Meta's Content Monetization Program for AI creators
Reels Play is gone. Here is what the unified Content Monetization Program requires from AI content, and which Meta surface actually repays the upload effort.
The Reels Play bonus ended on 31 August 2025. What replaced it is structurally different in a way that changes which content is worth making: instead of a bonus paid for hitting view targets, Meta now runs a unified Content Monetization Program that folds in-stream ads, Reels ads, Stars and bonuses into a single payout.
For AI-first creators, that merge matters more than it sounds. A bonus program pays for views. An ads program pays for ad impressions against views, which is a different thing, and it rewards a different shape of video. The formats that farmed Reels Play — short, loopable, high-volume — are the worst-performing shape under the program that replaced it.
What the merge actually changed
| Before | After |
|---|---|
| Reels Play bonus, paid on view milestones | Reels ads revenue, paid through the unified program |
| In-stream ads as a separate program | In-stream ads inside the same payout |
| Stars managed separately | Stars inside the same payout |
| Invite-based bonuses per surface | One program, one balance |
The practical consequence is that you stop optimizing per-program and start optimizing per-surface, because the surfaces pay at wildly different rates and now sit in the same statement where you can compare them directly.
One more change worth internalizing: the program weights shares and comments above likes, and it does not count every play. Reported thresholds are around three seconds for a Reel and thirty seconds for long-form video, with bounce views and bot traffic excluded. A loop that plays four times in twelve seconds was excellent under a view-milestone bonus. Under an ads program it scrapes over the Reels floor and earns cents, because the money sits in the long-form tier where the threshold is ten times higher.
The two bars AI content has to clear
Meta allows AI-generated content in the program. It attaches two conditions, and they are the ones to design around.
1. The content must be original to the creator. Not licensed, not reposted, not a re-upload of something that circulated elsewhere first. This is the same substance as YouTube's originality standard even though the wording differs: the platform is filtering for content that exists because you made it, not content that exists because a template ran. If your Facebook page is a mirror of your TikTok archive, that is a re-upload problem before it is an AI problem.
2. The content must be labelled. AI-generated material carries a disclosure requirement, and the label is not optional because the payout is attached. Getting this right once and applying it mechanically is far cheaper than doing it per-post — the cross-platform labelling checklist maps one disclosure decision to each destination, and the AI content label entry covers what the label is actually asserting.
What is not on this list: any specific model, any restriction on generated footage as such, any prohibition on synthetic voice. The bars are provenance and disclosure, not technique.
The account-level thresholds — followers, watch time, eligible country — are set per surface and Meta adjusts them. Read them in your Professional Dashboard rather than from a blog, including this one. Any specific follower number you see quoted for CMP eligibility is someone's snapshot of a moving target.
Which surface is worth the upload
Here is the part that decides your production plan. Meta publishes no RPM figures, so everything below is third-party reported and should be treated as an order-of-magnitude signal, not a rate card.
| Surface | Reported rate | Who reported it |
|---|---|---|
| Facebook long-form, in-stream ads | $0.50–$8+ RPM, widening with niche and geography | KiwiBox |
| Facebook Reels | ~$0.02–$0.20 per 1,000 views | KiwiBox |
| Facebook Reels, competing estimate | $0.30–$5.00 RPM, split by niche | fluxnote |
| Instagram Reels | No direct ad revenue share on Reels | fluxnote |
Two caveats before you draw the obvious conclusion. The first is that the two Reels estimates disagree by more than an order of magnitude, which is the most honest thing on this page: nobody outside Meta knows, and a number quoted to two decimal places from creator self-reporting is still a guess. The second is that these are creator-side RPMs, after Meta's cut — KiwiBox puts the split at roughly 55% to the creator — so do not apply a further percentage to them.
With both caveats applied, the ordering is still not close. Facebook long-form carrying in-stream ads is reported in dollars per thousand views. Facebook Reels is reported in cents on one estimate and low dollars on the other, and on either estimate its range sits below the long-form range. Instagram Reels is reported as having no direct ad revenue share on Reels at all — whatever it earns arrives through branded content, Stars or off-platform conversion, not an ad split.
The verdict for an AI-first operation: Facebook long-form video carrying in-stream ads is the only Meta surface where the upload effort is repaid at a meaningful rate. Facebook Reels is a distribution surface that pays a little. Instagram Reels is a distribution surface that pays approximately nothing, and should be judged entirely on whether it builds an audience you monetize elsewhere.
That is a genuinely unintuitive conclusion for creators who have spent two years treating Reels as the centre of Meta strategy — and it is the same conclusion the case for Facebook video reaches from the distribution side. If you are producing vertical clips and cross-posting everywhere, the highest-value change you can make is to also produce one longer Facebook piece per week that can actually carry ad breaks.
Building for the counted view
If the long-form Facebook piece is where the money is, the 30-second counting threshold is the production constraint that follows. Three rules fall out of it:
Front-load a reason to stay, not a reason to loop. Loop-bait wins on a bonus program and loses on this one. The hook has to promise a payoff that arrives after thirty seconds, and then deliver it.
Write a narrative spine, not a montage. Generated footage assembles very easily into thirty seconds of pretty shots with no through-line, and that is exactly the shape that gets abandoned at second eight. A reel built from a script with a beginning and a payoff holds where a montage does not.
Instrument the drop-off, not the view count. Completion rate and the shape of the retention curve tell you whether you are clearing thirty seconds. Raw views do not, because a play that ends at twenty-nine seconds is not a counted view on the long-form tier.
For the long-form Facebook piece specifically: length has to be sufficient to carry ad breaks, and the breaks have to land somewhere the viewer is willing to sit through them. That means chaptered structure with real beats, which in turn means writing to a runtime before generating anything.
Once the format is settled, the cadence is a scheduling problem — one longer Facebook upload plus the vertical cuts, scheduled across destinations from one production pass, with the label applied per destination.
FAQ
Is AI-generated content allowed in Meta's Content Monetization Program?
Yes, with two conditions: it must be original to you rather than reposted or licensed, and it must be labelled as AI-generated. There is no restriction on which tools produced it. The failure mode that actually costs creators money is the originality one — mirroring an archive from another platform onto a Facebook page reads as re-uploaded content regardless of who made it originally.
Did the Reels Play bonus come back in another form?
No. It ended on 31 August 2025 and the bonus mechanic was folded into the unified program rather than reissued. Bonuses still exist inside the program, but they are one component of a combined payout rather than a standalone view-milestone scheme, so planning a channel around bonus arbitrage no longer works.
Why does Instagram Reels pay so much less than Facebook?
Because the two surfaces monetize differently rather than at different rates. Facebook carries in-stream ad inventory against longer video; Instagram Reels is reported as having no direct ad revenue share for creators at all, so there is no equivalent line to compare. Judge Instagram on audience-building and off-platform conversion, and see Instagram Reels growth for what that looks like as a plan rather than an afterthought.
Are the rate figures in this post official?
No. Meta publishes no RPM or CPM figures for creator payouts. Every number here is third-party reported from creator self-reporting, which is why the two Facebook Reels estimates in the table above disagree by more than an order of magnitude. Use them to rank surfaces against each other, not to forecast revenue — and if you want the underlying distinction, CPM versus CPV explains why the two are not interchangeable.