Meta Ads Issues in April 2026: Pixel, Tracking, and Delivery Problems Operators Are Hitting

Last updated: April 22, 2026. This is a live log of the most common Meta Ads issues we are seeing across customer accounts and operator forums right now. Updated as Meta ships fixes or as new bugs surface.

If you landed here because something is broken in your Meta Ads account this week - you are not alone. April 2026 has been the worst stretch for Meta Ads stability since iOS 14.5 dropped. The list below covers what is breaking, what is a real bug vs operator misconfiguration, and what the fix looks like as of today.

Top issues we are tracking right now

Issue 1: Pixel events not showing in Events Manager

Symptoms. Your pixel fires on the page (Tag Assistant confirms), but the event does not show in Events Manager - or shows with a multi-hour delay. This started in late March and got worse the week of April 14.

What is actually happening. Two issues are tangled together. First, Meta's Aggregated Event Measurement (AEM) priorities resilient to pixel code updates have started to silently reset on some accounts. Second, the Andromeda delivery system rollout has changed how events deduplicate against the Conversion API - some accounts are seeing CAPI events double-counted, which then gets corrected by AEM dropping events Meta deems duplicate.

Fix order:

1. Open Events Manager > your pixel > Settings > Aggregated Event Measurement.
2. Verify your top 8 events are configured and prioritized. Re-set if anything looks off.
3. Open Events > recent activity. Filter by your event name. Check that it is firing within minutes, not hours.
4. If using Conversion API: open Test Events. Send a test event with a known event_id. Confirm dedup is working.
5. If still broken: file a support ticket via Business Help Center. Reference April 2026 AEM regression - support is escalating these faster.

Issue 2: Andromeda delivery and ROAS regression

Symptoms. Reported ROAS in Meta Ads Manager dropped 15-40% across accounts that did not change anything. CPM looks similar to last month, but conversions reported are lower.

What is actually happening. Andromeda is Meta's new ad delivery and personalization infrastructure that started rolling out December 2025 and reached most accounts by mid-March. Andromeda is supposed to improve performance long-term, but in the rollout window it is causing two problems for many accounts:

• Reporting accuracy regression. Andromeda is more conservative about which conversions it attributes back. ROAS reports look lower; actual blended business performance often unchanged when checked against GA4 or server-side analytics.
• Audience targeting churn. Andromeda's personalization model is re-evaluating audience signals. Some accounts see audience size estimates shifting daily and learning phase taking longer.

Fix order:

1. Triangulate ROAS with GA4 and your CRM/server data. Most accounts find blended performance is unchanged - the drop is reporting, not real.
2. Check the Attribution Setting at the campaign level. Some accounts see this reset to default 7-day click + 1-day view; if you previously ran on a different window, your reported numbers will look very different.
3. Hold off on major budget cuts until the rollout stabilizes - typically 4-6 weeks per account. Cutting based on reported ROAS during the regression often causes real performance to drop.
4. Check campaigns using Advantage+ Shopping or Advantage+ Audience. Andromeda is most disruptive to these.

We have a deeper post on this specifically at /blog/meta-andromeda-issues-may-2026 - the Andromeda regression is the single biggest issue facing Meta operators in April-May 2026 and worth a dedicated read.

Issue 3: Learning phase resetting on minor edits

Symptoms. Every time you adjust the budget by 5% or swap a creative variant, the ad set goes back into Learning. After 3-4 cycles, the campaign never gets out of Learning Limited.

What is actually happening. Meta tightened the "significant edit" threshold during the Andromeda rollout. What used to be a non-resetting edit (small budget change, minor headline swap) is now sometimes counted as significant, restarting the 50-conversion learning window.

Fix:

1. Bundle edits into a single window once per week. Multiple edits across a week can each trigger Learning resets.
2. Use 'Duplicate to test' for major creative changes - this leaves the original ad set learning while the duplicate explores.
3. Check your conversion volume. If under 50 conversions per week per ad set, Learning Limited may not be a bug - it is the floor.

Full troubleshooting on Learning Phase issues at /blog/meta-ads-learning-phase-stuck.

Issue 4: False ad disapprovals on retail accounts

Symptoms. Multiple ads disapproved for "low quality" or "engagement bait" with no recent change to creative or copy. Started in volume on April 14.

What is actually happening. Meta retrained one of its ad-quality classifiers and rolled it out to retail accounts mid-April. The new classifier is over-aggressive on certain ad formats: carousels with promotional copy, short-form video with text overlays, and product catalogs with "X% off" language.

Fix:

1. File an appeal immediately. Most disapprovals are reversing in 24-48 hours after appeal.
2. Do not edit the disapproved ad to try to fix it - this often confirms the disapproval. Appeal first.
3. If multiple ads disapprove on the same account in a 24h window, file a Business Help Center ticket - account-level review may be needed.
4. Document everything. Screenshots of the disapproval reason and timestamps help if account-level escalation is needed.

Issue 5: Conversion API and pixel deduping breaking

Symptoms. Conversions inflate (pixel + CAPI both counting the same event) or deflate (neither counts). Started showing up in early April.

What is actually happening. Andromeda changed the dedup window logic. Events that previously deduped within 60 seconds via event_id are now sometimes treated as separate if other parameters differ.

Fix:

1. Check Events Manager > Test Events. Send a test conversion through both pixel and CAPI with the same event_id.
2. Look at the Deduplication Score in Events Manager. Should be 80%+ for healthy setup.
3. Verify event_id is consistent. A common failure: pixel uses one event_id (often timestamp-based) and CAPI uses another.
4. Check fbp and fbc cookie passing. CAPI events that arrive without these cookies cannot dedupe against pixel.

For full setup guide on Conversion API: /blog/what-is-conversion-api.

Issues 6 to 10: Smaller bugs

How Hyper helps

Most of the issues above require checking Events Manager, AEM priority lists, attribution settings, and CAPI dedup scores - manually, every time something looks off. Hyper agents monitor all of these continuously and surface the issue with the fix already attached. The case study at /blog/ai-marketing-case-study shows the full monitoring layer.

For the Andromeda regression specifically - we have a Hyper Effects YouTube walkthrough showing how the agents detect ROAS drops that are reporting-only vs real, so operators do not panic-cut budgets during the rollout window.

Frequently asked questions