Skip to content
Parfumedia

Data & Methodology

This page explains where Parfumedia’s data comes from and how we process it, how we show community statistics, how similar perfumes are computed and how often the data is updated.

Where does the data come from?

Our database is built from perfume data collected over the years from various sources. We don’t publish that raw data as it is; we distill it:

  • Merging and deduplication: records describing the same perfume are merged into one and duplicates are removed.
  • Normalization: brand names, notes and concentrations are brought into one consistent form.
  • Translation and correction: note names and other labels are translated into Turkish for the Turkish version of the site, and wrong or inconsistent note names are corrected.
  • Enrichment: every perfume gets a permanent page address, and its data completeness is assessed against the thresholds below.
  • Analysis: community vote distributions are computed together with their vote counts, and similar perfumes are derived from note and accord overlap.

Each perfume record consists of brand, name, release year, gender, concentration, perfumer, the note pyramid (top, heart and base notes), main accords and community statistics.

Community statistics

Ratings and vote distributions come from the votes of a user community; they are not Parfumedia’s own assessment. That is why every statistic is shown with its number of votes: a result based on a few votes is less reliable than one based on thousands.

  • Rating: the community’s average rating and the total number of votes.
  • Distributions: the share of votes per option in categories such as longevity, sillage, season and occasion; the option with the most votes is highlighted.
  • This data is not presented as reviews and is not marked up as rating structured data for search engines.

How are similar perfumes computed?

Similarity is a weighted sum of two measures: note overlap (60%) and accord overlap (40%). Overlap is the number of elements two perfumes share divided by the number of distinct elements they have in total (Jaccard similarity).

Among candidates with close similarity, the perfume with more votes, the better-known one, is moved slightly ahead. This popularity effect can raise a candidate’s ranking score by at most 30%, so it never overtakes a much closer match.

The calculation is based on notes and accords only. Scent perception differs from person to person, so treat the results as suggestions.

How are dupe sources assessed?

Every dupe row carries a source label showing where the information comes from. The label tells you how reliable the match is.

Stated by the maker: the brand that makes the dupe states, on its own website or packaging, which perfume the product is inspired by.

What counts as an independent source? A source that is independent of whoever makes or sells the dupe and describes the match based on its own testing. Pages by the dupe's maker or seller, pages tied to an affiliate deal, and content copied from another source do not count as independent. Generic list articles without first-hand testing do not count either.

Several independent sources: at least two independent sources report the same match separately from each other.

Single source family (community): the match comes from one community source only. Sites that copy each other count as one source. Treat these matches with more caution.

We show the date each source was checked next to the dupe. If you spot a wrong or outdated match, please write to us.

Data completeness and search engines

Every perfume page is open to visitors. However, we only recommend pages with enough data to search engines and list them in the sitemap. A perfume page counts as indexable when it has:

  • at least 10 community ratings,
  • at least 3 notes,
  • vote distributions in at least 2 different categories.

Pages below these thresholds are marked “noindex” for search engines. Each page’s status is recalculated automatically as new data arrives.

Update cadence

The database is updated through periodic data imports. After data updates, indexability and similar-perfume lists are recalculated. Because pages are cached for speed, a change can take up to about a week to appear on the site.

Reporting errors

If you spot wrong or missing information, send the perfume’s page address and your suggested correction to [email protected]. We review every report and apply the corrections we can verify.