GEO for Home and Fragrance Brands: Describing Scent to a Machine That Cannot Smell

Every category in this series has a retrieval problem. Fragrance has the hardest one, because the attribute that decides the purchase cannot be transmitted as text at all.

You can photograph a chair. You can specify a leather. You can state a calorie count. You cannot render a smell, which means the entire burden falls on description, and the description convention the industry inherited from perfume marketing is atmospheric prose that contains no extractable fact whatsoever.

The result is predictable. Ask an engine to recommend a candle that smells like a wood fire without being sweet and the citations come from retailers and review sites, because those sources describe scent in comparative, structural language while the brands that made the candles wrote about winter evenings.

Why Evocative Copy Fails Retrieval

There are 3 separate failures inside the atmospheric convention.

The first is that it is unfalsifiable, and unfalsifiable statements make poor citations. A sentence about a villa at dusk cannot be right or wrong, so an engine assembling an answer has nothing to anchor to.

The second is that it produces no comparison attributes. When an engine builds a shortlist, it needs axes. Scent family, note structure, intensity and burn time are axes. Mood is not, so a brand publishing only mood appears in no comparison at all.

The third is that it does not match query language. People search for what a thing smells like using reference points and plain words: smoky, not sweet, like a real fireplace, similar to a discontinued product. Copy written in a different register than the query is a poor match regardless of how good the product is.

None of this argues for abandoning beautiful copy. It argues for publishing the structure underneath it as well, so the page carries both the persuasion and the facts. Our post on GEO for beauty and cosmetics brands covers the ingredient-transparency version of the same lesson.

Notes and Accords as Structured Text

The note pyramid is a convention that already exists in the industry and is frequently rendered as a graphic, which wastes it.

• Publish top, heart and base notes as text lists, individually named.
• Name the accords, meaning the combinations that produce the overall impression, since accords describe the experience better than an ingredient list does.
• State the dominant note explicitly. A composition with 9 notes still smells mostly of 1 or 2 of them, and saying which is more useful than the full list.
• Distinguish what is naturally derived from what is synthetic, without treating either as a virtue, because both have legitimate roles and buyers ask.
• State the concentration where it applies, since an eau de parfum and an eau de toilette behave differently and the difference is a number.
• For candles and diffusers, state the fragrance load as a percentage where you are willing to, because it is the single most predictive figure for throw. An 8 percent load and a 12 percent load are different products and no amount of copy closes that gap.

Then put the important ones into markup. additionalProperty accepts scent family, dominant note, concentration, burn time and wax type as structured pairs, which is how these attributes reach a comparison table. Our guide to product schema covers the implementation.

Description by Comparison

Comparison is the only scent description that reliably transfers, because it borrows a reference the reader already has in their memory.

• Anchor to a widely known scent. Smells closest to fresh cedar shavings with a faint smoke behind it does real work.
• Anchor to a category reference. Sits between a classic fig and a green herbal, closer to the fig.
• Anchor to your own range, which is both useful and internally linkable. Drier than our bergamot, with more woodiness and less citrus.
• Describe against a discontinued or widely loved reference where the comparison is honest, since replacement searching is a genuine and high-intent behaviour.
• State what it is not, which narrows expectations fast. Not sweet, not powdery, not gourmand.
• Give a plain-language summary alongside the technical one, because most buyers do not know what a chypre is and will not look it up.

That last point is a real barrier. The industry's own vocabulary is precise and largely unknown to buyers, so publishing both registers, the technical term and the plain description, captures 2 different query populations with 1 sentence.

Quantifying Intensity, Longevity and Throw

Several attributes buyers care about are genuinely quantifiable and almost universally left as adjectives.

• Intensity on a stated scale, with the scale defined. A 4 out of 5 means nothing unless you say that 5 is noticeable from an adjoining room within 10 minutes.
• Throw, meaning how far the scent carries, expressed in room size. Scents a room of up to 300 square feet is a fact.
• Burn time in hours, and the wax and wick that produce it. A 60 hour burn and a 40 hour burn at the same 8 ounce fill are telling you something real about the wax, and buyers comparing 2 candles at 48 dollars deserve both numbers.
• Longevity on skin or in a room, in hours, with the condition stated. Noticeable for about 5 hours in a closed room and about 2 in an open-plan space is a pair of figures that answers the question properly.
• Cold throw versus hot throw for candles, described separately, since they differ and buyers who know the distinction are the ones buying repeatedly.
• Time to fill a room, in minutes, which is a real question for a diffuser and is almost never answered.

These figures are the raw material of every comparison an engine could build. A brand publishing 6 quantified attributes appears in comparisons; a brand publishing 0 appears only when it is named directly. That is the difference between being 1 of 5 options a buyer considers and being a destination they had to already know about.

Room, Surface and Use Context

Home care and home fragrance both live or die on context, and context is where the query volume actually is.

• Which room, and why. A kitchen scent and a bedroom scent are different recommendations for reasons worth stating.
• Which surface, for cleaning products, named specifically including what to avoid.
• Which season or time of day, which is a real purchasing heuristic and an easy content set.
• Layering guidance across products in your own range, which is both useful and a merchandising asset.
• Practical constraints: pets, small children, asthma, open-plan spaces, rentals with no ventilation.
• Occasion use, from hosting to gifting to the specific case of masking cooking smells, which is searched far more than brands expect.

The pets question deserves special handling. It is asked constantly, it has a genuine safety dimension for certain ingredients and certain species, and a brand that answers it plainly, with a clear instruction to consult a veterinarian where relevant, owns a query set that competitors avoid entirely.

Ingredient Transparency and Clean Claims

Fragrance ingredient disclosure is an area where regulation permits collective declaration, which makes voluntary transparency a genuine differentiator rather than a compliance exercise.

• Publish the full ingredient list as text, including fragrance components where you are willing to disclose them. The FDA notes that fragrance may be declared collectively, so going further is a choice worth stating.
• Name what you exclude specifically, with the compound named rather than a category of concern.
• Be careful with clean, natural, non-toxic and eco-friendly, all of which sit under the FTC's Green Guides when used as environmental claims and require substantiation.
• Avoid claims that push a cosmetic toward drug status, which the FDA's labelling claims guidance addresses directly.
• Publish safety testing and certifications with the certifying body named.
• Publish the wax, wick, base and packaging materials, since these are real differentiators and easy to state.

The pattern holds across every regulated category in this series. The substantiated version of a claim contains a compound, a threshold or a body, and those specifics are what make it citable.

Mining Reviews for the Real Vocabulary

Your customers have already solved the description problem and written the results on your own site.

Read 200 reviews and extract the recurring descriptors, counting them. When 30 of 200 reviewers independently reach for the same comparison, that is not an anecdote, it is the consensus description of your product written by the market. They will not be your marketing words. They will be phrases like smells like a fancy hotel, or like a bookshop, or reminds me of my grandmother's house, and those phrases are how the product is actually being searched for.

• Publish the recurring comparisons as part of the product description, attributed as how customers describe it.
• Note where customer perception and the official description diverge, which is genuinely useful and rare.
• Capture the negative descriptors too, since too strong for a small room is guidance rather than a complaint.
• Feed the vocabulary into scent family pages and comparison content.

Do this with real reviews only. Manufactured review language is detectable and in a small category it is obvious, and our post on synthetic reviews covers what gets caught.

Scent Families as Navigation and Retrieval

Scent family is the natural organising structure for the category and most brands use it as a filter rather than as content.

Build a real page per family. Explain what the family smells like in plain language, name 3 or 4 classic reference points, list which of your products sit in it and how they differ from each other, and describe who tends to like it and what they often move to next. A range of 24 products usually resolves into 5 or 6 families, which is 6 substantial pages rather than 24 thin ones.

This does 3 jobs at once. It captures a large set of unbranded informational queries. It gives an engine a structured explanation of a concept it will otherwise take from a retailer. And it creates the internal linking spine that connects individual products into a coherent, understandable range, which is what our post on GEO for DTC brands describes as the difference between a catalogue and an entity.

Common Mistakes

• Atmospheric copy as the entire description. Unfalsifiable, unquotable, and produces no comparison axes.
• Note pyramid published as a graphic. The one piece of structured data the industry already produces, wasted.
• No dominant note stated. A 9 note composition smells mostly of 2, and buyers want to know which.
• Throw and burn time as adjectives. Both are numbers and both are decisive.
• Clean and non-toxic used loosely. Environmental and safety claims carry substantiation obligations.
• Ignoring the pets question. High volume, real safety dimension, avoided by nearly everyone.
• Marketing vocabulary that does not match review vocabulary. The search language is in your own reviews.
• Scent families as filters, not pages. Discards a large unbranded query set and the linking spine with it.

Implementation Sequence

1. Convert every note pyramid from image to text, with top, heart and base notes named individually.
2. State the dominant note and the accords in plain language alongside the technical terms.
3. Add comparison anchors to every product: a known reference, a position between 2 references, and a statement of what it is not.
4. Quantify intensity on a defined scale, throw in room size, burn time in hours, and longevity in hours.
5. Publish room, surface, season, layering and pets guidance as real content.
6. Publish the full ingredient list, name exclusions by compound, and substantiate any clean or environmental claim.
7. Mine 200 reviews for recurring descriptors and publish the vocabulary as customer language.
8. Build a page per scent family with references, internal comparisons and audience guidance, and mark up scent attributes with additionalProperty.

Frequently Asked Questions

Why do retailers get cited for scent descriptions instead of the brands?

Because retailers describe scent structurally and comparatively while brands describe it atmospherically. An engine needs an anchor, and a note list with a dominant note and a comparison reference gives it one. Mood copy gives it nothing to extract.

Does publishing technical fragrance terms alienate ordinary buyers?

Not if you publish both registers. Name the family or accord for readers who know the vocabulary, and give a plain-language description alongside for the majority who do not. Two sentences capture 2 different query populations.

How do you quantify something as subjective as throw?

By stating the room size it is designed for and the conditions assumed. Scents a room of up to 300 square feet is a testable, useful figure. It is approximate, which is fine, and approximate with a number stated beats precise-sounding adjectives with none.

Should we disclose fragrance ingredients when the law allows a collective declaration?

Disclosure is a real differentiator precisely because it is not required. If you are not willing to publish the full composition, say what you exclude by name and explain what you do disclose, which is still far more than the category norm.

Is customer review language really worth publishing?

It is the vocabulary the product is being searched with, written by the people who bought it, sitting on your own site for free. Publishing the recurring comparisons as customer language captures queries your marketing copy is not phrased to match.