Fix Meta Ads delivery and learning issues

#01Diagnostic order (do not skip)

#02Learning limited playbook

1. Confirm optimisation event volume last 7 days in Events Manager (not only Ads Manager).
2. Consolidate ad sets that share the same event and offer.
3. Raise budget toward the learning threshold or temporarily optimise to a higher-funnel event.
4. Stop daily creative swaps; batch changes.
5. Ensure CAPI+Pixel deduped events are not undercounting.

#03Not spending playbook

#04High CPA playbook

1. Verify you are optimising to the right event (Lead vs Schedule).
2. Refresh diversified creatives (lessons 8–9, 16).
3. Check landing conversion rate in GA4 — media cannot fix a broken form.
4. Review placement quality before blanket exclusions.
5. Run Lift before declaring the channel dead (lesson 15).

#05Worked example: Ledgerline

Ledgerline is a fictional accounts-payable software company. Details are illustrative.

Walkthrough

1. An agency duplicated campaigns thrice "to beat learning." Volume fragmented further.
2. Ledgerline in-house ran the diagnostic order, fixed CAPI first, then structure.
3. Delivery recovered in a week without new interest stacks.
4. Competitor A kept editing bids hourly; charts never stabilised.

#06Escalation path inside your team

Write this path down so freelancers do not silently create duplicate Pixels.

#07Diagnostics order that saves hours

1. Is the ad actually running or in review?
2. Is the account payment healthy?
3. Are events firing (Test Events)?
4. Is the audience too narrow (special category + exclusions)?
5. Is creative approved and diverse?
6. Did someone edit optimisation or bid strategy yesterday?

Most "Meta is broken" moments fail at steps 3–5.

#08When to recreate vs repair

Repair when assets and history are valuable (Pixel, catalogs, established ad accounts). Recreate only when the account is irreversibly restricted or ownership is corrupted. Recreating Pixels casually is a common self-inflicted wound (Pixel lesson).

#09Learning limited — deeper playbook

If an ad set is Learning Limited:

1. Confirm weekly optimisation-event volume vs Meta's ~50/week guidance for exiting learning.
2. Consolidate ad sets that split the same event.
3. Broaden Advantage+ audience constraints only where policy allows.
4. Check that the optimisation event is not too rare (upgrade event only with CAPI readiness).
5. Avoid editing every day "to help."

If you need rare events (SQO), consider optimising for a higher-funnel event while reporting on SQO in CRM until volume supports the rarer event.

#10Delivery chart reading tips

Flat delivery after a change often means review, payment or severe audience constraints. Spiky delivery can mean budget pacing, auction volatility or creative intermittent rejection. Screenshot the delivery chart into the incident log with the change you made that day.

#11Pixel and CAPI incidents that look like delivery problems

Symptoms: sudden CPA spike, fewer results, or "no conversions" while clicks continue.

Checklist:

1. GTM/container publish errors (GTM versions)
2. Dedup broken (double counting or zero counting)
3. Domain or consent mode changes blocking tags
4. Event name renamed without updating Ads Manager optimisation
5. Test Events still works but production traffic filtered

Open an incident with analytics before rewriting the whole campaign structure.

#12Audience size illusions

Estimated audience size in the UI is not a promise of cheap delivery. Special categories, exclusions, languages and placement limits stack. If delivery is thin, export the effective constraints rather than raising bids blindly.

#13Creative rejection loops

If every new ad rejects:

• Check Page publishing authority
• Check for trademarked terms
• Check landing page redirects and cloaking suspicions
• Try a minimal text/static ad to isolate creative vs account issues

Persistent unexplained rejections warrant Meta Business Support with business verification documents ready.

#14Common mistakes

• Bid panic before Events Manager checks.
• Audience paranoia excluding everyone interesting.
• Creative neglect for months then blaming Advantage+.
• Unlimited admins causing mystery edits.
• No change history review.

#15Try it yourself

Pick one underperforming ad set. Run steps 1–5 of the diagnostic order and write the first failing step. Fix only that step this week.

#16Quick quiz

1. What is the first diagnostic layer?

Answer: Account health — payments, limits and account quality — before bid tweaks.

2. Why does duplicating campaigns often fail?

Answer: It resets learning and splits conversion volume needed to exit learning.

3. What budget maths relates to learning?

Answer: Roughly aiming for ~50 optimisation events per ad set per week.

#17What should you check before the next lesson?

• One diagnostic completed on a live ad set.
• Change history reviewed for surprise edits.
• Ready to operationalise the monthly routine.

Previous: Meta Ads and AI search discovery overlap. Next: Monthly Meta Ads review routine. Or return to the Meta Ads for Marketers track. Related: Objectives, budgets and bidding, Events Manager and EMQ, GTM debugging.

#18Common questions

What does learning limited mean? The ad set is not getting enough optimisation events roughly within a week to exit learning — often from low budget, rare events, too many ad sets or constant edits.

Why is my ad set not spending? Common causes include too-tight cost goals, tiny audiences from hard limits, payment/account quality issues, disapproved ads, schedule conflicts or bid caps.

Should I duplicate campaigns to fix learning? Duplicating can reset learning and fragment volume. Prefer consolidating ad sets, fixing events and refreshing creative over endless duplicates.