MKT 375 Social Media Platforms

platform

an intermediary, connecting the buyers and sellers (two-sides of the market)

• one side of the market gains while the other side grows !

whats the importance of a platform

loyalty, rules of interaction, manage the demand on both sides so that each side has enough choices

decision 1

who should be allowed to join

decision 2

which side needs to be incentivized (the one that has more outside options)

decision 3

what content should people see

central goal of content filtering, ranking, and recommendation algorithms

sustained user attention ( social media platforms make money through ads, so the longer people are on the platform the more money the platform gets)

before 2015

chronological feeds

2016-2020

engagement-based sorting (personal taste, friends/connections)

2021-2024

AI-powered feeds (predict what the user will like next)

2025

real-time personalization

Instagram algorithm

social graph driven (relationship-based recommender), discovery layered

core signals of instagram

who you follow, past interactions, strength of tie, content engagement —> existing social connections

tiktok algorithm

user-content matching system

core signals of tiktok

watch time, rewatches, video features, early engagement velocity (shares in first minute) —> anyone can go viral

Linkedin algorithm

quality and credibility

quality (linkedin def)

Proprietary algorithm separates content into spam, low-quality and high-quality content

engagement (linkedin def)

probability: ML tools to predict engagement especially within the first hour

credibility (linkedin def)

Author expertise, insightful content, meaningful interactions

weak signals tiktok

social graph, follower count

algorithm Spotify

structured —> content-based filtering (all track, songwriter, artist information)

3 main unstructured factors of spotify algorithm

danceability, energy, valence

colaborative filtering

used by Spotify algorithm to help fetch content that is liked by similar users

algorithm youtube

more exploratory (5-10% of the time), minority viewpoints, new creators, long tail topics, exposure diversity

How do we know a platform monetizes well?

average revenue per user (ARPU) = revenue during a quarter/ daily active people

instagram, tiktok

attention monetized, advertisers pay

youtube premium

better experience monetized, users pay

meta verified, x premium

status/identity monteized, users/creators pay

linkedin premium

professional utility monetized, users/firms pay

tiktok shop

transactions monetized, sellers/merchants pay

memberships, gifts

creator-fan relationship monetized, fans pay

targeted advertising

primary source of income for social media platforms

platform revenue

= users*engagement*ad load*value per ad

recommendations matter…

because they increase engagement, inventory, resulting in greater revenue (attention holding)

more ads

results in more short-term revenue, worse user experience, less engagement, loss in long-term revenue

balance

maximize long-run platform value (not the number of ads)

freemium based model

most common, most features are free and you pay for some premium features

• which platform agents have willingness to pay?

ad removal

youtube premium, reddit premium

status & identity

instagram meta verified, x premium

power-user tools

linkedin premium, sales navigator

creator monetization

Tiktok gifts, youtube memberships