Google Shopping product titles: fix them in the shop, not in the feed

The title is a shop field, not a feed field

My advice: improve product titles where they originate, in the product name in the shop. A supplemental feed that permanently overrides the title is convenient to set up and expensive to run. It creates a second source of truth that has to be updated with every new product, every renaming and every change of range.

The reason lies in the data logic. Merchant Center maps the title attribute to the Schema.org property Product.name, and that very property is required in the structured data for merchant listings. A good name in the shop therefore works in three places: on the product page, in the markup for the organic listings and in the feed. An override in Merchant Center only improves the feed.

On top of that, according to Google, Search sometimes adjusts the displayed title itself and uses content from the landing page, images and other signals for this. If the page carries a thin name, Google gets little from it, however good the feed title is.

How long can a title be?

1 to 150 characters are allowed, and Google truncates longer titles. Much less is visible: according to Google, users notice at most 70 characters, depending on screen size. Google still recommends using all 150 characters, because the title is matched against the search query.

This gives a simple rule for the structure:

My assessment: the matching part is not a place for chains of synonyms. Repetitions such as "Sneaker trainer sports shoe" make the product name unreadable in the shop, and that is exactly where the title is supposed to live. What does not help the customer on the product page belongs rather in product_type or the description.

According to Google, the title must not contain price, sale price, offer period, shipping, delivery date, the company name, promotional text or capitalisation for emphasis.

Which order of attributes makes sense?

Google only says: put the most important details first. Feed tool vendors such as Lengow recommend a fixed formula of brand, product type, core attribute, model, size and colour. That works as a starting point. The better question is which feature carries the purchase decision in your category, because that has to go into the first 70 characters.

Depending on the range, I would weight it like this:

For variants, according to Google, the distinguishing detail belongs in the title and additionally in the matching attribute such as color, size or age_group. How to build variant groups cleanly is covered in variants in the feed.

The table is a working hypothesis, not a measurement. Only a test shows what clicks better in your range.

May the feed title differ from the shop?

Yes, within limits. According to Google, feed titles and landing page do not have to be identical, but must clearly refer to the same product. If you write "Running shoe women" in the feed and the page shows a men's shoe, you risk disapproval. If you add colour and size in the feed that the shop name leaves out, you stay within the rules.

Unlike with price, there is no automatic correction from the landing page for the title. The automatic item updates only apply to price, sale price, availability and condition. A deviation in the title therefore stays until someone fixes it in the feed or in the shop.

My assessment: the real risk is not disapproval but gradual drift. A static supplemental feed overrides titles for the item IDs that were in the range when it was created. If new products are added or the editorial team changes a name in the shop, feed and page drift apart without anyone noticing. The larger the range and the more frequent the changes, the faster this happens.

How to use the supplemental feed properly: as a test bench

A supplemental data source adds to or overrides attributes of the primary source, but cannot add or remove products. That is exactly what makes it the ideal test tool: you can serve a new title for part of the products without touching the shop.

The process I recommend:

1. Draw the test group. Split the products into two halves using a stable hash per product, with variants of the same group in the same half. Hand-picking the "important" products distorts the result.
2. Mark the group. Write the membership into a custom label so that you can separate both halves in the report. Which other labels make sense is covered in custom labels.
3. Apply new titles to the test half only, through the supplemental data source.
4. Evaluate cleanly. Feedonomics recommends a run time of two to four weeks, comparing CTR, conversion rate and ROAS per group, evaluating in Shopping campaigns rather than PMax, and freezing budgets and ROAS targets during the test. The difference between the campaign types is described in Performance Max or standard Shopping.
5. Adopt the winner, delete the override. If the new title formula is better, you write it into the shop as the product name and remove the supplemental data source. After that there is only one source again.

When evaluating, I look first at revenue and contribution margin per group, not just CTR. A title that draws more clicks to cheap items can raise CTR and lower profit. Because Smart Bidding and budget shift between the halves, the result is a guide, not a clean experiment.

When a permanent override is right after all

There are cases in which I depart from the principle:

• The shop does not allow good names, for example because the ERP fixes the product name and a change takes weeks in IT. The override is then the realistic solution, but with a fixed rule instead of manual upkeep so that new products are picked up automatically. Whether the shop plugin is enough for this is covered in shop plugin or feed tool.
• Marketplace or manufacturer data supplies titles that you may not change.
• The product page deliberately needs a short name, for example in a premium design, while the feed spells out the features. This is within Google's framework as long as both clearly mean the same product.

In all three cases the deviation is a decision, not a legacy. Document why the override exists and review it with every change of range.

The check for your feed

Before you rewrite anything, a comparison is worthwhile: export title from Merchant Center and the product name from the shop and compare both per item ID. There are three possible results. Both the same and good: do nothing. Both the same and weak: improve in the shop, after a test. Feed deviates: find out where the override came from and decide whether it is a test, a deliberate exception or a legacy. More on the feed as a whole is in the feed is Google's model of your range, and on rewriting with AI in improving product data with AI.

As of: Merchant Center rules for titles, AI labelling and automatic updates change from time to time. The limits above match Google's help pages at the time this article was published.