Meta Ads Manager Issues September 22, 2026:
What We Know So Far

What Meta has said

One post, twenty-eight words, no scope:

Note what is missing, because it is missing every time: no affected regions, no share of accounts, no campaign types, no estimate. "May be impacting" is doing a lot of work. The only hard information in the post is the severity tag and the timestamp.

On that tag - Meta's scale runs Low, Medium and High disruptions, and it describes Meta's view of the blast radius, not yours. A Medium incident that happens to catch your top-spending campaigns is a bad afternoon regardless of the label. Treat the tag as a signal that something real is broken, and your hourly data as the measurement.

Timeline

For reference, Meta's three most recent acknowledged Ads Delivery incidents each ran between roughly 1.5 and 2 hours from first post to resolution. That is a pattern, not a promise - and the acknowledged window is only ever the tail of the real one, since the clock starts when Meta posts, not when delivery degrades.

The status feed was on time this round

We track this because it broke badly once. During the July 16 Ads Delivery outage, Meta's dashboard carried the incident at 09:40 UTC while its RSS feed did not publish the entry until roughly 11:11 UTC - a 91-minute hole that covered most of a two-hour outage, and that every RSS-based monitor fell straight into.

Today the two sources agreed exactly. That is worth saying out loud: Meta's publishing pipeline is capable of being instant, which makes the July gap look like a fault rather than a design. We stopped depending on the answer either way - AdStatus has polled Meta's dashboard data source alongside the RSS feeds since July, so alerts fire on whichever source moves first.

4:33 PM ET is the expensive version of this

Meta's last two delivery incidents landed while North America was asleep. This one did not.

The upside of a business-hours incident is that you can see it happening. The downside is that it is happening to your best hours. If you run US traffic and your afternoon looks soft today, you now have a timestamp to put next to it.

Meta's quiet September just ended

Six days ago we published our September 1-15 reliability report. Nine incidents across Google, Microsoft, Amazon, Shopify and OpenAI. Meta logged zero - its last entry before that was August 21.

We flagged that silence rather than celebrating it, for a specific reason. On September 15, 2025, Meta logged an Ads Delivery outage. It turned out to be the first of eleven incidents over the following eight weeks - the pre-BFCM run that made Q4 2025 the worst stretch of Meta reliability in our records.

One incident is not a cluster, and we are not going to pretend otherwise. But it is the same surface, at the same point in the calendar, heading into the same peak season - and it arrives after three weeks in which Meta's status page said nothing while the rest of the industry logged nine problems. Worth watching over the next fortnight rather than filing and forgetting.

For what a bad Meta peak season actually looks like from the advertiser's seat, see our BFCM 2025 breakdown.

What to do right now

1. 1 Do not restructure anything. This is the one that costs real money. Soft delivery during a platform incident looks exactly like soft delivery from a bad audience or a tired creative, and the instinct is to start editing. Every budget change, bid change or creative swap you make in this window risks a learning reset you will not be able to attribute later - and the platform issue will resolve on its own.
2. 2 Screenshot the hourly view now. In Ads Manager, break spend and impressions down by hour from roughly 3:30 PM ET onward and capture it. Live evidence is the kind you cannot go back for, and it is what any later credit conversation will turn on.
3. 3 Watch for the catch-up spike after recovery. The classic delivery-incident signature is a suppressed hour or two followed by aggressive pacing to make up the daily budget. The dip is the platform's fault; the spike lands on your card. If you have thin margins on a campaign, that is the moment to be watching it.
4. 4 Hold today's numbers before you report them. Delivery disruptions routinely come with attribution gaps. Cross-check this afternoon's conversions against your own backend before anyone draws a conclusion from a September 22 CPA.
5. 5 Tell the client before they tell you. A two-line note with the incident timestamp and the metastatus link, sent while it is happening, is worth more than a perfect explanation tomorrow.

On the budget mechanics specifically - how overspend accumulates during delivery incidents and why refund requests usually fail - see our March 16 incident analysis.

Sources

• Meta Status - Ads Manager dashboard
• Meta Ads Manager RSS feed
• AdStatus: September 2026 Platform Reliability Report (Sep 1-15)
• AdStatus: July 16, 2026 Ads Delivery Outage & the 90-minute RSS lag
• AdStatus: March 16, 2026 Ads Delivery Outage & Budget Risk