to roll back to any version that ran cleanly, are worth more than they sound.
Who has to be in the room to run it?
If the answer is a specialist, that is a hiring decision, not a software decision. If an agency runs many catalogues, client separation and pooled limits decide whether the tool scales with the book.
What changes when AI agents read your feed?
Some buying journeys now start inside an assistant rather than a search box. Someone asks for a specific thing at a specific price that ships this week, and software goes looking on their behalf. That software leans on structured product data, your feed and the schema on your product pages, alongside whatever else it can read, because structured data is the version it can trust and compare.
That makes agent-readiness the same job with a longer payoff. The missing GTIN that costs you a comparison on Shopping is the same missing GTIN that makes you invisible to an agent doing the comparison. Nothing about this requires believing a particular timeline: the groundwork pays today and pays again later.
Where it does change the tooling is access. If an agent is going to help with feed work, it needs to be able to read the audit and propose the fix, safely, without being able to break a live feed on a hunch. That is why SKULayer ships an MCP server where every write is previewed first and only reaches your live feed on the next refresh.
How SKULayer approaches it
Connect the feed you already have and SKULayer audits every item against the checks above, then reports findings that say what the problem costs and how to fix it. Fixes are rules: one ordered stack that shapes a single clean catalogue, compiled per channel, rather than a separate field-by-field mapping for every channel and country. Write a rule yourself, click a one-click pack, describe it in a sentence and let SKULayer draft it, or ask a connected agent. Every path previews against your real items first, and your live feed only changes when a new version passes every check.
Pricing is on the pricing page, not behind a call. If you are comparing against an incumbent, there are direct comparisons with Shoptimised, DataFeedWatch, Channable, Feedonomics, GoDataFeed, Productsup, AdNabu, and FeedOn, each one sourced and date-stamped. For the longer walkthrough, the blog covers why feeds fail Google's checks and what agent-ready product data means in practice.