Skip to content
GetProfit

Shopping Titles: From Product Codes to Clear Descriptions

Using a textiles example, we explore which details to put first in Shopping titles, how the order varies by category and how to assess changes.

A supplier’s product name is convenient for stock records: it helps us find the right item in a price list. But someone on Google Shopping is seeing the product for the first time. If all they see is a collection name and a colour, they have to guess what is being sold and whether it suits their needs.

I would treat the title as a short explanation of the product. First, the shopper needs to understand what it is. Then they need to see the details that distinguish a suitable option from the others. Which details come first depends on the category.

Let’s take one unclear title

Imagine a hypothetical home textiles shop. The title “Luna, beige” has come into the feed from the supplier’s price list. This is a fictional product. Suppose its verified product information tells us that the offer is for one cotton duvet cover measuring 200 × 220 cm. It is not a duvet or a set with pillowcases.

The title does not currently tell us that. So I would add the product type first, followed by size and material. I would put the collection name and colour later. We cover the wider feed fields in our Shopping product data optimisation guide; here, let’s focus on assembling one clear title.

Clarifying the offer
The title should tell us what is for sale
From the supplier’s price list
Luna, beige
After checking the specifications
Duvet cover 200 × 220 cm, cotton, Luna, beige
Product typeSizeMaterialCollectionColour
A fictional product. The example is for one duvet cover, without a duvet or pillowcases. Specifications must be confirmed before use in a live feed.

Look at what changed. We did not add “best”, “bestseller” or a promise of a discount. We simply answered a few questions: it is a duvet cover, of this size, made of cotton and beige in colour. Someone looking for a set can also recognise more quickly that this is a different offer.

In practice, each of these details must come from the specification or product information for that exact variant. A photograph cannot reliably tell us a fabric’s composition. And the same collection name does not mean every item in it includes the same pieces.

Attribute order depends on the shopper’s choice

For a duvet cover, size helps someone work out whether it fits the duvet they already own. That is why it comes near the beginning in our example. But I would not automatically apply that order across the whole catalogue.

Now imagine a shopper who has already chosen a particular coffee machine model. The brand and exact model may be more useful to them than the colour. For someone choosing a shower curtain, the product type, size and material matter. The collection name can wait.

Order by category
Start with what drives the choice
Duvet cover: will the size fit?
Type → size → material → collection and colour
Coffee machine: is it the right model?
Brand → model → device type → important differences between variants
Shower curtain: does it suit the task?
Type → size → material → pattern and colour
Examples of attribute order, not mandatory Google templates. We choose according to the shopper’s task and verified product data.

These are starting points for different tasks, not mandatory Google formulas. To choose an order for our own category, let’s look at the available search queries, catalogue filters and questions about the products. If people search for specific models, we need to help them recognise the model. If they choose by size, we should not hide it behind the supplier’s internal code.

When information is missing, I would pause to clarify it. A polished title that describes the wrong contents will create more problems than a short but accurate one. And there is no need to string related terms together, such as “duvet cover, bed linen, set, duvet”: these refer to different products.

Length is a limit, not a target

According to the Merchant Center specification, a title can contain up to 150 characters. Important details should come earlier: the full text may not fit in a particular placement. There is no need to fill the limit for the sake of length.

The title must describe the product accurately and match the information on the page. We do not add prices, delivery details or promotional claims. Google may also adapt the displayed title, so the text in the feed is not guaranteed to appear word for word.

If we use generative AI for titles, the same specification requires us to submit them through structured_title, setting digital_source_type to trained_algorithmic_media. This also applies when transferring generated examples into a live feed. A person still needs to check the specifications: the technical label does not fix an invented fabric composition or size.

Let’s check a small group before rewriting everything

I would start with one category where the same problem is clearly visible. For example, the duvet covers list a colour and collection but lack the product type and size. Let’s save the original titles, the list of products in the group and the date of the change. We keep the product identifiers: we are changing titles, not creating different products.

After the update, we check that the new values have reached Merchant Center, that the products are eligible to show, and that they describe the same variants as the site. Then we look at the impressions, clicks and orders the group receives. Practical ways to handle changes like these are covered in our product data playbook.

Checking the change
From new text to an observable result
1. Record
One category, a product list, the old titles and the date. Keep the identifiers.
2. Check the data transfer
The new values have reached Merchant Center; the product is eligible and matches the page.
3. Assess
Impressions, available queries, clicks and orders — alongside price, stock availability, budget and demand.
More clicks ≠ more profit. One purchase ≠ a proven effect of the title.
Comparing periods alone does not establish causation. Google may adapt the text the shopper sees.

More clicks alone do not mean sales have improved. And a comparison of “the week before” with “the week after” does not separate the effect of a title from changes in price, stock availability, advertising budget or demand. Leaving a similar group unchanged for comparison is a useful check, but does not automatically make it a clean experiment.

So I would record in advance what we want to find out: did the new title arrive, is the product clearer, and have the available queries and the group’s results changed? With few orders, we keep observing rather than declaring success after a single purchase.

What we need at the end is a clear template for the category and verified product specifications. We can then describe the next product using the same logic — preserving its differences, rather than simply adding more words to its product code.