Meta Muse Ad Budget 39%: What Sensor Tower Data Does and Does Not Show

A headline on October 20, 2026 said Meta "raises Muse ad budget to 39%". It sounds like Meta is putting ads inside its Muse agent. It is not. The figure comes from a Sensor Tower estimate that Meta now directs 39% of its own advertising spend toward promoting Muse, up from 23% in September, as relayed by Trending Topics. Meta is the advertiser, and Muse is the product being advertised.

This post sorts the claims in that report into what is an estimate, what Meta has said itself, and what is still unknown. If you want the earlier story on Muse's growth, see our posts on its App Store climb and the reported plateau in daily users.

What the report claims, and what kind of claim each one is

Every number in the first seven rows traces to one data provider. Sensor Tower describes its figures as estimates meant to show trends, and we found no second outlet confirming the 39% figure. We also found no Meta statement on its own advertising allocation. Treat the numbers as plausible and directional, not audited.

Ads for Muse versus ads in Muse

Ad dashboard card with coin slot, representing Meta's reported Muse ad budget

The two readings of "Muse ad budget" lead to opposite conclusions, so it is worth being precise.

• Meta buys ads to promote Muse. This is what the Sensor Tower figure describes. A brand that moves from the 73rd-largest US advertiser by impressions to the 21st in a month is spending heavily to drive installs. It is consistent with the download surge.
• Muse shows ads to users. Meta says it does not. Per Digiday, the company has said Muse does not share users' conversations or browsing sessions with its ad systems, and Muse has a free tier plus $20 and $100 monthly plans.
• Muse earns on transactions. Zuckerberg has floated a small fee paid by merchants, not users, when Muse helps someone buy, sell or save money. Digiday calls this a toll on transactions rather than an ad auction.

Our earlier explainer, what Meta got right with Muse, covers the ad-funded token economics argument. Whether Muse ever shows ads is an open question, and Digiday frames it that way, but nothing confirmed says it will.

The retention numbers cut both ways

Magnifier over a claim bubble with a check mark, used to verify Meta Muse retention claims

The headline retention number looks good: 30% of September installers still active after seven days against an average of 19% for ChatGPT, Grok and Claude. But the report adds two details that complicate it.

1. After the first day, 73% of Muse users returned. After 14 days only 23% were active.
2. Rivals lost about 12 percentage points over that same stretch, while Muse lost about 50.

In plain terms, Muse converts curiosity into a first week of use better than its rivals do, but it loses users faster after that. This is the pattern you would expect from heavy paid promotion: ads bring in many people who try the app, and a share of them drift away. It does not prove the product is weak, since a launch-month cohort is mixed with novelty users, but it explains why Meta would keep spending to refill the funnel.

Sensor Tower's comparison with other apps over the first 40 days is striking: ChatGPT had 3.7 million US downloads in its first 40 days, Grok 500,000 and Claude 200,000, per the same report. Those are launch-curve comparisons from different eras and distribution channels. Muse arrived with Meta's reach and, per the same estimates, heavy paid promotion, so the comparison measures distribution as much as product quality.

Why the money matters to the business case

Meta's reported 2026 AI capital spending guidance is $135 billion to $145 billion, roughly double the prior year, and Digiday reports that by the second quarter it consumed 98% of operating cash flow. Spending a growing share of advertising dollars on Muse is one way to show that the investment reaches users. It is also a cost.

The report also notes a revenue challenge: Meta plans to earn mainly from transaction fees, but US banks and brokers are blocking AI agents like Muse from their accounts. If agents cannot complete purchases and payments, the fee model has less to charge on. That connects to the wider fight over agent access that we covered in Paul Graham's take on Amazon banning agents.

Privacy context readers should keep in mind

Promotion is only one part of the Muse story. Two incidents are relevant if you are deciding whether to install it: the Time investigation into contact dossiers and the pause on Muse share links after exposed chats. Meta's statement that conversations are not shared with ad systems is a company position; the investigations above are the independent checks to read alongside it.

A short timeline of the Muse launch

The sequence matters. The ad budget share rose from 23% to 39% in the same weeks that the daily user count stopped climbing, which is at least consistent with promotion being used to hold growth up. That is our reading of two separate estimates, not something either source states.

What builders and marketers can take from this

If you build an agent product, the lesson is the retention shape, not the download count. A first-week retention of 30% is strong; a 14-day figure of 23% suggests the second-week job-to-be-done is weaker. Agents that earn daily use usually have a recurring task (inbox, calendar, shopping list) that the user hands over every day. Track how many users delegate a second and third task, as we argued in the plateau post.

If you advertise, note that Muse is not an ad channel. The small-business integrations let Muse analyze ads and draft campaigns through connected ad accounts, but whether it can edit budgets or bids is not stated in the announcement, as far as we could find. Nobody can buy placement inside Muse conversations today.

If you follow Meta's finances, the 39% share is a signal about internal prioritization: Meta is using its own ad inventory to promote an AI product that does not yet carry ads. The business case rests on transaction fees that depend on third parties letting agents in.

Why "claimed vs verified" matters here

The mix-up between "ads for Muse" and "ads in Muse" is the sort of error that spreads because the headline is short. A reader who sees "Meta raises Muse ad budget" can reasonably conclude Muse is monetizing through ads, which would be a major shift for an assistant marketed as ad-free. The accurate reading is narrower: a data vendor estimates that Meta bought more ads for its own app. When you see a figure like this, ask who measured it, what the unit is (share, dollars, impressions), and whether the company has confirmed it.

How to check the claims yourself

1. Look for Sensor Tower's own publication of the data, since the figures here are secondhand through a news summary.
2. Check Meta's Ad Library for Muse campaigns, which shows active ads by advertiser. It will not give budget shares but will show whether Muse promotion is running at scale.
3. Compare with Apple App Store and Google Play rank histories for Muse. The report says it led Google Play for 30 days and sat at number one on the App Store on all but one of them.
4. Ask whether the claim is about ads for Muse or ads in Muse before repeating it.

What we do not know

• Meta's actual dollar spend on Muse promotion. The 39% is a share estimate, not a dollar figure.
• Whether Sensor Tower's ad-budget model covers Meta's own properties or only external placements.
• How much of the 2.5 million daily users are paying subscribers on the $20 and $100 tiers.
• Whether the post-launch drop-off will stabilize.

We will update this post if Meta or Sensor Tower publishes more detail.

Related reading

• Meta Muse hits number one on the App Store
• Muse daily users plateau
• What Meta got right with Muse
• Muse invite codes and a billion tokens per user
• Time investigation: Muse contact dossiers
• Muse share links paused

Figures are Sensor Tower estimates as reported on October 20, 2026 and have not been confirmed by Meta.