Muse cancels subscriptions. Your retention math just broke.

If part of your recurring revenue comes from customers forgetting, an agent is about to audit it. CNBC reported that Meta's Muse agent, given access to bank and card statements, turns into a budgeting coach that finds recurring charges and cancels the ones the customer no longer uses. It will also draft the message asking for a lower price or a retention offer. Subscription churn has always been throttled by hassle. Muse is a hassle-removal machine, and it is shipping to consumers in the US and Canada right now.

What actually happened

The mechanic is unremarkable and that is the point: hand an agent your statements, and it reads every recurring line item and tells you which ones you have not used. The friction that used to protect the charge — remembering, finding the account, clicking through a four-step retention flow — is exactly the kind of work agents are good at.

The numbers behind why this matters are not Meta's. Mastercard and FT Strategies found 44% of US consumers spent more on subscriptions in 2025, averaging $1,887 a year, up from $1,416 in 2024. The same research puts average monthly churn at 20% and reports that 74% of consumers are more likely to subscribe when cancellation is simple.

Then there is the uncomfortable one. In Selling Subscriptions, Stanford economists Liran Einav, Ben Klopack, and Neale Mahoney estimate that inertia and cancellation friction roughly double subscription revenue. Their identification is clean: they look at months when a payment card is replaced and renewal has to be active rather than automatic. In those months, Mahoney told CNBC, people are about four times more likely to cancel.

Read those two findings together. Half of subscription revenue, industry-wide, is a tax on forgetting. An agent that reads statements is a permanent card-replacement event.

Why agent-driven cancellation matters for your business

Separate your real retention from your friction retention. Most operators have one churn number and treat it as a measure of product value. It is two numbers wearing one coat: customers who stay because it is worth it, and customers who stay because canceling is annoying. Only the first survives contact with an agent. Pull the cohort that has not logged in for sixty days and is still paying — that is the exposure, quantified, and you can do it this afternoon.

A drafted retention request is a probe of your discount ladder. When agents start writing "I'd like to cancel unless you can do better on price," you are not handling one negotiation. You are handling a systematic, tireless scan of every save offer your support team is authorized to make. Decide your floor and write it down, because your support rep is going to be asked for it fifty times a week instead of twice.

The friction strategy now loses on both sides. Making cancellation hard is already regulated in several markets and is now being routed around technically. And the Mastercard research says 74% of consumers are more likely to subscribe when canceling is simple — the friction was never free, it was suppressing acquisition to protect churn.

Build the pause. A customer whose agent is hunting for savings will accept "pause for three months" far more readily than a human clicking through a cancellation flow will, because the agent is optimizing a number, not escaping a conversation. Pause keeps the relationship and the payment method. Cancellation loses both.

If you sell retainers, this reaches you too. A monthly retainer is a subscription. Agents auditing business card statements will find yours, and the defense is the same as it always was: a customer who can see what they got last month does not need an agent to tell them whether it was worth it. Send the invoice with the work attached.

Sources: CNBC, Mastercard and FT Strategies, Einav, Klopack & Mahoney, "Selling Subscriptions," American Economic Review.