How to remove duplicate attributes and tidy up your catalog

How to Remove Duplicate Feature Names Using AI

Elbuz is a solution that will help you clear product characteristics from duplicates to create convenient and correctly working filters on your site. After setup, the system will remember your choice and will no longer create duplicates for processed attributes.

1. What are duplicate attributes and why do they appear?

Duplicate attributes — these are characteristics of goods that have different names but the same meaning. The system perceives them as different properties, which prevents the catalog from working correctly.

Simple examples of duplicates:

• Color,Color,Colors,Primary color
• Size,Size,Dimensions (WxHxD)
• Material,Compound,Material,Material of manufacture

Why do they arise?

• Parsing from different sites: Each online store or marketplace uses its own names for characteristics. When collecting data, this entire "zoo" of names gets into your catalog.
• Import from different suppliers: Each supplier has its own price list format and its own attribute naming standards.
• Manually adding products: Different managers may call the same attribute differently (for example, "Screen Diagonal" and "Diagonal").
• Typos and different languages: Human factor, for example, "Material" and "Material", or the use of Russian and English names (Color And Color ).

What problems do duplicates cause?

• Unable to create valid filters: The site's filter will contain several items with the same meaning ("Color"), which will confuse the buyer.
• Buyers can't find the product they need: By selecting "Color: Red" in the filter, the customer will not see products that have a color specified in the "Color" attribute.
• The catalog looks sloppy and unprofessional.
• It is difficult to analyze product data.

2. How Elbuz fights duplicates

Our system uses artificial intelligence (AI) to analyze and group similar attributes. All you have to do is check the result and confirm it.

Key features:

• AI analysis of names: The system automatically finds and offers to group attributes of the type "Manufacturer" and "Brand".
• Grouping duplicates: All found synonyms are combined into one group for easy processing.
• Selecting the main attribute: In each group, you decide which name will be the main one (for example, “Color”), and which will become its synonyms.
• Saving rules: This is a key feature. The system remembers your choice. During subsequent import or parsing, if the system encounters the attribute Color, it will automatically match it with your main attribute.Color and will not create a duplicate.

Operating modes:

• By category (recommended): You select a specific category, for example, "Smartphones". AI will analyze only the attributes of products from this category. The result will be much more accurate, since AI works in a narrow subject area (for example, it will understand that "Screen diagonal" and "Screen resolution" are different things).
• Globally (for all categories): Analyze all attributes in the entire catalog. Use this mode with caution, as the AI may incorrectly group attributes from different product groups.

3. Step-by-step instructions for removing duplicates

Step 1: Preparation

1. Go to the attribute management section where you will find the duplicate removal tool.
   
2. From the list, select the product category you want to process.We strongly recommend starting with individual categories.
   
3. If you want to analyze the entire catalog, select the "Category List" option.
4. The system will show how many unique attributes were found for analysis in the selected category.

Step 2: Run the analysis

1. Click the button "Run Analysis".
2. Wait for the completion. Artificial intelligence will analyze the names of all attributes and group similar ones by meaning.
3. A list of found duplicate groups will appear on the screen.

Step 3: Check the results

Each group is a block with similar attributes. Inside you will see:

• List of all attributes, which the AI considered duplicates.
• Main attribute: It is marked with an asterisk ⭐. Its name will remain in the catalog after processing.

Step 4: Setting up groups

This is the most important step. Carefully check each group suggested by the AI.

• View attribute values
  • Click on the button 📋"Show values" next to any attribute. You will see a list of all its unique values (for example, for the attribute "Color" these will be "red", "blue", "white").
  • Why is this necessary? To make sure that the attributes in the group are really the same. For example, if the "Size" group contains the "Weight" attribute, looking at the values ("50x50 cm" and "1.5 kg"), you will immediately understand the error.
    

• Change main attribute
  • If you want to keep another name as the main one, click the ⭐ button "Make it the main thing" for the desired attribute in the group. For example, make "Brand" the main one, not "Manufacturer".

• Remove attribute from group
  • If the AI mistakenly added an attribute to a group that is not a duplicate, click the ❌ button "Remove from group". It will not be affected when saving.

• Delete the entire group
  • If the entire group is not formed correctly (for example, "Width" and "Length" are mixed), click on the 🗑️ icon to the right of the group title to remove it completely.

Step 5: Select groups to save

• By default, all groups are selected. Uncheck the boxes to the left of any groups you are unsure about or do not want to process at this time.
• You can use the buttons "Select All Groups" / "Deselect All" for convenience.
• Click "Preview", to see how the attributes will be renamed and to estimate the scope of the changes.

Step 6: Save changes

1. When you are confident in your choice, click "Save selected groups".
2. The system will ask you to confirm the action.
3. Warning! This action is irreversible. After saving, the old attribute names will be replaced with the main ones, and the system will remember the matching rules. It is impossible to cancel the operation en masse.
4. Wait for the process to complete. Done!

4. Recommendations and best practices

Best practices:

• Start with categories. Process the catalog in parts. AI works more accurately, and it is easier for you to check 5-10 groups in the "Components" category than 100 groups from the entire site.
• Always check the values. Always press 📋 before saving a group "Show values". This is the best way to avoid mistakes.
• Choose the right key attributes. Use clear, complete names (e.g. "Screen Diagonal" is better than "diag.")
• Be careful with similar attributes. "Width" and "Height" are different attributes. "Case Color" and "Strap Color" are different too. If in doubt, it's best to remove the attribute from the group.

Typical mistakes:

• ❌ Group attributes that have different meanings (“Memory size” and “Engine size”).
• ❌ Select an incomprehensible name as the main one (“attr_123” or “prop_size”).
• ❌ Save groups without checking their contents by viewing values.
• ❌ Process all categories at once without any experience.

5. After processing duplicates

What happens in the system:

• Products are updated: In all products where duplicate attributes were used, they are replaced with a single main attribute.
• The rules remain the same: The system creates a mapping map (e.g.Color →Color,Prod. →Brand ).
• New products are processed automatically: The next time you import products with the attribute Color the system will automatically apply the rule and use your main attribute Color.No more duplicates!

Creating filters on the site: Now that you have one attribute "Color" instead of five, you can easily create a filter on your site by it. Your customers will thank you!

6. Troubleshooting

• If the AI did not find any duplicates:
  • Perhaps they really aren't in the selected category.
  • Try choosing another, more "cluttered" category.
  • Make sure that the category actually has products with attributes.
• If the grouping is incorrect:
  • Don't save it. Use the ❌ buttons to remove unnecessary attributes or 🗑️ to remove the entire group.
• If you need to undo changes:
  • There is no mass cancellation. That is why checking before saving is so important. The error can only be corrected manually through standard attribute management, which is very labor-intensive.

7. Conclusion

Elbuz's duplicate removal tool is a powerful assistant that will solve the problem of feature confusion once and for all. It will not only clean up the current directory, but will also automatically maintain order in the future.

The main rule: it is better to process fewer groups, but efficiently, than many, but with errors.

Save a link to this article