Instagram Launches New Tools to Help Refine Content Recommendations
The secret to TikTok’s enormous popularity is its finely trained algorithm, which quickly picks up on what users like and dislike and transforms that information into an increasingly compelling stream of short video snippets.
Instagram is now relying more on AI-driven suggestions, which it claims have greatly improved engagement since being included into user feeds, because it is aware of this.
But it’s evident that’s insufficient because Instagram today unveiled several new features intended to make it easier for users to give more direct input into what they see in the app, further tailoring their IG feed to their tastes.
In order to speed up your algorithmic training process, Instagram is initially testing the option to label several posts in Explore as ‘Not Interested,’ as you can see in the first image above.
“We’ll immediately hide those posts and refrain from showing you similar content in the future.”
That could make it easier for you to get rid of a lot of clutter at once, which should, in theory, demonstrate to Instagram that you have no true interest in the subjects you choose to emphasize.
Which should be effective, but despite directly telling Instagram that I’m not interested, I still receive a lot of arbitrary or marginally relevant recommendations in Explore.
Sending the app bulk answers can perhaps help emphasize this point.
The ability for users to indicate Instagram they don’t want to see recommended photos with specific words, phrases, or emojis in the caption or included tags will also soon be tested, as shown in the second image.
“Whether you’re seeing something that’s not relevant, or have moved on from something you used to like, you can use this feature to stop seeing content that’s not interesting to you.”
The new self-reporting features should work together to give users more control over how Instagram is relevant to them and to give Instagram’s engineers a better understanding of which related recommendations are helpful and which annoy users so they can improve the automated content highlights in-stream.
Although TikTok’s AI system appears to be considerably better at recognizing variable aspects in postings and responding to direct human engagement, it falls short in that users must actively supply such input.
The truth is that many users simply won’t utilize any manual tools that IG offers on this front; yet, by taking into account the knowledge that users do offer, perhaps this can assist to better inform its automatic suggestions for everyone in any case.
Along with these new possibilities, Instagram has also published a brief description of its existing suggestion system. As previously mentioned, this system uses machine learning to discover other items that you might be interested in based on your past activity within the app.
As explained by Instagram:
“One of the ways we personalize your feed is by predicting how likely you are to do something with a post you see. The more likely you are to take an action, and the more heavily we weigh that action, the higher up you’ll see the post in your feed.”
On this front, Instagram says that there are five specific interaction metrics that it uses to guide its recommendation system:
Dwell time on posts
The likelihood of a user commenting on a post
The likelihood of a user liking a post
The likelihood of a user re-sharing a post
The likelihood of a user tapping though to the creators’ profile
In the recent past, it has seemed like re-shares have been given more priority, which would align with Instagram’s broader mission to help amplify creators in the app.
These are the key interactions that the platform’s algorithm focuses on in deciding what to show each user more of, although Instagram doesn’t explicitly state that any one of these factors is weighted more heavily than the other. If you’re looking to optimize your IG posting process, you should know this.
How can you leverage it as a Marketer on Social Media? Improve dwell time by presenting aesthetically engaging content (easier said than done, I know), and increase interaction by encouraging comments, maybe by asking community questions.
(It’s also worth noting that ‘saves’, which had been highlighted as a key metric of focus by some social media marketing commentators, are not specifically mentioned in this new overview.)
You may gain a better understanding of how Instagram looks to emphasize particular posts in accordance with user preferences by taking use of the new tools and insights that together offer some further information on the platform’s recommendation tools and process.
These additional manual input components will benefit the algorithmic systems over time, though it’s unclear if they can compete with TikTok on this front.