My Facebook Ads Got Worse Overnight: Is It Fatigue, Budget or Tracking?

Yesterday your cost per purchase sat where it has been all month. This morning it has jumped, and nobody touched the creative or the targeting. The obvious fixes are a new ad or a budget cut. Both reset things you may need later, and both are expensive if you have the wrong cause.

An overnight drop is rarely ad fatigue, because fatigue builds over days as the same people keep seeing the same ad. A sudden break with steady clicks and collapsing conversions usually means tracking has failed. A break the day after an edit to spend points to budget. Whichever metric moved first tells you which one you have.

Why fatigue rarely happens overnight

Creative fatigue follows an exhaustion curve. An ad reaches the people most likely to respond to it first. As delivery continues, more and more impressions go to people who have already seen it and already decided. Frequency climbs, click-through rate erodes, and cost per result drifts upwards. Each day is a little worse than the one before.

That shape is what you diagnose from. Fatigue is a slope, not a cliff. Pull the daily chart for the last three or four weeks. If it shows a steady decline that only crossed your tolerance yesterday, you are looking at fatigue that was noticed overnight, not fatigue that happened overnight. That is common, and it is worth ruling out first, because the fix is new creative rather than a tracking audit. How fast an ad reaches that point depends on spend and audience size, which how long a Facebook ad lasts before it needs replacing covers in detail.

The signature to look for is rising frequency alongside falling CTR. If reach is also shrinking while frequency rises, the ad set has used up its audience rather than its creative. Small retargeting pools are the one place fatigue can feel sudden, because a narrow audience saturates in days rather than weeks. Outside that case, a genuine one-day step change almost always has a cause that also happened on one day. If what you see is a gradual slide, start with the wider diagnosis in why Facebook ads stop working.

What does a tracking break look like?

A broken pixel is the likeliest cause of a true overnight cliff, and the most dangerous, because nothing announces it. There is no error message and no paused campaign. Events stop arriving, or arrive malformed, and the budget keeps spending.

The pattern is distinctive. Spend, impressions, CPM, clicks and CTR all look normal. Conversions fall sharply from one day to the next, sometimes to zero. Everything upstream of your website is fine. Only the part that depends on the site reporting back to Meta has broken.

The cause is usually a change on your side that nobody connected to ads:

• a theme update
• a new checkout flow
• a CMS upgrade
• a developer tidying up scripts
• a reconfigured consent banner

Events can also break while still appearing to fire. A purchase event might lose its value parameter, or a standard event might get renamed during a rebuild.

The longer it runs, the worse it gets. Meta's delivery optimises on the conversion signal it receives. Lose that signal and the system loses its feedback on who converts. A reporting fault can then turn into a real performance decline over the following days, even though budget and creative never changed.

The check takes minutes:

1. Compare the conversions Ads Manager reports for the past week against your backend (Shopify orders, CRM records or payment processor) for the same days.
2. If real sales held steady while reported conversions fell, the problem is measurement.
3. If both fell together, tracking is fine and the drop is real. Look at the site, the offer, stock and checkout before you look at the ads.
4. Open Events Manager and confirm your pixel and Conversions API events are still arriving with the parameters you expect.

When is budget the real cause?

Budget causes an overnight drop when someone changed it: you, an automated rule, or a colleague. That makes it the easiest of the three to confirm. Open the account's activity history and look for budget edits in the two days before results broke.

Two moves cause most of the damage:

• A large increase. It forces the ad set to find more conversions at once, which means buying impressions further from the people who were already converting, at a higher marginal cost.
• A big enough edit. This can send the ad set back into the learning phase, where delivery is less stable and results swing for several days.

Scaling too quickly is a cause in its own right, separate from anything wrong with the ad.

There is also a cost change you did not make. If CPM jumps sharply while CTR holds, the auction got more expensive. A competitor's sale, a seasonal peak or more advertisers chasing the same audience all have this effect. Your budget now buys fewer impressions, and results fall even though nothing in your account changed. That is a market condition to price in, not a fault to fix. Raising budget to compensate rarely helps. If the offer or price is the real constraint, more spend just buys more of the same result at a higher cost.

How do you tell the three apart?

Match what you see to the pattern, then run the cheapest check first.

| What you see | Likely cause | First check | |---|---|---| | Frequency up, CTR down over days | Creative fatigue | Daily chart, last 3–4 weeks | | Reach shrinking, frequency rising | Audience saturation | Audience size and overlap | | Clicks steady, conversions fall in a day | Tracking break | Backend sales vs Ads Manager | | Results drop the day after a spend edit | Budget or learning reset | Activity history | | CPM jumps, CTR stable | Auction pressure | CPM against prior period | | Ads Manager and backend both fall | Site, offer or checkout | Site conversion rate |

Start with the backend comparison, even if you suspect something else. It is quick, and a tracking fault contaminates every other number on the screen. Frequency and CTR stay accurate, but cost per result and ROAS do not. Diagnosing fatigue from a broken ROAS figure is how teams end up replacing creative that was working.

Then hold still. The worst response to an overnight drop is three simultaneous changes: new creative, a budget cut and a fresh audience. If performance recovers, you will not know which change did it, and each one can reset learning on its own. Make one change for one cause, then give it a day or two before you read the result. Meta results fluctuate day to day. A single bad day with no step change in any one metric is often noise. A decline that persists for a couple of weeks is the one to investigate properly.

Conclusion

An overnight drop is a timing clue. Fatigue is gradual. Tracking faults and budget edits happen on a specific day, and so do the drops they cause. Compare reported conversions against real sales first, check the activity log second, and only then decide whether the creative has run its course. A wrong diagnosis usually costs more than the original problem, because the fix it prompts resets something that was working.

See how Companion's daily brief flags what changed overnight, grounded in your live spend, performance and order data rather than a weekly report.