A Meta Ads campaign drops overnight. Cost per lead doubles. Nothing in the campaign changed. What do you do?

This post walks through a real case study of exactly that scenario — a $20,000/month lead generation campaign that fell from a $50 cost per lead to $114 in the space of 24 hours — and the three-stage framework used to diagnose, attempt to fix, and ultimately resolve it. The resolution involves an unconventional approach that has no logical explanation but worked when everything else failed.

The framework applies to any unexpected Meta Ads performance drop, not just edge cases.

The Case: A Campaign That Had Never Failed

The context matters here. This was not a new or unstable campaign. It had been running for a professional services business with a monthly budget of over $20,000, consistently improving its cost per lead over several months. January and February had settled at a campaign best of around $50 CPL, with not a single day above $100 in either month.

On March 1st, with no changes made to the campaign, ad set, or ads, cost per lead jumped to $114 for the first five days of March — with four of those five days above $100.

This is the kind of drop that immediately raises a question: is this explainable, or is something genuinely wrong?

Stage 1: Check for Logical Explanations

The first and most important discipline when facing an unexpected performance drop is to exhaust logical explanations before assuming something unusual has occurred. The majority of performance drops have explainable causes, and each one has a structured response.

CPM (cost per thousand impressions)

CPM had increased approximately 42% above its normal range, reaching $57 when it had previously peaked at $43 in December. An elevated CPM increases the cost of reaching the audience, which flows through to a higher cost per lead.

However, a 42% CPM increase alone cannot explain a more than doubling of cost per lead. CPM was a contributing factor, not the full explanation.

Conversion rate

The landing page conversion rate had more than halved — from approximately 7% to just over 3%. This, combined with the CPM increase, explained the dramatic cost per lead spike. The ad was costing more to deliver and converting at a lower rate from the audience it was reaching.

Ad content and engagement

The primary ad had accumulated over 6,000 comments. A review showed no significant negative sentiment that would explain a conversion rate drop. But during that review, something notable was observed: new comments had stopped completely on March 1st. An ad receiving a dozen or more comments per day was now receiving none at all.

Platform distribution breakdown

This is where the real story emerged. Breaking down spend by platform revealed a complete structural shift:

• January–February: Facebook = 92% of spend. In-app Facebook Feed = 64% of placements.
• March 1–5: Instagram = 86% of spend. In-app Facebook Feed = 0% of placements.

The placement that had been driving 64% of all impressions — and the vast majority of conversions — had received zero impressions since March 1st. This was not a variance. This was a complete structural redistribution with no changes made to trigger it.

Stage 2: Apply Logical Interventions

Value rules to redirect placement distribution

With the diagnosis pointing to a placement distribution issue — too much spend on Instagram, zero on the in-app Facebook Feed — the logical response was a value rule to reduce the bid on Instagram placements. This would make Instagram relatively less cost-efficient for the algorithm and encourage budget to flow back to Facebook.

A value rule was applied to decrease bids on Instagram Feed, Stories, and Reels by 50%.

What happened

The value rule partially worked: Instagram spend dropped and Facebook spend increased. But the budget that moved to Facebook went almost entirely to Facebook Marketplace — a placement that had received minimal spend historically. The in-app Facebook Feed remained at zero impressions even with Instagram de-prioritised.

This was the confirmation that something was structurally wrong beyond normal distribution behaviour. The algorithm was refusing to deliver impressions to its previously dominant placement, and a direct value rule intervention could not override it.

Stage 3: The Last-Resort Unconventional Fix

When logical interventions fail to produce expected results, and when the behaviour pattern is inconsistent with how the platform normally operates, the situation may reflect a platform bug rather than a performance issue. In these cases, there are unconventional approaches that occasionally work — not because they have a logical mechanism, but because they appear to reset whatever corrupted state is causing the problem.

These should only be attempted after logical explanations have been exhausted and logical interventions have failed. They should never be the first response.

What was tried

The primary ad — the one that had previously driven the majority of impressions and 6,000+ comments — was duplicated. Minor changes were made to the creative. The new version was published.

The result

The impact was immediate. Within a short time of the new ad being approved, impressions began flowing back to the in-app Facebook Feed. Marketplace was no longer dominant. After monitoring for a couple of days, the value rule was removed.

March 9–13 results: Facebook 87% of spend, in-app Facebook Feed 66% of placements, cost per lead $46 — lower than the January and February average.

Why This Worked: A Best Guess

There is no clean logical explanation for why duplicating and slightly modifying the ad restored normal delivery. The most likely explanation is that the original ad had entered a corrupted distribution state — something in the algorithm’s optimisation data for that specific ad had become misaligned in a way that was preventing it from accessing its previously dominant placement. A fresh version of the ad started the formation process from a clean state, allowing the algorithm to rediscover the Facebook Feed placement as an optimal delivery channel.

This is a hypothesis, not a confirmed mechanism. Meta has not explained why this occurs. But the pattern — a top-performing ad suddenly losing access to its primary placement, restored by a fresh creative variant — has been observed across accounts, suggesting it is not a one-off anomaly.

The Full Diagnostic Checklist for Meta Ads Performance Drops

Use this checklist in order when facing an unexpected performance drop:

Initial diagnostics

• Check CPM — is it elevated above the recent baseline? By how much?
• Check conversion rate — has it dropped? For which stage of the funnel?
• Check creative engagement — are comments, likes, and shares continuing at normal rates?
• Check platform and placement breakdown — has spend distribution shifted between Facebook and Instagram, or between specific placements?
• Check for recent changes — any edits to the campaign, ad set, ads, landing page, or website in the relevant timeframe?
• Check for external factors — competitor activity, seasonal shifts, relevant news events

If the explanation is found

• CPM spike with normal conversion rate → accept as temporary cost increase, monitor
• Conversion rate drop with normal CPM → review landing page, offer, and ad creative
• Placement distribution shift → apply value rules to adjust bid levels for underperforming placements
• Creative engagement stopped → refresh creative, test new angles

If no logical explanation is found and interventions have failed

• Duplicate the primary winning ad with minor creative changes and publish fresh
• If that doesn’t resolve it, consider duplicating the ad set entirely
• Monitor placement distribution data closely after any change

Key Takeaways

• Most Meta Ads performance drops have logical explanations — always start there
• Platform and placement breakdowns are essential diagnostic tools — surface-level metrics won’t show you where distribution has shifted
• Value rules are the preferred first intervention for placement distribution problems — they adjust bids without removing placements entirely
• When logical interventions fail and the behaviour looks like a bug, duplicating the top-performing ad with minor changes is a last-resort option worth trying
• Months of stable performance data is essential context — without a clear baseline, unusual behaviour is difficult to identify

FAQs

1.My Meta Ads cost per lead suddenly doubled. What should I check first?

Start with CPM (cost per thousand impressions) and conversion rate. If CPM has spiked significantly, check whether it’s a temporary competitive fluctuation or a sustained change. If conversion rate has dropped, investigate your landing page performance and whether the audience reaching your page has shifted. Then check your platform and placement breakdown — a shift in where your ads are being delivered can explain both CPM changes and conversion rate changes simultaneously.

2.My Meta ad used to perform well but suddenly stopped getting impressions. What’s happening?

This can happen for several reasons: creative fatigue (the audience has seen the ad too many times), a shift in the algorithm’s distribution logic, or — in rarer cases — something closer to a corrupted delivery state. Start by checking placement breakdowns to see where impressions have gone (or haven’t). If the ad is receiving zero impressions on placements that previously dominated, try duplicating the ad with minor changes and publishing fresh.

3.What are value rules and how do they help with placement distribution issues?

Value rules allow you to adjust your bid for specific audience or placement criteria without removing them entirely. If Meta is sending too much budget to a specific placement — like Instagram — you can apply a value rule to decrease the bid on that placement by a percentage. This makes it relatively less cost-efficient for the algorithm and encourages budget to redistribute toward other placements. Value rules are the preferred approach over hard placement restrictions because they work with the algorithm rather than imposing hard constraints.

4.Why would duplicating an ad fix a delivery problem?

There is no confirmed mechanism. The working hypothesis is that a top-performing ad can enter a corrupted distribution state — the algorithm’s optimisation data for that specific ad becomes misaligned in a way that prevents it from accessing certain placements. A fresh duplicate starts the formation process from a clean state, allowing normal delivery patterns to re-establish. It doesn’t always work, but it’s worth trying when logical interventions have failed.