The Market Exists. The Model is Yet to be Proven
A telling statistic summarizes the current state of friendship apps: in 2025, they generated $16 million in consumer spending in the United States, along with 4.3 million downloads. These numbers validate demand. However, when we break it down, the average revenue per download is less than $3.75. In comparison, the dating market—which closed 2025 at $11.61 billion globally—shows that the gap isn't about growth but about financial architecture.
The sociological context is clear: the Global Solitude Report 2025 indicates that 35% of adults feel chronically lonely, and the World Health Organization declared loneliness a public health threat in 2023. This represents a documented and urgent real demand. The pressing issue lies in conversion: translating an emotional need into a stream of recurring and predictable revenue requires an entirely different mechanic than dating apps, even though both share similar technical infrastructures.
Bumble BFF serves as the most instructive case. It launched its friendship feature in 2016 within its dating app, separated it as an independent product in 2023, rebranded it as Bumble For Friends in 2024, and by 2026, it operates with themed groups—the so-called "Circles"—for activities like tennis or book clubs. Researchers from the Internet Institute at the University of Oxford found that users utilizing this group function are 40% more likely to maintain active friendships at six months than those using individual matching. This figure matters, but not for the typical reasons many would analyze.
Why Oxford’s Retention Data Changes the Cost Equation
The 40% increase in retention in groups isn’t just a metric of satisfaction: it signals where the real business margin resides. Dating apps fundamentally monetize hope: users pay to keep trying. Friendship apps, when executed effectively, reduce the need for ongoing searching. This reverses the classic incentive of the freemium model.
If a user of Bumble BFF finds their tennis group in six weeks and doesn’t need to return to the app, the monthly subscription model has a very low ceiling for revenue capture. The logical response to this issue is found in the original article’s data: integration with Eventbrite and Ticketmaster, establishing a marketplace for activities within the platform. This shift turns the app from a mere matching product into a distribution layer for physical experiences. Revenue then stops depending on how long the user feels lonely and starts depending on how many activities they coordinate.
This distinction is crucial. In the first model, the user churn—who finds friends and leaves—is considered a loss. In the second model, that same user who finds friends and organizes group outings becomes a high-value client. The operational question Bumble Inc. is still answering is whether it can pivot successfully in time, before the marginal cost of acquiring new users exceeds the event transaction revenue.
Peanut, which focuses on women undergoing life transitions—motherhood, menopause, career changes—has a different and potentially more sustainable logic in the short run. Life-stage segmentation fosters communities with lower churn because the user’s identity is anchored to a prolonged condition, not a temporary state like "being single." This allows for monetization through niche memberships with a higher willingness to pay, akin to how specialized media operate compared to generalized ones.
The Trap of Scaling Dating Infrastructure for a Different Product
There’s a structural risk not immediately visible in the data but which emerges when analyzing product architecture: friendship apps are built on dating infrastructure, and this technical inheritance carries hidden costs.
The swipe mechanism was designed for markets with high churn and implicit rejection tolerance. In the context of seeking a partner, a no-match is socially neutral. In the context of seeking friends, particularly for populations already experiencing social anxiety or isolation, that same mechanism generates different friction. Bubblic documented in user tests that swiping works for low-commitment browsing, but the conversion to a real meeting remains the central bottleneck.
This is where the bet on AI with contextual compatibility—analyzing conversation styles, values, and availability—has its strongest business logic. Not as a marketing technology, but as a reducer of operational costs: if the system pre-filters more accurately, it lowers the number of failed interactions per user, which shortens the time to the first real meeting and, by extension, the abandonment rate before converting to a paying user.
Nextdoor operates under a different logic than the rest. Its advantage lies not in the algorithm but in physical proximity as a barrier to entry. A neighborhood cannot be replicated by another app. This gives it a defensive position that no interest-based app can easily match, but also limits its scaling capabilities beyond urban density and quality of local communities.
The Real Business Lies in Coordination, Not Matching
What the $16 million in consumer spending in 2025 suggests is not that the market is small, but that the value extraction model is still in the testing phase. Dating apps took nearly a decade to discover a freemium formula that works. Friendship apps are compressing that cycle with more data and better infrastructure, but the underlying problem is different.
Dating has a natural re-engagement cycle: a person loosens up again and returns to the app. Friendship, if the product is effective, doesn’t have that cycle. The sustainable model for this segment points toward platforms that charge for coordinating an active social life, not for searching for whom to have it with. Events, group activities, bookings, access to curated communities: that’s the layer where users are willing to pay repeatedly because the value is ongoing, not aspirational.
Apps that solve this architectural problem—transforming user success into recurring revenue for the platform instead of churn—will be the ones to justify valuations comparable to the dating market. Those that do not will continue reporting record downloads with margins that don’t add up.
