Advertisers report Meta ad sets stop spending without errors

• Active Meta ad sets were reported dropping to zero spend with no visible error, according to binghott's first report and binghott's follow-up about random ad-set drops.
• One likely trigger was overfilling an ad set: binghott's 51st-ad reply identified the added unit as the 51st ad, while Meta's Marketing API reference caps an ad set at 50 non-archived ads.
• A separate settings bug report claimed a disabled feature came back on after publishing and refreshing; binghott's video-proof claim said there were no edits left to publish.
• The public status trail lagged user reports: binghott's confirmation request asked Meta to confirm the issue, and his status-disclosure reply said even resolved incidents need disclosure.

The cleanest technical breadcrumb is the number 50: Meta's own Ad Set API docs list “Maximum number of ads per ad set” as 50 non-archived ads, and binghott's 51st-ad reply put one failure at exactly the next ad. Meta's Advantage+ creative page says some enhancements may be on by default and can be turned off at any time, while its turn-off guide tells advertisers to check default-on enhancements before publishing. The advertiser-side complaint was stranger: binghott's video-proof claim alleged a setting changed after publish, with nothing left in draft.

The first report was narrow: ad sets that had been spending “perfectly fine” two days earlier had fully stopped spending yesterday and today, according to binghott's first post. A later reply added that overall spend was still present, but a few previously spending ad sets dropped to zero “suddenly and seemingly randomly,” per binghott's follow-up.

Meta's delivery-status help page says the Delivery column is where advertisers check campaign, ad set, or ad status and find information about problems. The reported failure mode was more maddening: “Active” status, zero spend, no error, no flag.

The workaround report reads like a bug repro:

• More ads were added to a top-spending ad set.
• The ad set stopped spending.
• Ads Manager showed no errors or flags.
• Turning the ad set off and on did nothing.
• Increasing cost caps did nothing.
• Duplicating the ad set once did nothing.
• Duplicating it again and turning off some ads worked.
• Turning off some ads in the original ad set made the original spend again.

The missing constraint appeared later: binghott's 51st-ad reply said the new ad was the 51st ad. Meta's Marketing API docs list a hard cap of 50 non-archived ads per ad set, so the weird part was not the limit. It was that Ads Manager apparently allowed the state, then left the ad set active with no useful warning.

Binghott's reply to another user's reaction was shorter: Yeah, ridiculous.

A second report moved from delivery to control. Binghott's video-proof claim said he turned off a specific ad feature, published successfully, refreshed, and saw the feature back on with no edits made and nothing left to publish.

He later said the video helped prove an employee had made the intended edit and that Meta effectively overrode it, according to binghott's follow-up about the video. Another reply said a Meta rep had “block-listed” accounts from some ad display features, but binghott's account-level reply said he still did not trust that those restrictions were being applied.

The reporting gap is part of the complaint. Binghott's reply on proof said it would be almost impossible to prove how much money had run through those settings because Meta does not report on them.

The same thread did not turn into a blanket anti-automation argument. Binghott's take on Meta optimizations said Meta may be improving performance for many advertisers because most advertisers are too self-limiting, even though he called many of the AI creative changes he had seen “terrible.”

On performance diagnosis, he separated platform bugs from noisy metrics. Binghott's CVR and CPM reply called CVR and CPM components, with CPA and spend as the metrics he cared about, while his CPM reply said higher CPMs can reflect warmer audiences or more expensive placements. In another exchange, binghott's website-issue reply argued that higher CPMs plus lower CVR usually points away from Meta and toward the site.

His position on manual intervention was also qualified. Binghott's nuance reply agreed with an “almost never” framing on intervening against Meta, but said there are times Meta gets it very wrong; his intervention reply added that some interventions are less severe than turning ads off.

Meta's public Ads Manager status page listed “No known issues” for Ads Creation and Editing, Ads Delivery, and Ads Reporting during the research pass for this story. Its campaign-not-running help page covers normal checks such as delivery status, spending limits, lifetime budget, and budget scheduling when no budget has been spent.

Advertisers were asking for something more specific than a help-center checklist. Binghott's confirmation request asked Meta to confirm the issue publicly, and his status-disclosure reply argued that even a resolved issue should still be disclosed after the fact.