Skip to main content

Backfill product data with AI Suggestions

Quickly fill in missing weight and replacement cost data for multiple products in your inventory at a time with our AI-powered Suggestions.

Samantha Schurr avatar
Written by Samantha Schurr
Updated yesterday

This article is specifically about Rheaply’s inventory management app, Visibility (root.rheaply.com), and may not be relevant if your organization is not using that functionality.

Learn how you can leverage our AI Suggestions functionality to backfill missing product data like weight and replacement cost. This functionality should help you and your team save time and improve the quality of the data on the platform (directly leading to an increase in measurable ROI).


This article is helpful if: you need to complete product data in bulk, you want consistent replacement costs across your inventory, or you’re standardizing weights for reporting.

Who can do this? Inventory creators, editors, and admins. Keep in mind you can only edit your team’s products, unless you’re an admin or editor.

Where is it done?

Inventory (select products > Bulk actions > Backfill with Suggestions)


Select products to backfill

  1. Navigate to Inventory and use the checkboxes to select the products you want to update.

  2. Click the Bulk actions menu and select Backfill with Suggestions.

  3. Review the info, then click Generate Suggestions.

  4. Your Suggestions will be queued and processed in the background while you return to managing inventory. Once your Suggestions are ready, we’ll notify you.


Review Suggestions

  1. Click Review Suggestions once you receive your notification that your suggestions are ready.

    1. You will also see a callout at the top of the inventory page when suggestions are ready.

  2. You’ll be taken to a page where you can review, edit, and approve or deny each suggested value.

  3. You can also filter to show only suggestions of a specific confidence level for each field, in case you want to approve or deny a whole set of products based on their confidence score.

  4. Once you finalize your review and any changes make sure to hit the checkmark to apply your changes or Accept all to accept all final changes.

Notes:

  • Any data that was initially suggested by AI and accepted by your team will be highlighted when viewing that product's details.

  • You, and only you, will be able to review the suggestions that you generate. No other user will be able to review your suggestions.

  • It is possible (though unlikely) that users update the same product simultaneously, resulting in a value that differs from what was expected to be saved. Make sure to coordinate with your team if you plan on generating suggestions for a large batch of products.


How it works

We use a two-tiered approach to generate product data suggestions:

  • High-confidence method: We leverage OpenAI to search publicly available information online and return likely values for weight and replacement cost.

  • Medium-confidence method: If no direct match is found, we generate an estimate based on the average of comparable products or the product’s category.

Reasoning

Confidence

Data was found on a manufacturer's or dealer’s website.

🟢 High

Data comes from a reputable source on the internet.

🟢 High

Data is estimated based on comparable products.

🟡 Medium

Data is estimated based on the average for the given product types or category.

🟡 Medium

Suggestions are always surfaced for your review and only applied once you click Accept. This ensures all of your product data is human-verified.


Tips & Tricks

Below is a collection of tips & tricks to help ensure you have a smoother experience:

Group products by category or location – Use AI suggestions for groups of similar products (by location or category) to speed up the review process.

Double-check your work – Double-check suggested values for unique or rare items, as averages may not reflect true replacement costs.

Be quick about it – Review suggestions promptly to ensure your inventory data remains consistent and ready for reporting.

Use filters + the accept all functionality – Filter for high-confidence suggestions and accept an entire column to save time. Or filter for medium-confidence suggestions and update or deny those suggestions in one go.

Did this answer your question?