Meta Ads Creative Fatigue Analysis: From Exported Report to Next-Test Plan

Upload comparable Meta Ads exports. Runbear ranks the biggest delivery and response changes, flags possible creative fatigue, drafts three creative directions, and returns an HTML report for review. It works from exports only and never changes the ad account.

What the agent does

• Performance: Rank the campaigns, ad sets, ads, and placements with the clearest changes.
• Fatigue review: Read repeated delivery and weaker response together, with competing explanations.
• Creative: Draft distinct hooks, copy, and visual directions tied to the observed pattern.
• Plan: Define controls, variables, measures, missing inputs, owners, and approvals.

When the scheduled review finishes, Runbear posts a short summary and the HTML report.

Open the report for the comparison, caveats, and next test. Account changes still require approval.

Prepare comparable inputs

Use two complete periods from one account. Keep timezone, currency, attribution, reporting level, objective, result definition, and columns consistent.

• Upload ad-level and placement reports with IDs, delivery, spend, reach, frequency, clicks, CTR, CPM, results, and cost per result.
• Add the brand brief, approved claims, formats, current copy, hook, and preview image.

Review a fatigue signal

No single threshold proves fatigue. The Tidewell figures below are fictional.

Evidence: Video 03 frequency rose while outbound CTR fell and cost per purchase increased. CPM barely changed; Static 02 remained stable.

Inference: This may be creative fatigue, but the export does not prove the cause.

Unknown: Audience edits, learning status, offer, landing page, measurement health, and creative changes.

The agent proposes three tests while keeping the product and offer fixed:

• Problem-first video: Open on a crowded bag, then show the locking lid.
• Three-moment video: Cut from desk to train to trail.
• Feed card: Show approved features in a scannable checklist.

All three remain drafts until the team approves the claims, assets, budget, and test design.

Run it every week

Schedule the review after refreshing the exports. Runbear analyzes the uploaded files and returns a short message with an HTML report. If the periods or columns do not match, it stops and tells you what to replace.

Setting it up

1. Create a Runbear agent named Meta Ads Report Analyst.
2. Export and upload matching ad-level reports for the two completed periods, plus the brand brief and creative references.
3. Run one review, then schedule it after the export refresh and use the instruction below.

meta-ads-report-analyst-instruction.txt
You are the Meta Ads Report Analyst.

Use only the supplied exports, brand brief, and creative references. You cannot access or change Meta Ads.

For each run:
1. Confirm the account, periods, timezone, currency, attribution, reporting level, objective, and result definition. Stop if the files are stale, incomplete, or not comparable.
2. Rank the largest delivery and response changes across campaigns, ad sets, ads, and placements.
3. Assess possible fatigue from several signals. Do not use a universal threshold. Separate Evidence, Inference, and Unknown.
4. Draft up to three distinct creative directions tied to the evidence and approved claims.
5. Return a short message and attach meta-creative-review.html with the comparison, fatigue signals, competing explanations, unknowns, controls, measures, owner, approvals, and next test.
6. Mark every creative or account action Awaiting approval.

Never create, edit, publish, pause, or apply changes to campaigns, ad sets, ads, audiences, placements, bids, budgets, tracking, or experiments.

Boundaries

• The agent only uses supplied files. It cannot refresh reports, prove fatigue, invent claims, or change Meta Ads.