𝘛𝘩𝘦 𝘳𝘦𝘷𝘪𝘦𝘸𝘴 𝘢𝘳𝘦 𝘤𝘰𝘮𝘱𝘳𝘰𝘮𝘪𝘴𝘦𝘥. 𝘛𝘩𝘦 𝘳𝘢𝘵𝘪𝘯𝘨𝘴 𝘢𝘳𝘦 𝘮𝘪𝘴𝘭𝘦𝘢𝘥𝘪𝘯𝘨. 𝘛𝘩𝘦 𝘢𝘳𝘤𝘩𝘪𝘷𝘦 𝘪𝘴 𝘶𝘯𝘳𝘦𝘭𝘪𝘢𝘣𝘭𝘦. 𝘈𝘯𝘥 𝘵𝘩𝘦 𝘤𝘰𝘮𝘮𝘶𝘯𝘪𝘵𝘺 𝘬𝘦𝘦𝘱𝘴 𝘨𝘰𝘪𝘯𝘨 𝘣𝘢𝘤𝘬 𝘢𝘯𝘺𝘸𝘢𝘺.
Somewhere on Fragrantica right now someone is publishing a five-star review of a fragrance that has not yet been released. You cannot verify whether they have smelled it. Apparently neither can Fragrantica.
And yet Fragrantica has more influence over what fragrances you buy than any brand, any influencer, or any critic, but with far less accountability than any of them.
Over the years I have asked many serious frag heads privately what they think of Fragrantica reviews. They usually complain that the site is biased or flawed, but are not able to express why. What they are trying to say is what this article is about.
Before publishing this piece I reached out to Fragrantica for comment, including a draft of this article and my main criticisms. Zoran Knezevic, co-founder of the platform alongside his wife Elena, replied swiftly and personally. His thoughts wherever pertinent are included.
I know, I know … Fragrantica is the first place many serious fragrance lovers go to examine perfume. It’s also the last place most of them fully trusts. That contradiction has never been examined honestly. It should be.
Here is my take: Fragrantica is broken. Not irreparably, not maliciously, and not in ways that make it useless, but broken in specific, documented, consequential ways that the community has sensed for years without anyone saying clearly, in detail.
Here is what I believe is actually wrong.
𝐖𝐡𝐚𝐭 𝐅𝐫𝐚𝐠𝐫𝐚𝐧𝐭𝐢𝐜𝐚 𝐛𝐮𝐢𝐥𝐭
Before sites like Fragrantica existed, serious fragrance knowledge was gatekept by geography, by budget and by access to specialist retail that simply does not exist in most of the world. The platform democratised fragrance discovery in a way that deserved recognition then and still deserves it now. Its archive is vast, its database genuinely useful and the treasure hunting quality — scrolling through obscure releases, finding unexpected community around a forgotten fragrance — is an undeniable pleasure.
I use Fragrantica. I have also made purchasing decisions based on its reviews. Some of the advice was spot on, and some of it left me gnashing my teeth in regret after buying something. The most instructive was my negative experience with Nishane Hacivat — overwhelmingly reviewed as a pineapple forward take on Creed Aventus, smoky in an interesting way, a niche house doing something genuinely original based on a modern classic. Convinced what I was reading matched my limited in person impressions with the fragrance, I spent approximately $150 CAD to purchase a 50 ml bottle.
What I discovered in Hacivat was a fragrance in which the pineapple note appeared almost exclusively when applied on fabric while applying it on skin revealed something else entirely: a smoky, mossy, patchouli-dominant composition that bore almost no relationship to what the reviews had promised.
One reviewer among dozens had noted Hacivat’s split personality, but his insights did not get the weight they deserved. His observation had been buried under the weight of the consensus which the platform’s architecture had made it easy to miss. I do not know who he is, but he was right in his criticism. I eventually sold the bottle to a friend who appreciated it more than I did.
When I checked Fragrantica’s AI-compiled summary of Hacivat while writing this article, it still described Hacivat as having a “juicy, high-quality natural pineapple note,” but there was nothing mentioned about skin versus fabric behaviour.
Fragrantica itself notes that its AI summary may be inaccurate and that full reviews should be read instead. But this raises an obvious question: if the summary may be inaccurate, then why is it the first thing a buyer sees?
The problem is not that Fragrantica exists. The problem is that it has become the primary research tool for millions of fragrance buying decisions while remaining structurally unequipped to support the weight of that responsibility. Here is what that looks like in practice.
𝐖𝐡𝐚𝐭 𝐅𝐫𝐚𝐠𝐫𝐚𝐧𝐭𝐢𝐜𝐚 𝐛𝐫𝐨𝐤𝐞
The first problem is one the community has sensed without naming. I call it the conformity machine.
When a highly anticipated fragrance releases, early reviews arrive before most buyers have smelled anything. Those reviews establish a tone, a dominant narrative about what the fragrance is, what it resembles, what it promises. Subsequent reviewers absorb it. The result looks like community consensus. In fact, it is frequently social pressure wearing consensus as a disguise.
Fragrance community member Stallionmang said it plainly in his review of the newly released Lattafa Khamrah Waha: “Can you guys all relax and talk about the fragrance if you’ve actually tried it?” I agree, he perfectly summed up the entire problem.
The conformity machine creates perfect conditions for the second problem, one that is corroding the platform’s usefulness faster than any other.
As a professional writer I can usually identify AI generated prose within 10 seconds of reading. AI written content tends toward a narrated third person voice with no identifiable human behind it. It avoids strong opinions and circles the same observation repeatedly without advancing anywhere.
Consider this staccato construction, rife on Fragrantica: “The fragrance doesn’t stop. It keeps going. It projects. And it has strong performance.” Those four sentences read less like a human and more like word salad, in my opinion.
Knezevic offered a counter-argument to my opinion worth taking seriously. “Someone, often not a native English speaker, tells a chatbot what he really thinks about a fragrance and lets it shape the words,” he wrote. “The review still carries his true opinion, even if it wears an LLM accent.”
If a reviewer’s authentic opinion is present, even shaped by an LLM, does the prose style matter? I think it does, for two reasons. The LLM does not just shape words. It shapes what gets said and what gets omitted. A chatbot trained on thousands of fragrance reviews will reproduce the consensus vocabulary and suppress the dissenting observation, which is precisely the conformity problem already identified. And the reader has no way of knowing whether the opinion behind the LLM prose is genuine or whether the LLM filled in opinions the reviewer never formed. That distinction is invisible on the page.
To his credit Knezevic agreed with my follow-up clarifier that I was not against AI, but that AI should assist writers rather than replace them. On that principle we are in complete agreement. AI can be very useful in the hands of professional writers, amplifying already strong prose when used properly. Where we differ is on how consistently human input is being upheld in reviews currently visible on the platform.
AI assistance in writing is not the problem. I use it myself as a tool that my human judgment shapes into something worth reading. The problem is when AI is used to replace human thought. A review generated and published without a single moment of human critical engagement is not a review, it’s a content placeholder that happens to mention a fragrance.
The third problem is less visible and more insidiously consequential. Fragrantica creates separate pages for reformulations only when brands officially announce the change, which means quiet reformulations. The most common and consequential kind of reformulations for the most part go entirely undocumented on a single page. The platform’s reformulation record is therefore dependent on the honesty of the very industry it is supposed to assess independently. When a house changes a formula without announcing it to the world, earlier Fragrantica reviews do not disappear. On the contrary, they continue propping up a rating for a product they never assessed. I documented this in my review of Calvin Klein Escape earlier this year. The archive had preserved the praise, it had not preserved the caveat.
The fourth problem flows directly from the third. A single thumbs up count cannot distinguish personal preference from craft judgment, skin chemistry from compositional quality, a single spray in a store from a month of daily wear. The overall rating tells you whether people liked something. It cannot tell you whether it deserved to be liked, whether what you buy today resembles what the reviewers experienced, or whether the fragrance coheres at all. It is a number that feels like information and functions as noise.
The fifth problem is the platform’s most damning failure because it resembles education while producing the opposite. Fragrantica invites users to confirm or deny individual notes, a feature that gamifies ingredient identification and creates the illusion of critical skill. Knowing that a fragrance contains iris and ambroxan is not the same as knowing whether the iris and ambroxan cohere. The platform rewards the former and has no mechanism for the latter.
Actual expertise in knowing how the notes fit together, which is the kind of criticism I desperately crave, does not exist on Fragrantica in any meaningful form.
The sixth and final problem explains why none of the previous five have been corrected. There is apparently no accountability for how users produce their reviews. A review written after one spray in a store sits identically next to one written after years of daily wear. Furthermore, a review from someone who received the bottle free from a brand is indistinguishable from one written by an independent buyer. Other serious review platforms — in film, in literature, in food — require disclosed methodologies and declared conflicts of interest. Fragrantica requires none of these things and calls the result a community.
There is a further dimension nobody discusses because it is the most uncomfortable to name.
I write about fragrance from the intersection of different worlds — Pakistani, American and Canadian, devout and non-religious, Eastern and Western — and that dual position has taught me how differently the same fragrance can read depending on where you stand culturally when you smell it.
A reviewer in New York City who finds an oud fragrance medicinal is telling you something real. So is a reviewer in Ankara for whom that same oud reads as familiar and correctly weighted. Both reviews appear identically weighted on the page.
The same applies across age and economics. A senior collector who insists pre-reformulation fragrances are categorically superior is operating from a specific historical relationship with fragrance that younger buyers do not share. A young reviewer who defends a clone as equal or superior to the original may be doing so partly because the original is financially inaccessible — a perfectly understandable position that nonetheless shapes the review in ways the reader cannot see. Nobody is obligated to disclose who or what they are. But without that context the reader is absorbing opinions they cannot properly weigh.
These six problems are the natural consequence of a platform that prioritised scale over standards, growth over accountability and engagement over truth. Fragrantica became the biggest fragrance resource in the world. It never asked whether biggest was the same as best.
The platform optimised for everything except the one thing its community actually needed: the ability to trust what it found there.
None of this is an indictment of every person who has ever left a review on Fragrantica. The platform has genuinely professional contributors — knowledgeable, independent, generous with their time, asking nothing in return except to be useful to someone they will never meet. The tragedy is that the platform’s architecture makes them genuinely difficult to find. Their honest assessments sit alongside pre-release impressions, AI generated placeholders and socially pressured consensus with no signal to help a reader tell the difference.
The problem is not the absence of good people on Fragrantica. It is that the platform has built no mechanism for making their voices rise above the noise.
𝐖𝐡𝐚𝐭 𝐦𝐮𝐬𝐭 𝐜𝐡𝐚𝐧𝐠𝐞 — 𝐚𝐧𝐝 𝐰𝐡𝐚𝐭 𝐲𝐨𝐮 𝐜𝐚𝐧 𝐝𝐨 𝐧𝐨𝐰
Fragrantica will not change because critics like me say it should. It will change when the pressure to change becomes greater than the comfort of staying the same. That pressure has to come from somewhere. It might as well start here.
These changes will not be welcomed by a platform that has prospered without them. They are nevertheless necessary. Knezevic himself acknowledged in our exchange that there is always something to improve and that some of my criticism of the platform he already thinks about a lot. I take that at face value. A co-founder who has built something used by millions of people and still thinks about how to improve it is certainly worth respecting.
The good news is that Fragrantica can improve. I offer four starting points that would vastly improve the site. The solutions I mention below are not technically complex, but they do require the site takes action.
First, require real names. Nobody’s safety depends on their right to tell strangers on the internet that a fragrance has poor longevity without attaching their name to the claim. Real names introduce consequence into a space that currently has none.
Second, require ownership disclosure. A mandatory checkbox — I own this fragrance, I have sampled it, I have not yet tried it — costs the platform nothing to implement and gives readers information they currently have no way to access. Its absence is the most revealing thing about what the platform prioritises.
Third, introduce editorial oversight for AI-generated content. Knezevic confirmed that Fragrantica is actively rolling out measures to protect community integrity as the technology evolves and does not tolerate reviews with no genuine person behind them. That is encouraging. The question is whether those measures will be visible enough to readers to restore trust, or whether they will remain invisible infrastructure that leaves the problem looking the same from the outside.
Fourth, introduce a reviewer credibility system, a score based on review history and demonstrated expertise, not follower counts. Reviews from high credibility users should surface first. Parfumo has implemented exactly this kind of system. That Fragrantica has not done the same is a choice, not a limitation. This single change might have surfaced the one Hacivat reviewer who correctly identified the fragrance’s split personality.
Knezevic closed his response with an invitation: “Send your readers our way, the door is open, and the work speaks for itself.” I am taking him up on it. Here is how to use Fragrantica without being used by it.
𝐇𝐨𝐰 𝐭𝐨 𝐮𝐬𝐞 𝐅𝐫𝐚𝐠𝐫𝐚𝐧𝐭𝐢𝐜𝐚
Trust Fragrantica’s staff written content above all else — notes, launch years, perfumers, concentration formats, identifiable by the editorial byline at the top of the page rather than a username. For reviews, prioritise those from trained and recognised fragrance experts over anonymous community posts. These insights written by professionals who understand the craft of perfumery are the platform’s most reliable and insightful offering.
Community discussion threads, despite their limitations, often surface dissenting voices that formal reviews miss, but finding them is difficult. To separate the wheat from the chaff, I think Fragrantica should monitor early reviews of newly released fragrances heavily and remove those clearly written with an agenda. The conformity machine is most active in the first weeks of a release. Treat them as impressions at best and planted consensus at worst.
I recommend ignoring the overall rating of any fragrance that has been in circulation long enough to have been reformulated without makers’ announcing it. Such ratings are an average of opinions about multiple different products presented as a single product with one verdict. That is not information. That is noise wearing a number.
Finally, look for dissenting voices. The reviewer who noticed something the consensus missed may be on to something. The observation that received three thumbs up while the agreeable paragraph above it received three hundred is frequently the most honest thing on the page. Fragrantica’s architecture buries it. Your job as a reader is to dig.
Fragrantica has everything it needs to be the platform its community deserves, and I believe Knezevic knew it when responding to my critique of his site. His willingness to engage with my article honestly and without defensiveness suggests not only that the conversation is not closed but that the site’s leadership has the will to make Fragrantica better. What is needed now is clearly not goodwill but tough decisions to make the site better for the tens of millions of people who come to the platform every month trusting it to serve them well.
The community has the power to accelerate those decisions, by writing reviews that disclose who they are and how they tested what they reviewed, by holding each other to standards the platform does not yet impose, and by giving their attention and their money to platforms that earn both.
And you, reading this, have already taken the first step. You know what the problems are. The next time you open Fragrantica, you will not be able to unknow them. Use that.
𝘈𝘭𝘪 𝘉𝘰𝘬𝘩𝘢𝘳𝘪 𝘪𝘴 𝘵𝘩𝘦 𝘍𝘰𝘶𝘯𝘥𝘦𝘳 𝘢𝘯𝘥 𝘌𝘥𝘪𝘵𝘰𝘳-𝘪𝘯-𝘊𝘩𝘪𝘦𝘧 𝘰𝘧 𝘈𝘭𝘪 𝘗𝘦𝘳𝘧𝘶𝘮𝘦𝘸𝘢𝘭𝘢, 𝘢𝘯 𝘪𝘯𝘥𝘦𝘱𝘦𝘯𝘥𝘦𝘯𝘵 𝘧𝘳𝘢𝘨𝘳𝘢𝘯𝘤𝘦 𝘱𝘶𝘣𝘭𝘪𝘤𝘢𝘵𝘪𝘰𝘯.
