What people try to understand in their dreams

Figures come from Google Search Console (sc-domain:midnightmind.app). Window: 6 July – 2 August 2026 (28 days), before a later traffic drop. Each query is weighted by impressions per day. We do not use user dreams: no consent, and the internal corpus would be too small to cite.

GSC does not expose every query (some are anonymized). The 4,877 named queries weigh 1,039.31 impressions/day, versus 15,816.8 for the property. Every percentage on this page is inside that named corpus — not a survey of a country, not all site traffic.

A symbol is what remains after stripping “dreaming of”, “rêver de”, “meaning”, and a few frames (“en islam”, Ibn Sirin). Language is detected from rêv/rev/songe versus dream. Categories use a fixed taxonomy, assigned by a language model in batches, then checked on 50 rows (96% agreement).

199 queries, 6.4% of named impressions. The word “islam” alone is 5.5%; Ibn Sirin appears in 31 queries; the biblical frame, 2.

That frame is almost only French-language: 11.7% of FR impressions versus 0.2% in English.

Inside the “islam” subset, the head of the list is singe, examen and cimetière — concrete images, not a sacred lexicon.

The rest, 92.7%, is labelled curious: no token from the list in the query. The list lives in the script, not in the model.

French is heavier: 10.1% versus 4.3% in English. Cemeteries drive much of that gap. It is a request for a frame, not a nightmare census.

These figures do not say what the French dream about. They say what they type into Google when a dream has marked them enough to search — and, more precisely, what they type when it is our blog that answers. It is a narrow window. It still has something to show.

First, the concrete. People do not search for “dreaming of fear” or “dreaming of freedom”. They search for rain, rice, a cemetery, a monkey. Abstract categories — emotion, death, spirituality — together weigh less than 5%. The dream people want to understand is the one that has an image.

Next, the frame. More than one Francophone query in ten adds “en islam” or cites Ibn Sīrīn. In English, almost none. This is not a detail: a significant share of French-language readers are not looking for an interpretation, they are looking for an interpretation in their tradition. And the symbols in question are not religious — they are the monkey, the cemetery, the failed exam. Everyday images, read inside a received frame.

Finally, unease is the minority. 7% of queries contain a word from the list we fixed in advance — death, drowning, falling, attack. The cemetery accounts for half of that on its own, and it is more often a place to decode than an anxiety to soothe. People who look up the meaning of a dream are, in their great majority, curious rather than frightened.

We publish these data because they did not exist elsewhere, with their limits: a single site, less than 7% coverage of actual queries, four weeks of summer. The full file is downloadable. If you draw something else from it than we did, write to us.