Skip to main content

How to automatically filter and approve posts using AI Relevance Scoring

Learn how to use AI to score your content based on custom topics and automatically display only the most relevant posts in your widget.

Didi avatar
Written by Didi
Updated yesterday

Managing a social media feed can be time-consuming, especially if you want to display only specific types of content—like posts about "Nature" or "Sales"—while keeping irrelevant posts hidden.

With our AI Relevance Scoring, you can teach our AI to evaluate your posts on a scale of 1 to 100 based on a specific topic. You can then set rules to automatically approve and display only the posts that hit a high score.

Here is a step-by-step guide on how to set this up.

Activate AI Scoring in Account Settings

First, you need to enable the scoring system for your entire account.

  1. Open your Profile section and click on the "settings" tab.

    Opening the profile section, and clicking on the "settings" tab.

  2. Locate the AI Filtering option and activate it.

  3. Add your Prompt: In the text box, describe exactly what the AI should look for. You must specify the scoring range (1-100).

    • Example Prompt: "Evaluate how relevant the post captions are to Nature, start scoring with 1 (Not relevant at all) and 100 (highly relevant)."

  4. Select Scope: Choose whether you want to:

    • Score posts (Old and New): Applies to all existing content and future posts.

    • Score only new posts: Applies only to content pulled after this moment.

    Save your changes.

    Opening the AI Filter section, enabling the AI relevance option. Writing the prompt you want to filter with and saving the changes.

Configure the Source Filters

Now you need to tell the specific social media source (e.g., your Instagram account) to use this scoring logic.

  1. Go to the Sources tab in the dashboard.

  2. Find the specific source you want to filter (e.g., your Instagram Business account), and click on "Edit"

    Opening the sources tab and clicking on "edit" source.

  3. Open the Filters settings for that source, and activate the AI Relevance feature. Enter the Prompt you used in Step 1 to ensure consistency.

    • Example: "Evaluate how relevant the post captions are to Nature, start scoring with 1 (Not relevant at all) and 100 (highly relevant)."

    And click Save.

    Opening the filters section and enabling the AI Relevance option. Writing a prompt for the filtering and saving the changes.


Set Up Auto-Approval in Your Widget

Finally, you need to configure your widget to use these scores to decide which posts to show.

  1. Open the widget you want to edit.

    Opening the moderations tab and the filters section

  2. Go to the Moderation tab and click on "Filters".

    Opening the moderations tab and the filters section

  3. Disable "Auto-add new posts": This is crucial. Turn this OFF. This ensures that new posts are not added blindly; they will wait for the AI's decision.

  4. Set Relevance Score: Use the slider to choose your threshold (e.g., 70%).

  5. Enable Auto-Approval: Toggle on the setting "Auto-approve new posts with selected AI filters".

  6. Click Save.

    Disabling the setting for adding new posts, setting the relevance score and then enabling the option for automatically approving new posts with the selected AI filters

What happens next?

Once you save these settings, the automation begins:

  1. When a new post is published on Instagram, we pull it into the system.

  2. The post is initially disabled (hidden) because "Auto-add" is off.

  3. Our AI analyzes the post based on your prompt (e.g., "Is this about Nature?") and gives it a score (e.g., 85/100).

    Example: Since 85 is higher than your 70% threshold, the system automatically approves the post and it appears on your website widget. If a post receives a low score (e.g., 20/100), it will remain disabled and will not show up on your site.

If you have any questions about writing the perfect AI prompt, click the bottom right chat icon to talk with our customer success team or send us an email at support@embedsocial.com

We're here for you 🙏

Did this answer your question?