Last updated Nov 15, 2022
Every day millions of people, including influencers, are uploading photos of products, discussing products, and engaging with brands across our platforms. Over 90% of accounts follow at least one business on Instagram and people are being inspired by influencers and creators to shop directly on our platforms more often. Social Commerce Model is an AI System that helps connect people to products from businesses and creators that are interesting to them. With Social Commerce, we’re able to pick up on patterns like influencers on Instagram posting new styles of clothing and connect that emerging fashion trend to relevant products making it easier for people to find them.
We’ve designed the Social Commerce Model with privacy in mind and focused on creating a model that meets people’s expectations and is easily explainable. For example, we limit the data that is used by the system: Social Commerce Model looks at content from creator or business accounts belonging to people who are over 18 years of age, where the post contains content with commercial intent, like a product image or a product tag. By using data responsibly and focusing on data that presents the most value, the Social Commerce Model is creating better connections to people, businesses and products through more innovative ways.
To show users the most interesting and relevant products, we use our Social Commerce Model to identify the most popular and trending products from public user content on Facebook and Instagram. These AI models are trained to recognize products that appear in public user content before identifying similar products for sale on our platforms. This helps us display products that are popular and trending to people that want to find them.
First, the AI system scans posts shared on public Instagram accounts by creators and businesses that promote particular products in their product catalogs.
After specific products are identified from posts, our system analyzes the product attributes such as styles, related hashtags, and more to best describe products in the post.
Next, our systems identify which product trends are appearing most frequently and getting the most engagement. This is how we can identify emerging trends.
The system also groups together products that are visually similar and available for purchase.
Finally, we use these learnings to understand which products are trending with creators and business accounts, predict future trends, and display more relevant products in your feed.
Social Commerce provides recommendations based on machine learning. Sometimes we combine the outputs of our Social Commerce AI with the expertise from our Meta fashion teams’ to create premium content experiences. We have taken into consideration the feedback and to avoid bias, and that’s why when we editorialize it, we make it clear on the labels. For example, we disclose when human editorial curation is at play in the titles of the recommendation modules in the shopping experiences. When you are browsing shopping content on Instagram, you may be seeing #trending collections within Editor’s picks.
Today, Social Commerce Models are mainly created from scanning products contained in photo and text content, but we’re excited about the possibility of identifying trends from videos, for example, from Reels.
Metadata about a post which serves as inputs into an AI Model, such as the time a post was created or how many comments it has.
A continuously updated list of advertisements and recent posts from the people you follow on Instagram.
A measure of what a user cares about, based on past engagement and activity.
A statistical representation of tasks learnt from datasets, which can provide meaningful insights and be used for predictions.
A group of ML models, AI and non-AI technologies that work together to accomplish specific tasks, such as ranking posts in a feed.