Content, Moonstone, Aura: Layer 1 of Discover's editorial foundation (55% of FR volume)

Most publishers optimize for Google Discover as if it were a single judge to win over. That's a modeling error. Before a human ever sees your article, it has to clear a foundational layer made of three systems — grouped under the names Content, Moonstone and Aura. This Layer 1 does no personalization: it simply decides whether your content has the right to exist in the editorial foundation. In French it alone drives roughly 55% of the feed's volume. Everything starts here.

Why a "Layer 1" and why 55% in French

Discover doesn't evaluate every article with the same tools across markets. In English, Google has an abundance of external signals: dense citation graphs, rich Knowledge Graph entries, massive engagement history. The algorithm can therefore lean heavily on a publisher's reputation. In French, those signals are rarer and more scattered. As a result, Google falls back more on intrinsic content evaluation — exactly what Layer 1 can judge without external context.

That's precisely why Layer 1 captures an outsized share of FR volume. Where an English article can "pass" thanks to its domain's authority, a French article has to convince on substance alone first. Understanding Content, Moonstone and Aura means understanding 55% of your Discover fate — and it's also the most directly actionable part, because it depends only on what you publish.

This layered architecture isn't an isolated theory: it sits within the broader chain of Discover's 20+ pipelines. Layer 1 maps to the first classification and quality-scoring links — the ones that filter before any personalization.

Content — the judge of raw substance

The first system, Content, answers a single question: does this article add something, or recycle what already exists? It doesn't measure length or keyword density. It evaluates visible editorial effort — a concept documented in the 2024 Content Warehouse leak under the name chard_score (Content Effort).

Concretely, Content rewards the markers of real work: an original angle (not the 400th rewrite of a wire story), clean sources and data, a structured argument rather than a stack of interchangeable paragraphs. Conversely, it demotes "thin" content: generic summaries, paraphrases of other outlets, assembly-line articles with no added value.

How Content evaluates you

• Angle originality: does your treatment measurably differ from other articles on the same topic?
• Real depth: do you dig beyond the raw fact (context, consequences, analysis)?
• Verifiable effort: citations, sourced figures, elements only real work produces.

Failure signature: your article is indexed but gets a laughable impression volume while peers on the same topic take off. That's the classic symptom of weak Content — the piece is judged "already seen."

Moonstone — multimodal understanding

The second system, Moonstone, reads your article as a whole: text + hero image + HTML structure, together rather than separately. This is the major break from older text-only classifiers. Moonstone checks the coherence between what the title promises, what the image shows, and what the body delivers.

This multimodal coherence has become a decisive filter on Discover, a format where the hero image fills 60% of the card. If your title announces one thing and your visual tells another, Moonstone detects the dissonance and demotes the article — not as punishment, but because it (rightly) predicts a poor click-through rate and a fast bounce.

What Moonstone checks

1. Title ↔ image alignment: does the visual actually illustrate the title's promise?
2. Intrinsic image quality: resolution (≥ 1200px wide), sharpness, no heavy overlaid text, identifiable subject.
3. Readable structure: single H1, clear heading hierarchy, clean markup Moonstone parses without ambiguity.

That's exactly what our Image Validator tests: it audits your hero thumbnail against the criteria Moonstone applies, including title ↔ image semantic alignment. An off-topic visual is one of the most frequent — and most invisible — causes of Layer 1 failure.

Failure signature: article indexed, hero image replaced by a generic thumbnail (or your logo) in the feed, abnormally low click-through. Moonstone didn't validate your visual.

Aura — the judge of editorial quality

The third system, Aura, is the closest to what SEOs call "E-E-A-T." It evaluates trustworthiness and reading experience: who signs the article, is the publisher a recognized entity, is the page clean or saturated with intrusive ads, does the content inspire confidence?

Aura aggregates signals found scattered across the Content Warehouse leak: helpful_content_score, author signals, page-experience quality. But its function in Layer 1 is specific: it acts as a minimal trust filter. An article can be substantial (Content OK) and coherent (Moonstone OK) yet still fail Aura if it comes from a source Google can't credit.

Concrete Aura levers

• Clear byline: identified author, ideally with an author page and history.
• Recognized publishing entity: does your site exist for Google? That's what your Google Web Profile checks — a foundational identity signal, detailed in our piece on the Knowledge Graph and entities.
• Clean E-E-A-T markup: complete NewsArticle or Article JSON-LD, date, author, publisher organization.

Failure signature: article never pushed to Discover despite decent Search traffic, even though Content and Moonstone seem satisfied. Aura won't grant you its trust ticket.

The three in cascade: why order matters

Content, Moonstone and Aura don't vote separately and then average out. They run in a blocking cascade: an article killed by Content never reaches Moonstone; one rejected by Moonstone never reaches Aura. That's what makes diagnosis hard for most publishers — they optimize the wrong link.

A perfect title (Moonstone) will never save empty content (Content). And excellent content (Content) won't break through with an off-topic image (Moonstone). Layer 1 rewards the coherence of all three, not the excellence of one.

The resulting method is simple: diagnose in order. Laughable impression volume despite indexing → suspect Content. Good content but generic thumbnail and weak CTR → suspect Moonstone. Everything looks fine but zero Discover presence → suspect Aura.

Action plan: clearing Layer 1

Here's the concrete sequence to maximize your odds across all three systems, in the order they judge you.

1. For Content — before writing, ask: "what does this article add that doesn't already exist?" If the answer is fuzzy, the article will fail. Add an angle, a data point, clean analysis.
2. For Moonstone — pick a hero image that actually illustrates the title, high-resolution, no overlaid text. Test the alignment with the Image Validator.
3. For Aura — check your markup and E-E-A-T signals with the Quick Audit (25+ checks in one minute), and make sure your site is a recognized entity via the Profiler.

Once Layer 1 is cleared, your article enters the personalization and ranking phases — but without this foundation, none of what follows happens. It's the highest-yield investment on Discover, especially in French where it commands 55% of volume.

Conclusion

Content, Moonstone and Aura aren't marketing concepts: they're the three successive doors every article must clear before existing in Discover. In French, they alone decide more than half the feed. The good news is that none of the three depends on your budget or your size: they reward real editorial effort, multimodal coherence and trust — three levers entirely under your control. Start by diagnosing which one is blocking you, then address it. The rest of the Discover chain is waiting on the other side.

Frequently asked questions