Google’s Hidden Data Layer Explained

Ecommerce businesses can sometimes find products appearing in Google with incorrect prices, outdated images or unexpected sale labels, even when their current product feeds and websites appear to be correct.

The reason may be what can be described as Google’s hidden data layer. Rather than relying only on the information a business currently provides, Google builds up its own history of product information by crawling websites, feeds and other online sources.

This can include previous prices, product images and information found on third-party websites. When Google’s records do not match a retailer’s current information, unexpected results can sometimes appear in Shopping, organic search or advertising.

For ecommerce teams, this means checking the current feed alone may not always be enough.

What Is Google’s Hidden Data Layer?

When a retailer submits a product feed, it provides Google with its current information, such as the product price, availability and image.

However, Google also crawls product pages and revisits feeds over time. This allows it to build a record of information it has previously seen.

Google can also discover product information elsewhere online. An old marketplace listing, for example, may still contain an image or price that a retailer stopped using years ago.

The result is that Google can have several versions of a product’s information from different points in time and different sources.

This is not necessarily a deliberate attempt by Google to override a retailer’s current data. Google is trying to build a reliable understanding of products, but differences can occur when information changes faster than Google’s records are updated.

Three areas are particularly important:

Price history: Google can compare current product information with prices it has previously seen on feeds, pages and structured data.
Image indexing: Google maintains its own record of product images and may continue showing an older image after a retailer has replaced it.
Data from other platforms: Product information may also be found across marketplaces and other online sources, creating additional signals that Google can use.

Google’s Product Price History

One common problem is a product being shown as discounted when the retailer is not actually running a sale.

There are several different price-related features in Google, and they can easily be confused because they can look similar in search results.

Sale Price Badges

A retailer can deliberately submit a sale price through its Merchant Center feed using the sale_price and sale_price_effective_date attributes.

Google checks whether the discount meets its requirements. For example, the discount must generally fall between 5% and 90%, while the original price needs to have been available for a qualifying period.

For UK products, Google requires the original price to have been submitted for at least 30 days during the previous 200 days.

Importantly, Google does not only consider the current feed. It can compare the submitted base price with its own historical records.

Price Drop Badges

Price drop annotations work differently.

Retailers do not directly submit these badges. Instead, Google can compare the current product price with the average price it has recorded over the previous 60 days.

If the reduction is significant enough, Google may automatically display a Price drop label along with a previous price.

The retailer does not provide that previous price directly. It comes from Google’s own records.

Price Drop Rich Results

Google can also display price drop information in organic search results.

For these results, product structured data needs to provide an individual offer price rather than relying on an AggregateOffer with a low and high price.

Again, however, the information Google displays is not necessarily limited to what is currently contained in the site’s structured data. Historical information held by Google can also influence what appears.

This is why checking only the current markup may not explain every price-related annotation appearing in search.

How An Unexpected Sale Badge Can Appear

Consider a retailer that sees a product displayed in Google Shopping with a message such as “Was £474, now £355”, despite never intentionally running a sale.

The first step would be to check the current Merchant Center feed.

In one example, the feed contained a price of £355 excluding VAT and did not include a sale_price attribute.

The latest crawl of the product page showed £426, which represented £355 including VAT.

Looking further revealed that the website’s HTML contained the word “Now” next to the price.

The product’s structured data also used an AggregateOffer containing different price tiers. One was labelled as a list price of £395, while another was labelled as a sale price of £355.

The £395 figure was particularly interesting because adding VAT brought it to £474, matching the historical price Google was showing.

An older feed export then revealed that the product had previously been listed at £474.

No sale price had ever been deliberately submitted. Instead, Google had several signals available:

A previous price in the feed
A list-price reference in the structured data
Wording in the page’s HTML suggesting a current price

Google was therefore able to combine these pieces of information and produce a sale-style annotation.

The lesson is that Google’s understanding of a product can extend beyond the information currently being submitted.

Old Product Images Can Remain Visible

Images create a similar problem.

When an ecommerce business replaces an old product image, it may assume that Google will immediately stop using the previous version.

That does not always happen.

Some retailers delete old image files from their servers, causing the old URL to return a 404 response. Others periodically clean up unused images. Some leave old files permanently available on their content delivery network.

The last approach can create problems because an old image remains accessible and can continue to be discovered by Google.

Different ecommerce platforms handle this in different ways. Shopify CDN image URLs, for example, can remain available after a product is removed. Magento images can remain in media directories, while WooCommerce images often stay in the WordPress media library after a product has been retired.

Renaming images can also create duplicates. If a new lifestyle image is uploaded under a new filename while the old file remains accessible, Google may have two image URLs associated with the same product.

Google maintains its own image index and can discover images directly from webpages rather than relying solely on the image URL supplied through Merchant Center or structured data.

This means updating a feed does not necessarily remove an older image from Google’s records.

A Third-Party Website Could Be the Source

The search for an incorrect product image may also need to go beyond the retailer’s own website.

An old image can remain live on a marketplace or another third-party website. If Google has crawled that page, it may use the image as another signal when deciding which image is associated with a product.

For example, an old product image uploaded to an Amazon listing years earlier could still appear in Google even after the retailer has replaced the image on its own website.

This can be particularly difficult to identify because ecommerce teams may not think to search marketplaces for images they have already replaced.

A useful first step is to take the filename of the unwanted image and search for it in Google Images. This can reveal where else the image is still being hosted.

Could Product Data Be Shared Between Platforms?

There is another potential layer involving major ecommerce platforms.

Some practitioners believe marketplaces such as Google and Amazon may exchange certain product data through APIs or other commercial arrangements. However, the specific mechanism described by practitioners is not publicly confirmed or documented by Google or Amazon.

One reported example involved a discrepancy between a Merchant Center feed and an Amazon feed. The two were treated as separate data sources, yet an error in one appeared to have wider consequences for advertising activity.

This should be treated as an edge case rather than an established rule.

There is, however, evidence that Google and Amazon have platform-level data integrations. Google’s Amazon Multi-Channel Fulfilment integration allows Amazon to provide fulfilment and delivery information to Merchant Center, which can be used for delivery estimates in Shopping.

That does not prove that product feed data is routinely exchanged between the two platforms, but it demonstrates that connections between the systems do exist.

How To Investigate Hidden Data Problems

When something looks wrong in Google Shopping or search results, checking the current feed is only the starting point.

Check Your Price Data

In Merchant Center, go to Products > All products, select the relevant product and look at the Product details section.

The Information found on your site area can show the price and availability Google found during its latest crawl, along with the date of that crawl.

Compare this with the price in your feed.

A difference of exactly 20% may indicate a VAT issue where the feed uses an ex-VAT price while the website displays the price including VAT.

Larger differences, or a price that matches an older figure, could indicate that Google’s historical information is contributing to what appears in search.

It is also worth checking your website’s HTML and structured data. If terms such as Sale, Now or Was appear in the markup when there is no genuine sale, consider changing them.

Keep historical copies of your feeds too. A simple archive of previous exports can make it much easier to identify when a price changed and what Google may have previously seen.

Investigate Old Images

If Google is displaying an outdated image, check:

Merchant Center assets
Primary and supplemental feeds
Feed rules
The live product page
Page HTML
Structured data
Old image URLs

Then search the image filename in Google Images.

If the image appears on an old marketplace listing or another third-party website, this may explain why Google continues to associate it with the product.

Businesses should also review their internal image processes. Ask what happens when a product image is replaced, whether unused images are removed and who is responsible for checking images across third-party platforms.

Compare Google And Amazon Data

Businesses using both Google and Amazon should compare the two feeds when unexplained disapprovals or performance problems occur.

Look at:

Product prices
Availability
Titles
Image URLs
Feed processing dates

Even if the two platforms are managed separately, comparing the information can help identify discrepancies.

Recording the date of each feed export is also useful. Two platforms may be working from different versions of the same product data even when both feeds are expected to be synchronised.

The Bigger Issue Is Data Fragmentation

Google’s hidden data layer is only part of the problem.

Many ecommerce businesses already have product information spread across several teams. SEO teams may manage structured data, PPC teams may manage Merchant Center feeds, developers control the website and ecommerce teams manage the product catalogue and images.

Each team may check its own information without looking at the other sources.

This creates a fragmented picture of what the business is actually telling Google.

Ideally, teams should compare the feed, structured data and webpage rather than treating each as a separate system.

Google is also working towards a more unified understanding of product information. Its Shopping Graph contains more than 50 billion product listings, with more than 2 billion refreshed every hour. Information can come from Merchant Center and Manufacturer Center feeds as well as manufacturer websites, product pages, YouTube, reviews and product testing data.

Google can then cross-reference this information with its wider understanding of products, brands and entities.

Exactly how Google resolves conflicting information between these different sources is not publicly documented.

Why This Matters More For Ecommerce

The importance of accurate product data is likely to increase as Google expands AI-powered shopping and automated purchasing.

Google has already introduced agentic shopping features that can allow shoppers to set price targets, receive notifications and have purchases completed through Google Pay.

For these systems to work reliably, Google needs an accurate understanding of each product, including its price, availability and imagery.

That makes consistent product information increasingly important.

The hidden data layer is therefore less about Google deliberately hiding information and more about the gap between what businesses currently provide and everything Google has previously discovered.

For ecommerce teams, the best approach is to maintain a consistent view of product data across feeds, websites, structured data and third-party platforms. Regularly checking historical information and old assets can also help prevent outdated signals from affecting how products appear in Google.

This can also be shortened into a more concise Property Accelerator-style news article if you want to keep the blog tighter.