Home fragrance
No brands in common, but the same buyer usually asks both.
We asked an assistant, not a panel. Sleep environment, measured 2026-09-06.
We asked a grounded assistant what actually makes a bedroom sleepable, in three phrasings, and wrote down every brand it named. Here is what came back, what a buyer should do about it, and why we make nothing here yet.
Across three phrasings of the question what actually makes a bedroom sleepable, one brand came back more than once: Plant Therapy. Every other name appeared exactly once, which means the field has a shortlist rather than a winner.
Nineteen distinct brands were named across three phrasings, captured 2026-09-06 against gemini-2.5-flash (google_search grounded). Eighteen of them appeared exactly once. Plant Therapy was named most often, holding 10% of all mentions. Brands we hold enough history on are linked to their record.
| Brand | Mentions | Also named for |
|---|---|---|
| Plant Therapy | 2 | named here only |
| Alaska Bear | 1 | named here only |
| Aromasong | 1 | named here only |
| Baloo Living | 1 | named here only |
| Boll & Branch | 1 | named here only |
| Brooklinen | 1 | named here only |
| Cozy Earth | 1 | named here only |
| Drowsy | 1 | named here only |
| IKEA ROSENMANDEL | 1 | named here only |
| Manta | 1 | named here only |
| Moonchild | 1 | named here only |
| Neal's Yard Remedies | 1 | named here only |
A mention is not an endorsement and nobody paid to be on this list. It is what one grounded engine said on one date.
Treat the repeated names as the shortlist and everything else as noise: a single mention from a grounded engine is closer to a coin flip than to a recommendation. Plant Therapy came back more than once, which is weak evidence but the only evidence the answer space offers.
Contestability here is 90 out of 100. That describes how far the field is from having a leader. It says nothing about how many people are asking, and it is not a forecast of anything.
We have not written the buyer's guide for this question yet, so this page carries the measurement and the field and nothing more. We would rather publish a short honest page than pad it out with material we have not actually researched.
A claim is worth what its instrument is worth. So here is the instrument.
Captured 2026-09-06 against gemini-2.5-flash (google_search grounded), three phrasings of the same buying question, brand names recorded verbatim. Mentions, not rankings.
The same buying question asked several ways: category, situation, and open.
Retailers excluded. Amazon is where an answer gets bought, not a brand competing to be it.
Share of all 20 mentions held by Plant Therapy, the most-named brand here.
How far the field is from having a leader. Not demand, and not a forecast.
Directional, not decision-grade. Prompt sampling reads an output distribution, and published work on ordered brand lists finds very low run-to-run consistency. We publish the limit because the limit is what makes the rest worth reading.
Eighteen of nineteen brands were named exactly once.
When almost every brand appears once, the engine is not reporting a consensus. It is reaching for whatever the retrieval layer surfaced that minute.