Adam Mosseri on Launching Threads: Social Graphs, Creators and Retention

Harry Stebbings talks with Instagram head Adam Mosseri about Threads shortly after its launch. Mosseri says the initial adoption exceeded his expectations, but warns that sign-ups and lasting use are different achievements. The conversation describes the product and plans as they stood in 2023.

Launching a new app is more likely to fail than succeed, Mosseri says, even when the early response is encouraging. He was excited about Threads but considered it a significant risk; the first eight days were only the start of a retention challenge.

Stebbings asks how the team bootstrapped the social graph. Mosseri explains that people could sign in with Instagram, carry over their identity and choose to follow accounts they already followed there, including accounts whose owners had not yet joined Threads.

That connection also allowed Instagram to tell a potential user that followers were already waiting on Threads. Mosseri considers this a compelling invitation, but says importing the graph was optional and users could curate different follow lists.

He expects the people someone wants to follow for public conversations to overlap with, but not necessarily match, the accounts they follow for photos and videos. At the interview, he anticipated a future option to keep the graphs in sync while preserving the ability to let them diverge.

The team discussed making everyone start without followers. Mosseri says they instead judged that enough people would find an imported graph useful, while allowing individuals to select only some accounts or build a different experience.

He compares the decision with earlier standalone Facebook experiments. Facebook Home and Paper, he says, were not sufficiently different from Facebook itself; other experiments were different but did not make use of the company's existing strengths or product connections.

In his view, Threads needed to be distinct enough to justify a separate app while benefiting from Instagram's existing network. He doubts the initial launch would have been as successful if every relationship had to be rebuilt from scratch.

Stebbings asks whether an interest graph has replaced the friend graph. Mosseri rejects a strict either-or: whom people care about, whom they want to hear from and what they engage with each day can differ. He describes a tension between immediate behavior and longer-term preferences.

Mosseri uses Reels and direct messages as an example of interest and friendship working together. A person may discover a video from a stranger, share it with a friend, and start a conversation that brings both back to the app.

He says video also creates business tradeoffs. Watching one video can take the time a user previously spent scrolling through several posts and seeing an ad; in his account, video tends to increase time spent while putting pressure on revenue and some other engagement measures.

The integrated inbox, he argues, can turn video discovery into sharing and conversation. He presents this as a product flywheel rather than proof that entertainment and social connection never conflict.

Asked whether the same combination was already working on Threads, Mosseri says it is too early to tell. At eight days old, his immediate concern is keeping the service running and learning how people use it.

Stebbings asks about creator activation. Mosseri describes the familiar two-sided problem: users need content when they arrive, while creators need an audience. His biggest launch concern was that people would open Threads, find nothing worth reading and leave.

The team focused on having a critical mass of creators and posts early, drawing partly on lessons from IGTV. Mosseri says feeds tend to rely heavily on content from a relatively small group of producers, so the initial mix can influence the tone of a service.

The partnerships team chose categories where Instagram already had active creators and commentary communities, including athletes, musicians, fashion and beauty. Mosseri says the launch effort did not specifically target technology, news or politics, though the service was not intended to exclude those groups.

Asked whether creators were paid to join, he says no. The team considered it, but he argues that paying everyone would not scale and that a durable platform must offer creators enough value to participate. He distinguishes that from targeted incentives used to learn or bootstrap a product.

A launch delay left the partnerships team with people ready to participate on short notice. Mosseri says inviting several thousand people shortly before launch helped build anticipation, alongside the choice to focus on particular communities.

The team then discovered that iOS preorders would be delivered at midnight in each local time zone, without the launch time being controlled as expected. In Japan the app appeared in the store before people could log in, forcing a choice between canceling preorders and moving the launch forward.

They opened the service at midnight UK time because the UK was the next major market to reach that point. Mosseri says the unexpected schedule change also contributed to the excitement around launch; it was a response to an operational constraint, not simply a planned tactic.

Stebbings asks whether bringing established Instagram personalities to a new platform amounted to importing an existing elite. Mosseri does not accept the label, but says Instagram already had a large creator community the team could reach. He also notes that creators increasingly use multiple platforms to diversify their risk.

Once a sufficient number of interesting people joined, he says, others wanted to participate too. He names Beyoncé and Taylor Swift as culturally influential people he would have liked to see on the service, and mentions football and Formula One as personal interests.

At that point, Threads had not launched in Europe. Mosseri says the delay mattered both for reaching users and for the availability of European football communities. This was a statement about the rollout at the time of the interview.

On creator income, Mosseri says paying every creator directly would be difficult to sustain. He favors tools such as subscriptions and branded-content partnerships, while allowing that targeted payments can help bootstrap participation or reward incremental value.

He also questions whether irregular, lottery-like payments give creators enough predictability to plan a business. His comparison of platforms reflects complaints and observations he had heard at the time, not a complete accounting of their compensation systems.

Asked for a Threads equivalent of an early Facebook retention threshold, Mosseri says the team does not yet know the number. He expects both a sufficient set of friends and a sufficiently useful interest graph to matter, with recommendations helping a new user find content.

He recalls that earlier Facebook thresholds differed greatly by country because culture and internet usage varied. The broader lesson for his team is to measure actual behavior rather than assume one universal number for every market.

Retention is the primary challenge for the next few months, Mosseri says. At the interview, he identifies a following feed, an edit button and European availability as missing or incomplete, and expresses interest in supporting ActivityPub and federation.

He says European rollout will require substantial work on data separation and verification of claims about data handling under upcoming laws. His estimate of many months is a historical prediction from the launch period, not a statement of present availability.