Skip to main content
Products AI Guide
Updated this week

Configuring Products AI

In this overview, we will explain how to set up the Producthero’s Products AI in the Producthero Platform.

Products AI is a tool that utilizes artificial intelligence to automatically optimize your Google Shopping product attributes. By following Google's best practices, it enhances your feed and boosts relevancy. With a fully optimized feed, Google can better match your products to search terms and users, leading to increased visibility and relevance. In the long term, this results in more impressions and a higher conversion rate.

Products AI will automatically improve the most important Google Shopping feed attributes. Feed management can be a time consuming and tedious task; however, improving the feed makes a significant difference to the performance of the Google Shopping campaigns.This time and effort can be invested into more relevant tasks, like strategy or data analysis.

Products AI will optimize the following attributes:

  • Title

  • Product_type

  • Description

  • Color

  • Size

  • Age group

  • Gender

  • Pattern

  • Material

Products AI is currently available for the following languages:

  • Danish

  • Dutch

  • English

  • French

  • German

  • Italian

  • Spanish

  • Swedish

By optimizing these fields using Products AI, you can optizime your feed in bulk so it follows the best practices of Google. This way you can save a lot of time on doing tedious tasks, while also improving the performance of your campaigns.

Topics covered in this guide:

  • Selecting products

  • Prompting

  • Credits

  • Reviewing changes

  • Uploading the supplemental source

To start with Products AI, go to Products AI by clicking the box in the left menu.

1.1 Selecting products

Once you have arrived you will see the products, their attributes and their labels.

You can select multiple attributes to filter the products you want to optimize. This includes:

  • State

  • Title status

  • Brand

  • Product type

  • Google Product category

  • Custom Label

This way, you can easily filter the right products you want to optimize. Some examples to get started with filtering:

  • Heroes + Poor title: by improving the attributes of best performing products, you can increase relevancy and therefore the volumes of your best performing products.

  • Zombies + not edited: improve relevancy so these products will gain more impressions, and have the opportunity to switch labels.

  • Category (Seasonality): optimize seasonality products in bulk by adding seasonal keywords to the titles to boost volumes during peak periods (Christmas, Summer, Valentines Day etc.)

  • Brands: filter on your best selling brand to increase relevancy

Once you have applied the filters you can select the products you want to optimize. You can do this by either selecting the products manually or by clicking the box in the top left corner. If you click the box in the top left corner, you will automatically select all products on the page, which is 25 products at max.

Once you have selected the products, you can click on the blue optimize button in the top right corner.

1.2 Prompting

In the optimize section, you have four options to determine how the AI model will optimize your products. This includes:

  1. Title template

  2. Language

  3. AI creativity control

  4. Customize prompt

Select title template

The title template follows the best practices of Google regarding the structure of your titles. In your titles, you want to follow this structure so the titles contains the most relevant search term a potential client is searching for. You can choose between the following title templates:

  • Apparel

  • Consumable

  • Hard Goods

  • Electronics

  • Books

  • Custom title structure

Language

With the language settings you can inform the AI model which language your feed is set to. This way, the model will optimize the content using the language you’ve set it to. The languages which are currently supported are English and Dutch.

AI creativity control

With the AI creativity control you can decide if the results of the optimizations are strict or random. Creativity control can be set with a slider that adjusts how much the model takes risks or plays it safe in its choices. When it’s set too strict, the AI model will follow the prompt as closely as possible. When it is set to random, it will be more creative.

Customize prompt

In the context of artificial intelligence, a prompt is an input or a set of instructions provided to the AI to generate a response. This prompt can be a question, a statement, or a task description that directs the AI on what type of content to create.

You can customize the prompt to give the AI model instructions on how to optimize your titles. This way, you are in control of the results. You can add additional instructions to write an optimized prompt, so you can guide the AI model into the right direction.

Examples of relevant prompts are;

  • Replace redundant keywords in the title for product attributes

  • Add the brand name to the end of the title

  • Add seasonal keywords to the title

  • Add more synonyms to the title for maximum coverage

  • Change the title structure so the most relevant keywords are shown first

Best practices for prompting:

  • Be clear and specific

  • Emphasis important parts in caps

  • Use line breaks for different parts of the prompts

  • Make sure to add limitations (no more than 150 characters for the titles, no more than 5 layers deep ">" for the product type, make sure all attributes are in Dutch etc.)

It is also a possibility to leave the prompt empty; our AI model will then optimize the title according to the best practices of Google.

Once you have selected the right settings and made a custom prompt, you can press the “optimize” button. The AI model will optimize the selected products.

This will take a few seconds, depending on the amount of products you have selected.

1.3 Credits

Products AI includes unlimited credits, enabling you to optimize all your products. This allows you to take your Shopping ads to the next level, achieve better results, and maximize your campaign performance without restrictions.

1.4 Reviewing changes

When the products are optimized, the results will be displayed as below. You can review your new content. By clicking on the arrow, you are able to review your content and compare them to the original content.

If you aren’t satisfied with the results, you can revert the content back to the original version.

After optimizing the content, one of the following statuses will show;

  • Green: Optimized without errors

  • Red: Optimized, but an error occurred. The title is likely too long, exceeding 150 characters.

  • Grey: These products aren’t optimized, meaning they haven't been selected for optimization.

1.5 Approving products

Once you are satisfied with the results, you’ll need to approve the products before they can be added to the supplemental source. Follow these steps to approve the changes:

Use the filter and select State > Optimized to view all optimized products.

In the top left corner, click the checkbox to select all the optimized products.

After selecting the desired products, go to the Optimize button and approve all the changes.

1.6 Uploading the supplemental source

After approving the results, you can upload the supplemental source to Merchant Center. In the top right corner, click on the three dots:

You can either copy the supplemental source, download as a .csv to preview or undo all changes:

If you have copied the supplemental source, you can follow the steps in this video for classic Merchant Center Next.

Currently, the supplemental source is not integrated into the supplemental source that’s being used for the Labelizer, Optimizer and Price Benchmark data. The supplemental sources will be merged together at a later stage, so you will have all optimizations in one place.

1.7 Uploading the Products AI supplemental source when another supplemental source is already added to Merchant Center

If you already have a supplemental source in Merchant Center with attribute data similar to Products AI (such as optimized titles from the Optimizer), you can create feed rules (Attribute Rules in MC Next) to ensure that the titles from Products AI do not override the values from the other supplemental source. You can do this by following these steps;

Click the gear icon in the top right corner and go to data sources.

Navigate to supplemental sources.

Copy the supplemental source from the Producthero platform. Make sure you use the right language / country combination. Then, select add product from a file and paste the link.

Then, proceed to select the right language, country and primary data source and select create data source:

After adding the supplemental source, go to Data Sources > Primary Sources and select the right primary source. Then, proceed to Attribute Rules. By default, a rule has been created, but you can modify it to ensure the correct data is selected from the right source.

For example:

  • If you’ve optimized titles with the Optimizer and want these titles to appear in search results, select "Take from: Producthero supplemental source."

  • If a product hasn’t been optimized by the Optimizer but has been by Products AI, select "If the attribute has no value yet, take from: Products AI."

  • If the titles haven’t been optimized by either the Optimizer or Products AI, select "If the attribute has no value yet, take from: [Your Primary Feed Source]."

By following this structure, you can build a hierarchy that ensures the correct attribute values are used and prevents the wrong supplemental source from overriding them.

Did this answer your question?