Last Updated on May 29, 2026 by Hafsa J.

46,000 Complaints, Few Recalls: Inside the FDA’s New Real-Time Cosmetics Dashboard

Since December 2023, almost forty-six thousand adverse-event reports have been filed with the FDA about cosmetic products. In that same window, the number of products actually recalled is a tiny fraction of that. So here is the question worth sitting with: if tens of thousands of people are reporting reactions, why do so few of those reports turn into recalls, and what does it mean for you that every one of those reports is now public, searchable, and updated daily?

That last part is new. As of September 2025, anyone with a browser can look up a cosmetic product and see the reactions people have reported against it. For most of the industry’s history that data was buried, available only through slow records requests. Now it is a dashboard. I want to walk through what that database actually is, what its big number really means, and the part most manufacturers miss entirely: that the same tool a worried consumer uses to check a product is also the earliest warning system you will ever get about your own.

What the dashboard is, and the one thing it is not

The tool is the FDA’s Adverse Event Monitoring System (AEMS) Public Dashboard for cosmetic products. If the acronyms feel like a moving target, that is because they are: it launched in September 2025 as the FAERS dashboard, drawing on the older CAERS database, and in March 2026 the FDA folded it into a unified system it now calls AEMS. Same data, evolving name. What matters is what it does: it lets anyone search cosmetic adverse-event reports by product name, by reaction, or by date, and it refreshes daily with the newest submissions.

It pulls from two streams. The first is mandatory: under MoCRA, the responsible person, the company whose name is on the label, must report serious adverse events to the FDA within fifteen business days. The second is voluntary: reports submitted directly by consumers, healthcare professionals, salon workers, and anyone else who experienced or witnessed a problem. The dashboard pools both into one searchable public record.

Now the single most important thing to understand about it, and the line I want you to hold onto through everything that follows. A report in this database is not proof of anything. The FDA states plainly that the reports are unverified and that their presence does not mean the agency has concluded the product caused the reaction. Someone can report a rash after using a moisturiser; that report enters the database; and it may have nothing to do with the moisturiser at all. The data is exactly what was reported, no more. Anyone reading it, a journalist, a competitor, a regulator, or you, has to treat it as a signal to investigate, never as a verdict.

That distinction is not a technicality. It is the whole reason the question I opened with has the answer it does. Tens of thousands of reports, few recalls, because a report and a confirmed, recall-worthy hazard are two very different things. The next section is about that gap, and why the big number is less alarming, and more useful, than it first looks.

The number nobody expected: ~46,000 reports

When a number like forty-six thousand lands, the instinct is to read it as a sign that cosmetics suddenly got more dangerous. That reading is wrong, and getting it right is the difference between panicking and using the data well.

Cosmetics did not become more dangerous in December 2023. What changed in December 2023 is that reporting became mandatory. Before MoCRA, a company that received a complaint about a serious reaction had no federal obligation to tell the FDA. Many did not. The reactions were happening; they simply were not being collected in one place. MoCRA’s reporting requirement, combined with the FDA’s decision to publish the results, did not create forty-six thousand new problems. It surfaced a volume of reporting that had always existed and had never been visible. The number is large because the net is finally wide, not because the water is suddenly full of sharks.

This sits inside a deliberate shift the FDA describes, in its own words, as a move toward radical transparency. The agency is no longer content to collect safety data quietly and act on it internally. It is putting the raw stream in public view, daily. Based on my experience watching how regulators usually handle this kind of data, that is a meaningful change in posture. The old model was “trust us, we are watching.” The new model is “here is what we are watching, watch alongside us.” For a manufacturer, that shift cuts both ways, and the rest of this article is about which way it cuts for you.

Now back to the original question: why so few recalls against so many reports? Because the journey from a report to a recall is long and evidentiary. A report is one person’s account. A recall, especially under the FDA’s mandatory recall authority introduced by MoCRA, requires the agency to establish a reasonable probability that a product is adulterated or misbranded and will cause serious harm. Most reports never clear that bar, and many never should, because they do not reflect a genuine product defect. The gap between the report count and the recall count is not evidence of a cover-up. It is evidence that a single report, on its own, proves very little.

But here is the trap inside that reassurance. The fact that most reports do not become recalls does not mean reports are noise to be ignored. Buried in forty-six thousand reports are the early, scattered signals of the few that will become recalls. The skill is not dismissing the database because the bar for a recall is high. It is learning to spot the pattern that is climbing toward that bar before it gets there. That is a reading almost nobody outside the regulator is doing, and it is exactly where a manufacturer’s advantage lies.

Three ways to read the same data

The dashboard is one database, but it is read by three very different people, and the gap between how they read it is where the opportunity sits.

The consumer reads it as a buying signal. They type in a product they are considering, see a list of reported reactions, and make a decision. This is the use the FDA’s transparency push is built around, and it is also the most prone to misreading, because a worried shopper rarely stops to ask whether a report was ever verified. A handful of unconnected reports can sink a purchase that the data never actually justified sinking. That is the cost of radical transparency, and it is not going away.

The regulator reads it for clusters. A single report is noise. The same product, the same ingredient, or the same reaction showing up repeatedly is a signal. The FDA uses these patterns to decide where to point finite enforcement attention, which facility to inspect, which product to scrutinise. The dashboard is, in part, a triage tool for the agency’s own resources, and a rising cluster against your name is a quiet way of moving up that list.

The manufacturer, in my experience, reads it least of the three, and has the most to gain from reading it most. Because the third reading is the one that turns this database from a threat into an instrument: early warning. The complaints accumulating against your product, or against a competitor’s near-identical formula, are visible to you on the same daily feed the regulator sees. You can watch your own category in real time. You can see a pattern forming against an ingredient you also use. You can catch the shape of a problem while it is still a scatter of reports and not yet a recall. The consumer reads to avoid; the regulator reads to act; the manufacturer can read to prevent. Almost none of them do.

That gap, between the manufacturers who treat this as a public-relations hazard to be feared and the ones who treat it as a surveillance feed to be worked, is the entire practical argument of this article. The next section is about how you actually do the second thing.

Turning a public database into your own early-warning system

Reading the dashboard as a manufacturer is not complicated, but it is deliberate. It is a habit, not a one-time check, and here is what that habit looks like in practice.

Start with your own products. Search them by name on a fixed schedule, monthly at least, and read every new report against them. Not to dismiss the ones that seem unfair, but to ask the uncomfortable question each time: is there anything here that, if it recurred, would point to something real in my manufacturing or my formula? One report of irritation is noise. Three reports of the same reaction to the same product line in the same quarter is a thread to pull, well before anyone outside your company starts pulling it.

Then widen to your category. Search the product types and the key ingredients you use, across the whole database, not just your own label. If a preservative or a fragrance compound you rely on is generating a rising pattern of reactions industry-wide, that is intelligence you want months before it becomes a reformulation scramble or a regulatory focus. You are using other companies’ reports as a free early test of the ingredients sitting in your own formulas.

This is cosmetovigilance, and under MoCRA it is not entirely optional. The responsible person already carries a legal duty to report serious adverse events within fifteen business days. A company that is not systematically watching the very database it is required to feed is in a strange position: contributing to the signal while ignoring it. The manufacturers who handle this well fold dashboard monitoring into their existing quality system, alongside the complaint-handling and post-market surveillance that ISO 22716 already expects them to run.

If you want a quick read on whether your quality system is actually set up to do this, rather than just intending to, the self-assessment below covers complaint handling and post-market surveillance among the GMP areas it scores.

So, back to the question this article started with. Tens of thousands of reports, few recalls, all of it now public and refreshed daily. The reason so few become recalls is that a report is not proof. But the reason you should care anyway is that the few which do become recalls were visible as scattered reports first, on the same feed you can read for free. A post-release contamination like the Suntegrity recall often shows its first traces here, in complaints, before the company runs the test that confirms it. And the same data feeds the FDA’s decision about when to use its recall power. The database is open to everyone. The only real question is whether you are reading it before the regulator does, or finding out what it said after they have already acted.