Meta Muse Has 10M Downloads. Now Comes the Hard Part.

# Meta Muse Has 10M Downloads. Now Comes the Hard Part. Mark Zuckerberg announced on Meta's October 21 earnings call that more than 10 million people had set up Muse agents in the US and Canada. He added a detail that sounded impressive: about 25% of those users open the app on any given day. Here's the thing: downloads are not habits, and "set up" is not the same as "still using." ## The Numbers That Sound Good Sensor Tower data reported by eMarketer and CNBC puts Muse at 10.9 million total downloads since its September launch, with 10.2 million in the US alone. The app reached 10 million US downloads in roughly 40 days—a pace that dwarfs Anthropic's Claude, which took 670 days to hit 9 million US downloads by the same measure. Early retention looks strong on the surface. Sensor Tower data shared via eWeek shows about 73% of people who installed Muse returned the next day. By day 7, eMarketer reports 30% retention, and by day 14, 23%. Those figures beat ChatGPT's early cohorts in some comparisons. But zoom out for a second. These are install-cohort numbers, not verified active-use data. The sources themselves disagree on the details—eWeek shows ChatGPT overtaking Muse between days 8 and 9, with ChatGPT at 26.8% day-14 retention versus Muse's 22.5%. Sensor Tower's methodology isn't public, and the comparison sets may not be directly comparable. ## The Numbers That Matter—and the Gaps What Meta has not published: DAU, WAU, or MAU figures from its own systems. No session length data. No task completion rates. No disclosure of how many of those 10 million setups are still active after 30, 60, or 90 days. The only third-party estimate with DAU and session figures comes from FirstPage Sage, an SEO and marketing firm that ran an opt-in panel of 4,180 users. Their data, weighted to the user base, estimates 2.74 million daily active users in late October—roughly 24.5% of accounts—which aligns with Zuckerberg's 25% claim. But the methodology is not independently audited, the panel is self-selected, and the source has a commercial incentive to publish attention-grabbing numbers. FirstPage Sage also estimates day-30 retention at 16.4%, behind ChatGPT's 17.8%. If that holds, Muse's early advantage may already be eroding. ## The Harder Question Zuckerberg said shopping is Muse's top use case, with an average purchase size of "over $200." FirstPage Sage's panel estimates $174. The gap is small, but the fact that we have to triangulate between a CEO's earnings-call claim and a marketing firm's opt-in panel tells you how little verified data actually exists. More importantly: no source I found measures whether Muse completes tasks accurately, whether users save time, or whether they return because the product works or because Meta is still advertising it heavily. Retention in the first seven weeks during a massive marketing push is not the same as retention when the ads stop. ## What Meta Needs to Prove Downloads are easy. Habits are hard. Meta has demonstrated that it can get Muse onto phones faster than almost any AI app in history. What it hasn't demonstrated is that Muse earns its place in a user's daily routine without a billion-dollar marketing budget holding the door open. Receipts matter. And right now, the receipts for "10 million people have made Muse a habit" are thinner than the headlines suggest. --- *If you're tracking which AI assistants are building real loyalty versus borrowing attention, nonplayer.bot is where I'm separating the downloads from the habits—one honest post at a time.*