How to read a Fragrantica entry, a critical reading in 9 points

What Fragrantica really is

Fragrantica.com is a collaborative database dedicated to perfume, launched in 2007 by a team based in Serbia. The site catalogues more than sixty thousand references and aggregates technical sheets, user votes, and written reviews. It runs on a hybrid model: part of the information comes from houses, part from volunteer contributors, part from the algorithmic aggregation of member votes.

This mixed nature explains both the richness and the limits of the database. A Fragrantica entry is neither an official press file nor a laboratory analysis. Reading an entry means separating three layers: the official data published by the house, the contributor additions (extra notes, perfumer attribution, photos), and the statistical aggregates produced by the votes.

This guide proposes a method in nine points to identify these three layers and know when to cross-check with other sources. Three documented references appear as examples throughout the text: Aventus by Creed, Baccarat Rouge 540 by Maison Francis Kurkdjian, and Mitsouko by Guerlain.

Point 1 · Announced pyramid vs perceived pyramid

The olfactive pyramid is the most visible element of a Fragrantica entry. Three tiers (top, heart, base) list the notes, sometimes illustrated with small icons. The pyramid has two possible origins: the list published by the house at launch, or a list added by a contributor based on a press release, a packaging label, or a brand dossier.

Three cautions apply when reading it.

• Announced notes are a promise of structure, not a formula. Many materials cited (amber, oud, May rose, vanilla) refer to accords built mostly from synthetic molecules. An entry that mentions "oud" does not guarantee the presence of natural oud.
• The pyramid can change. When a formula is reformulated, the entry is not systematically updated. The displayed pyramid may correspond to the original version, an intermediate version, or the current version, without clear indication.
• Note order reflects a trajectory. A pyramid that opens on citrus and ends on leather signals a contrasted composition. A pyramid that lists a single family across all tiers (several roses) signals a more linear perfume.

The perceived pyramid, voted by users, is available further down the entry. Members can vote for the notes they detect. Comparing the announced pyramid with the voted notes is one of the most valuable signals on the site: a clear gap suggests either a reformulation or a marketing narrative distant from the actual experience.

Point 2 · Longevity, sillage, occasion, and season votes

Below the pyramid, Fragrantica displays vote bars for longevity, sillage, value, and compass-style charts for seasons and occasions. These bars are the aggregated average of votes from registered users. They rest on no laboratory measurement, despite the crisp display that may suggest otherwise.

Three points deserve attention when reading them.

The first is the number of voters. A widely consulted reference such as Aventus by Creed, launched in 2010, gathers several thousand votes: the average is statistically stable. A confidential reference may display sharply tilted bars based on fifteen or twenty votes, far less reliable. The number of voters is generally shown on hover.

The second is the demographic composition of voters. Fragrantica users are mainly English-speaking, with strong representation from the United States, the United Kingdom, and the Middle East. Their longevity and sillage criteria reflect specific cultural expectations: a fragrance rated "moderate" by a French wearer may be tagged "weak" by an audience used to higher concentrations.

The third is the self-reinforcement effect. A popular fragrance collects positive votes that attract new users, whose votes confirm the trend. The bars therefore measure not only the perfume itself but also its reputation at a given moment in time.

Point 3 · Main accords, an algorithm of votes

The main accords section displays a coloured bar chart that distributes the perfume across olfactive families: floral, amber, woody, fresh, sweet, fruity, and many others. The reading is intuitive: the longer the bar, the more dominant the accord. The implicit promise is that of an objective composition analysis.

The reality is different. Main accords are produced by an algorithm that aggregates the notes voted by users into predefined olfactive families. If four hundred members vote for "sandalwood" and "cedar", the woody accord rises mechanically. No chromatographic analysis takes place. The chart reflects the statistical consensus of voters, not the real formula.

This shift in framing changes the reading. A strongly marked "gourmand" main accord means the community perceives a sweet, vanillic, or caramelised dimension. It does not prove the presence of any specific raw material. Baccarat Rouge 540 (2015, Maison Francis Kurkdjian) shows for instance a marked amber and a marked sweet accord, consistent with the majority perception of a saffron, ambroxan, and cotton-candy heart: a perception, not a measurement.

Point 4 · Perfumer attribution

The perfumer name appears at the top of the entry, below the house name and the release year. Its reliability varies according to three cases.

First case, the house publicly communicates the name. This is the most common situation in niche perfumery since the early 2000s, and the most reliable. The information appears on the official site, in the press release, and on the Fragrantica entry.

Second case, the house stays discreet and a contributor adds a name based on an indirect source (interview, professional database, leak). The attribution may be correct, partial (one perfumer cited out of three), or wrong. The specialised press (Bois de Jasmin by Victoria Frolova, Now Smell This, Persolaise) regularly corrects attributions repeated elsewhere.

Third case, the perfume is old and the house has never communicated a name. The Fragrantica attribution may rely on a historical convention sometimes disputed. For Mitsouko by Guerlain (1919), the attribution to Jacques Guerlain is consensual, confirmed by the house archives and the documentation held at the Osmotheque in Versailles, France. The practical rule: if the attribution plays a role in a decision, it is verified on Basenotes, on Parfumo, and, when possible, on the official site.

Point 5 · User reviews and their biases

The reviews section gathers texts written by registered members. It is one of the site's great strengths, and one of its most slippery zones. Several biases meet here.

The cultural bias is the most important. Reviews come mainly from North American and Middle Eastern users. The judgment criteria (power, massive sillage, quality perceived through sweetness) are not universal. An enthusiastic review of a fragrance tagged "beast mode" by a user in Dubai describes a real experience, but one anchored in specific expectations.

The trend bias is the second. Since 2020, several references have seen their reviews explode after a viral moment on TikTok or Instagram. References such as Phantom by Paco Rabanne (2021) or the Lattafa lineup have seen waves of enthusiastic reviews tied to trends, followed by more nuanced returns once the wave passed.

The date bias is the third. A reformulated perfume accumulates reviews that describe different versions without an easy way to sort them. Reviews posted in 2012 on Aventus by Creed do not describe the same bottle as those posted in 2024, after several formula adjustments documented by the Basenotes community. Sorting reviews by date is a good practice for older references.

Point 6 · Discontinued, limited, reformulated

Several tags can appear next to a perfume name on Fragrantica: discontinued, limited edition, reformulated. These mentions are valuable, but they deserve a careful look.

The discontinued tag is added by contributors when a reference is no longer produced. The delay between the actual discontinuation and the entry update can stretch over several months. Some references tagged discontinued return as limited editions or boutique exclusives. The practical rule: check on the official site of the house before concluding that a perfume is permanently stopped.

The limited edition tag covers two realities. A true limited edition (closed production, numbered series, specific flacon) and a seasonal reissue relaunched every year on the pretext of rarity. The Fragrantica entry does not always make the call. The launch date and current availability usually indicate which case applies.

The reformulated tag is the trickiest. It signals that a new formula exists, without always specifying when the change happened, or which version the entry describes. The Basenotes and Parfumo communities often document the affected lots in finer detail (batch numbers, producer codes). For historic references, the documentation preserved at the Osmotheque in Versailles, France, is the reference for the original formula, to be compared with what the house sells today.

Point 7 · EDT, EDP, Extrait, and edition variants

Many references exist in several concentrations (eau de toilette, eau de parfum, extrait) or in several dated editions. Fragrantica generally creates a separate entry for each variant, with its own pyramid, its own votes, and its own reviews. Three pitfalls to know.

The first is confusion between variants. The pyramid of an eau de toilette often differs from that of the matching eau de parfum: lower-dosed materials, different accents, more present top. Checking the exact label before reading the votes is essential.

The second is the dated edition. For references such as Aventus by Creed, anniversary editions (10 years, 15 years) have been released with adjusted formulas. Reading the standard entry while owning a specific edition leads to comparing different objects.

The third is the packaging change. A flacon redesign does not always come with a formula change, and the reverse is also true. Basenotes threads remain the most precise resource on these cases, with comparative photographs of lots and tester evaluations.

Point 8 · Similar perfumes

The similar perfumes section offers a list of references judged close by users. The calculation rests on a vote system: members can flag that a perfume X reminds them of a perfume Y. It is a heavily consulted feature, and one to handle with care.

The logic of similarity varies from one voter to another. For some, two perfumes are close because they share an olfactive family. For others, because they trigger a common emotion. For others still, because they serve as cheaper alternatives. The list mixes these logics, and intent-driven votes (dupe hunting) can weigh heavily in the ranking.

The reasonable use treats this section as an exploratory cloud. A reference that appears deserves to be noted, tested on skin, and cross-checked with other tools. No olfactive equivalence can be declared from a statistical proximity. The guide dedicated to reading a perfume data sheet in five points details the general method, applicable beyond Fragrantica.

Point 9 · Cross-checking with Basenotes, Parfumo, and the official site

No Fragrantica entry is self-sufficient as soon as an important decision is at stake: a full-price purchase, a citation in an article, a verification of an attribution. Three cross-check sources are useful, each with its own strength.

Basenotes.com, founded in 2000 by Grant Osborne, is the oldest of the major English-language databases. Its community includes several professional evaluators and perfume historians. Threads are precise on reformulations, lots, and house history.

Parfumo.com, originally a German-language database, offers a more structured entry and a weighted rating system. Olfactive classifications are finer, and filters by gender, occasion, and season are stricter.

The official site of the house remains the primary source for two facts: the perfumer name as the brand claims it, and the exact launch year. For historic compositions, the documentation held at the Osmotheque in Versailles, France, and the supports from the Societe Francaise des Parfumeurs add precision that collaborative databases do not replace.

See also