Content, Moonstone, Aura: Layer 1 of Discover's editorial foundation (55% of FR volume)
Discover's Layer 1 in French concentrates 55% of the feed's volume. It's composed of 5 pipelines forming a recycling loop. Here are the attested FR-specific figures and the actually actionable levers.
Discover's editorial Layer 1 is the set of 5 pipelines concentrating 55% of the French Discover feed's volume. This figure comes from 1492.vision's analysis of 42 million Discover cards between December 2025 and February 2026. Here's what the data reveals — and how to activate each of the 5 pipelines.
Discover's Layer 1 in French = 5 pipelines forming a recycling loop: content (entry, 30.7% FR volume) → moonstone (engagement amplification, 19.3% reach — 2× more than EN) → aura (diversification, 13.9% volume) → paginationpanoptic (scroll) → relatedcontentruby (click). Combined, this layer carries 55% of FR volume. Primary source: 1492.vision (42M cards). Metehan Yesilyurt's independent decompilation of the Google App SDK confirms the architecture.
Where the data comes from — three converging sources
Three public analysis bodies converge on Discover's pipeline architecture:
- 1492.vision — Sylvain Deauré & Damien Andell. The reference platform for Google Discover monitoring. Their public research on 42 million Discover cards (Dec 2025 - Feb 2026) identified and named Discover's 20+ pipelines and published their volume shares per market (FR figures differ significantly from EN ones). It's the primary source for this article.
- Metehan Yesilyurt — metehan.ai. Decompilation of the Android Google App SDK: 87,498 Java classes analyzed. Confirms the pipeline names observed by 1492.vision from the behavioral angle, via the code-source angle, and documents the absolute priority of
JSON-LDover Open Graph in Discover's markup parsing. - The May 2024 Content Warehouse leak. Discovered by Erfan Azimi, surfaced by Rand Fishkin (SparkToro) and analyzed by Mike King (iPullRank). 2,596 modules, 14,014 attributes. Documents the signals feeding content evaluation (
contentEffort,chard_score,OriginalContentScore) without naming the Discover pipelines.
SearchEngineLand's February 2026 study is a relay of 1492.vision with aggregated (all-languages) figures. We use the FR-specific 1492.vision figures here, which are more precise for a French publisher.
Layer 1's 5 pipelines (FR)
The key number: these 5 pipelines, together, serve 55% of the French feed's volume (source: 1492.vision). That's massive. They form a recycling loop where an article can traverse several successive stages.
1. content — the entry highway (30.7% of FR volume)
- FR reach: 9.9% (share of devices seeing each URL/day)
- FR volume: 30.7% — the largest pipeline by volume
- Median age: 11 hours
- Dominant FR domains: YouTube (10.6%), Le Monde (8.7%), Le Figaro (7.5%), L'Équipe (7.3%), Ouest-France (6.1%), BFM TV (5.4%)
Nearly every article transits through content. The question isn't getting in — it's getting out to higher-reach specialized pipelines (moonstone, mustntmiss). The documented signal feeding content is contentEffort: « LLM-based effort estimation for article pages » in the leak's QualityNsrPQData module (Hobo Web analysis). In plain terms: Content rewards real editorial effort and originality.
2. moonstone — the French-market jackpot (19.3% reach)
- FR reach: 19.3% — 2× more than English (9.4%)
- FR volume: 12.9%
- Median age: 17.6 hours
- Dominant FR domains: Ouest-France (9.0%), BFM TV (8.8%), Le Figaro (6.8%), Le Monde (6.2%), L'Équipe (4.8%)
- Overrepresented topics: horoscope (×3.5), betting/gaming (×3.3), entertainment, weather, people
It's the #1 lever for a French publisher. moonstone picks ~4.5× fewer URLs than content and pushes them to 2× more devices. It's the engagement broadcast machine. If Ouest-France dominates it ahead of engagement pure-players, it's via the « local crime + national angle, weather + regional people » mix. To activate: drive CTR + read-time on popular topics.
3. aura — the intellectual diversifier (13.9% of FR volume)
- FR reach: 5.4%
- FR volume: 13.9% — 2nd-largest pipeline by volume
- Median age: 1.46 days — 3× older than content
- Dominant domains: broad distribution, no marked concentration
- Overrepresented topics: business (×1.5), consumer electronics (×1.52), cycling (×1.42), rugby (×1.33)
aura is the anti-moonstone. Where moonstone concentrates audience on a few popular articles, aura diversifies: ~3.5× more URLs at lower reach. It's the pipeline surfacing content users didn't search for. 1492.vision's hypothesis: cross-user sourcing — what readers similar to you read, rather than what directly matches you. Activation lever: depth, originality, topical expertise.
4. paginationpanoptic — triggered by scroll (×7 in 3 months)
Infrastructure pipeline triggered when a user scrolls past the first screen. Growing fast (×7 in 3 months per 1492.vision). To activate as a publisher: a reading experience that holds attention (long articles, short paragraphs, visual breaks, frequent H2s). If your visitors bounce, this pipeline stays closed.
5. relatedcontentruby — triggered by click
When a user clicks on your article, your other related articles can appear in the feed. Lever: quality internal linking and regular publishing on the same topic to densify the topical cluster.
The cascade as a recycling loop
The 5 pipelines aren't independent: they form a loop. An article entering through content can be amplified by moonstone if engagement is strong, then diversified via aura to neighboring audiences, then extended by paginationpanoptic (scroll) or relatedcontentruby (click). Articles that « explode » in Discover aren't those passing one pipeline — they activate 3 to 8 simultaneously. 1492.vision observes 58% of FR URLs are in at least 2 pipelines, and 25% reach 4 or more.
The right question isn't « is my article in Discover? » — it's « how many pipelines is it visible in? ».
Layer 1 in FR vs EN — moonstone is the key asymmetry
1492.vision documents a major difference: moonstone has 2× the reach in FR (19.3%) than in EN (9.4%). The French market is structurally more dependent on the engagement-amplification pipeline than the English one. This explains why a French publisher mastering moonstone's codes (horoscope, weather, people, betting) can reach volumes their English equivalents never will using the same levers.
Another significant asymmetry: neoncluster (video broadcast, 13% reach in EN) is nearly absent in FR (only 36 occurrences in 3 months). The FR feed is therefore more editorial and less video than the EN feed.
Action plan — activating Layer 1
- For
content— flawless JSON-LD (absolute priority per Metehan Yesilyurt's decompilation), real editorial effort (documentedcontentEffortsignal), originality (OriginalContentScore). Our Discover eligibility audit runs 25+ checks in a minute. - For
moonstone— engagement (CTR + read-time). Topics that work: horoscope, weather, people, entertainment. Strong hero image. Our Image Validator tests title ↔ image alignment. - For
aura— depth (content 16% longer than average per 1492.vision), original angle, topical expertise. Business, electronics, science/tech overrepresented. - For
paginationpanoptic+relatedcontentruby— a holding reading experience, quality internal linking, regular publishing on the same topical cluster.
Sources
- 1492.vision — Sylvain Deauré & Damien Andell (primary source: 42M Discover cards analyzed Dec 2025 - Feb 2026, FR-specific figures per pipeline)
- Metehan Yesilyurt — metehan.ai (Google App SDK decompilation, 87,498 Java classes, independent validation of pipeline architecture)
- SearchEngineLand — Inside Google Discover: 20 pipelines, 42 million cards (relay with aggregated all-language figures)
- iPullRank, Mike King — Secrets from the Google Algorithm Leak (
contentEffort,chard_scoresignals from the 2024 Content Warehouse leak) - Hobo Web — What is Google's Content Effort Signal? (technical detail on the
contentEffortsignal)
Take action in 1 minute
Three free tools the editorial team uses daily — tested across French and international publishers.
📘 Want to go deeper? Grab the free Discover Essentials ebook (33 pages, 25-min read).
Frequently asked questions
What is Discover's Layer 1 editorial foundation?
It's the set of 5 foundational pipelines concentrating 55% of the French Discover feed's volume (per 1492.vision's analysis of 42 million cards). The 5 pipelines: content (mandatory entry), moonstone (engagement amplification), aura (diversification), paginationpanoptic (triggered by scroll), relatedcontentruby (triggered by click). It's a recycling loop: an article enters via content, may be amplified by moonstone, extended by aura, etc.
What are the FR-specific figures for each pipeline?
Per 1492.vision (FR dataset, December 2025 - February 2026): content 9.9% reach, 30.7% volume; moonstone 19.3% reach (2× more than in EN), 12.9% volume; aura 5.4% reach, 13.9% volume; paginationpanoptic + relatedcontentruby complete the combined 55%. Note: these FR figures are sensibly different from the aggregated (all-languages) figures SearchEngineLand publishes.
Why is moonstone "the king pipeline" in France?
Because it has 2× more reach in FR (19.3%) than in EN (9.4%) — a major asymmetry attested by 1492.vision. moonstone selects ~4.5× fewer URLs than content but pushes them to 2× more devices. Overrepresented topics: horoscope (×3.5), betting/gaming (×3.3), entertainment, weather, people. Dominant FR domains: Ouest-France (9.0%), BFM TV (8.8%), Le Figaro (6.8%). For a French publisher, it's the #1 reach lever.
What's the role difference between Content, Moonstone and Aura?
Content is the entry highway: almost every article passes through (30.7% of FR volume). Moonstone is the engagement broadcast machine: few URLs, lots of reach. Aura is the anti-moonstone: it diversifies by exposing ~3.5× more URLs on less mainstream topics (business ×1.5, consumer electronics ×1.52, science/tech). Aura has a median age of 1.46 days, 3× older than content — it's the pipeline that extends article lifespan.
What concrete actions activate Layer 1?
Four attested levers: (1) for content, clean JSON-LD markup (absolute priority documented by Metehan Yesilyurt's SDK Google App decompilation); (2) for moonstone, drive engagement (CTR + read time) on popular topics (horoscope, weather, people); (3) for aura, depth and originality (content 16% longer than average); (4) for paginationpanoptic and relatedcontentruby, scroll experience and internal linking.
