Pinterest provides guidance on optimizing algorithms for engagement.
Pinterest has revealed new details on algorithms to enhance user experiences.
Concentrating on clear engagement cues, such as Likes and comments, may result in negative consequences.
As elucidated by Pinterest:
“User interaction is a crucial indicator utilized by Pinterest and various online platforms to determine the content displayed to users. However, it is common knowledge that solely optimizing for user interaction can showcase substandard content (e.g., ‘clickbait’) or even harmful material. Pinterest’s CEO, Bill Ready, warns that if not vigilant, content prioritization may highlight the ‘train wreck we can’t turn away from.’ Conversely, ‘if you inquire about showing another accident after witnessing one, the majority would decline vehemently.'”
This poses a significant hurdle for social platforms, as the motivations of direct involvement, which boost activity, can frequently result in platforms favoring inappropriate content or unintentionally encouraging users to upload content that skirts the rules to attract clicks.
This is an issue without a complete solution, despite efforts from every platform.
Meta CEO Mark Zuckerberg addressed this challenge in 2018:
“One major challenge for social media platforms is the tendency of users to gravitate towards sensational and provocative content, which can negatively impact public discourse and contribute to polarization. This behavior is not limited to online platforms but has historical roots in traditional media as well.”
Zuckerberg involvement curve
To tackle this issue, Pinterest collaborated with UC Berkeley and the Integrity Institute to develop a fresh “Manual on Non-Engagement Cues,” offering guidance on enhancing algorithms with positive signals beyond traditional interactions.
Ultimately, this will result in improved user experiences.
According to the guide:
“There is compelling proof that arranging based on anticipated involvement proves effective in enhancing user loyalty. Yet, loyalty can be boosted by integrating additional cues, such as proxies for item quality, and seeking input from users through item-level surveys. Moreover, diverse engagement serves as a valuable quality indicator.”
To tackle the initial aspect, Pinterest conducts user surveys to gather insights on the user journey.
Pinterest questionnaire
Majority of social networks employ similar techniques, collecting diverse feedback beyond basic interaction. These methods, though subtle, greatly influence the content ranking criteria by offering extra insights into user preferences.
Using Meta as a case study, in 2021, Meta highlighted that:
“A primary feedback from our community currently is the desire to avoid politics and conflict dominating their interaction on our platforms.”
Such sentiments may not be evident in standard engagement metrics, which typically highlight increased interactions with political content. However, such explicit feedback has the power to influence the fundamental strategies of social platforms, offering deeper insights into user preferences beyond quantitative data.
The manual categorization of content is highlighted as an additional method of assessing interests, albeit requiring substantial human effort to implement.
Aligned with this strategy, Pinterest emphasizes non-engagement cues to enhance user satisfaction, implementing diverse tactics to tailor and enhance individual feeds.
As an instance, our groundbreaking inclusive product initiatives heavily rely on Non-Engagement Signals. Upon receiving a user’s preferences regarding body type, hair pattern, or skin tone for their feed, Pinterest can prioritize content accordingly.
However, concentrating on these aspects does pose challenges.
Initially, prioritizing non-engagement signals may affect short-term retention and metrics.
Eliminating clickbait is beneficial, yet reducing clickbait results in fewer clicks, requiring platforms to adapt to potential initial data fluctuations.
Especially challenging for numerous publicly traded firms.
Pinterest highlights the importance of adjusting algorithms to promote emotional wellness, emphasizing the time and experimentation needed for successful implementation, which can be challenging initially but yields enhanced user retention in the long run.
Some of the factors influencing this may differ based on the platform, with Pinterest’s commerce-centered approach potentially providing an edge compared to X (previously known as Twitter), known for its emphasis on current news interactions.
However, the push is towards social media platforms enhancing user experiences by prioritizing both immediate and future objectives, rather than solely aiming for maximum interactions.
The manual also highlights that generative artificial intelligence “might be employed to enhance signal quality and introduce novel user controls” in the future, enhancing these aspects.
It presents a compelling analysis, outlining various ways to enhance social media engagement and the motivating factors that platforms employ to showcase their effectiveness.
Ultimately, emphasizing engagement brings about increased clicks; however, the resulting incentives can ultimately deteriorate a platform substantially.