Why Your Meta Ads Stopped Working (And How to Fix It)

The most common cause of a sudden, dramatic performance drop is not a performance drop at all. A site update, a new consent banner, a theme change, an expired integration or a broken Conversions API connection can all stop conversion events being recorded correctly, and the account then appears to collapse overnight while sales continue normally.

The tell is a mismatch: platform-reported conversions fall sharply while your own order or enquiry numbers hold steady. That pattern means a measurement problem, not a marketing one — but it becomes a marketing problem quickly, because the optimiser is now learning from incomplete data.

Check event firing, deduplication and whether purchase values are still being passed before touching a single campaign setting. Rebuilding an account to fix a tracking bug is a genuinely expensive mistake.

If the decline is gradual rather than sudden — cost per result creeping upward over weeks, click-through rate falling, frequency rising — the cause is almost always creative fatigue. The audience has seen the ad, formed an opinion, and is no longer responding.

Every ad has a finite audience before this happens, and higher spend exhausts it faster. This is why accounts that scale need new creative entering weekly rather than quarterly: the question is not whether an ad will fatigue but how quickly.

The fix is a new angle, not a new edit. Recutting a fatigued ad with a different opening can buy a little time, but an audience that rejected the underlying message will reject it again in a different font. Produce genuinely different reasons to care: a demo if you have been running testimonials, an objection piece if you have been running demos.

Significant edits — large budget changes, new audiences, changed optimisation events, swapped creative on a live ad set — push campaigns back into learning, and performance is unstable while they re-stabilise. If the drop coincides with a change somebody made, this is usually the explanation.

The instinctive response, changing something else to fix it, restarts the clock again. Accounts get stuck in permanent learning this way, never stabilising because nobody leaves them alone long enough.

The remedy is patience and a change log. Note every edit with a date, so that when performance moves you can look at what actually happened rather than reconstructing it from memory a fortnight later.

If costs rose specifically after a budget increase, the account may simply have run out of the cheapest, most interested people. Scaling reaches progressively less-interested audiences, and cost per result rising with spend is normal rather than a malfunction.

The judgement is whether it is still under what a customer is worth. If yes, the higher cost is the price of volume and is fine. If not, the options are to scale back to the efficient level, improve conversion rate downstream, or find new creative angles that open up a different segment of the audience.

Frequency is the useful diagnostic here: a climbing frequency alongside rising costs means the same people are being shown ads repeatedly, which points at either saturation or too narrow an audience.

Auction prices move with advertiser demand. The fourth-quarter shopping season pushes prices up substantially as retail budgets arrive, and prices typically ease afterwards. A campaign that suddenly costs more in November has not necessarily got worse — the auction got more expensive.

Other predictable swings include major sales events, election periods in some markets, and category-specific seasonality. Comparing this year's period against the same period last year, rather than against last month, avoids misdiagnosing a seasonal pattern as a failure.

The response is planning rather than panic: budget for higher costs during peaks, build creative in advance, and avoid launching brand-new untested angles into the most expensive weeks of the year.

If click-through rate and cost per click are unchanged but cost per result has risen, the ads are doing their job and something downstream changed. A slower page, a new checkout step, an out-of-stock product, a price increase, a broken form, or a redesign can each move conversion rate enough to swing the whole account.

This is a frequently missed cause because everybody looks inside the ad platform first. Checking whether the landing page conversion rate changed at the same time usually resolves the question in minutes.

It is also the cheapest fix. Recovering conversion rate on the page is often easier and faster than trying to compensate for it with better ads.

Accounts drift towards complexity over time: a new campaign for each test, each product, each idea, none of them ever switched off. The result is a budget spread so thinly that no campaign gathers enough conversions to optimise, and the whole account's performance degrades without any single thing being broken.

Consolidation is unglamorous and frequently the highest-impact change available. Fewer, better-funded campaigns produce faster learning and steadier costs, even though the dashboard looks less impressive.

As a rule of thumb, if a campaign is not producing a meaningful number of conversions a week, it is not producing information either, and it either needs more budget or should not exist.

Work through the causes in order of cost to check rather than likelihood. Tracking first, because it is quick and the consequences of missing it are severe. Then the change log, to see whether the drop follows an edit. Then the creative metrics — click-through rate and frequency — to identify fatigue. Then downstream conversion rate. Then seasonality and account structure.

Change one thing at a time and give each change long enough to produce a readable result. The temptation during a bad month is to change everything at once, which guarantees you will not know what worked when it recovers.