Inside Google Discover: How 20 Pipelines Shape the English Feed

Same methodology, same period, English data. The EN feed reveals a structurally different product — with pipelines absent from French and radically inverted balances.

What you'll discover

The English Discover feed shares the same pipeline names as the French one. But the data tells a different story. Here are the ten key findings:

  1. neoncluster: 13% reach, 100% YouTube, EN-only — a video broadcast pipeline that doesn't exist in France
  2. The video cascade — creatorcontent → freshvideos → neoncluster, three stages, increasing reach, increasingly pure content
  3. AIO has arrived — discover_ai_summary is 99.997% AIO, mustntmiss EN contains 29% AIO content. In France: zero
  4. feedads: 58.4% reach — the highest of all pipelines, across all markets. 264 cards/device/day
  5. The EPL exclusion — Premier League systematically under-represented in 7+ pipelines. The most surprising finding
  6. shoppinginspiration: 13.1% reach — 2.5-day lifespan, but a silo
  7. moonstone EN: 9.4% — half of France's level. A structural divergence
  8. 37% of URLs in 2+ pipelines — vs 58% in France. Multi-pipeline is less common in English
  9. A living system — neoncluster 18x, paginationpanoptic 4x, feedads 4x in 3 months
  10. Position in the feed — pipelines also control placement in the feed, not just selection

Methodology

Our analysis is based on observing real Discover feeds, collected across hundreds of devices over three months (December 2025 – February 2026). In total: 42 million cards analyzed.

The methodology is identical to that of the FR reference article: for each card, we trace the pipeline responsible for its selection. The metrics — reach, speed, exclusivity, volume — are calculated the same way and normalized by observation density.

Explore the 20 EN pipelines visually: Open the interactive explorer →

Interactive pipeline explorer — EN view Screenshot of the interactive explorer — EN toggle. Live version →

The architecture: a layered system, different balances

The structure is the same as in French — six functional layers. But proportions change radically. Video is omnipresent. AIO exists. Advertising dominates.

Freshness × Reach — the EN pipeline landscape Each pipeline positioned by speed (X-axis, log) and reach (Y-axis). neoncluster occupies a unique position: high reach (13%), medium speed. feedads is the extreme outlier on the right (month-long campaigns). Breaking (nsh, mustntmiss) in the top left.

Freshness Ladder — article lifecycle in EN The lifecycle of an article in Discover EN: from breaking (nsh 2.2h, mustntmiss 3.5h) to long-tail (shopping 2.5 days, feedads 57 days). Each pipeline has its time window.

Volume by pipeline EN The 20 EN pipelines ranked by total volume (December 2025 – February 2026). content dominates at 34.2%, followed by feedads (11.1%) and aura (8.7%).

Layer 1 — The editorial foundation

Five pipelines form the base. As in FR, it's a recycling loop — content → moonstone → aura → paginationpanoptic → relatedcontentruby. But the balance shifts:

  • content — 34.2% of EN volume — the baseline. YouTube at 29.8%, Guardian 10.4%, BBC 7.9%.
  • moonstone — 7.8% of EN volume. Reach 9.4% — half of France's (19.3%). This is the most marked structural divergence between the two markets.
  • aura — 8.7%. Science, tech, finance over-represented. YouTube at 27.4% — a massive presence absent from FR.
  • paginationpanoptic — 5.5%. Growth 4x over the period.
  • relatedcontentruby — 6.7%. BBC (14.4%) and Guardian (13%) dominate.

Layer 2 — Breaking news and urgency

  • mustntmiss — 0.5% of volume but 7.3% reach. 29% AIO content — the highest rate of any pipeline. YouTube (9.3%), CNN (8.8%), NYT (8.1%), Guardian (7.3%). Growth: 2.6x.
  • newsstoriesheadlines — 0.4% of volume, high reach. Guardian (14.3%), BBC (13.1%). Google News clusters.

Layer 3 — Trends

  • deeptrendsfable — 0.5%. x.com (21.4%) as second source after BBC — a social trend signal. Same mechanism as in FR.
  • deeptrends — 0.2%. BBC (24.7%) dominates the second stage. Contracting: 0.5x over the period.

Layer 4 — Local and geo

  • geotargetingstories — 0.7%. x.com dominates at 43.2% — a unique EN fact. Social content is the top geolocated source.
  • webkicklocalstories — 0.3%. British local press: Oxford Mail, Dorset Echo, Liverpool Echo. Same hyperlocal logic as in France, but with UK regional titles.
  • astria — 1.7%. BBC dominates at 29.3%. Horse racing, astrology, Showcase.

Layer 5 — Social and video

This is where the EN feed fundamentally diverges from FR. In English, three pipelines form a video cascade with no French equivalent:

  • creatorcontent — 4.0%. YouTube dominates at 72.4% (vs x.com 73% in FR — complete inversion). Growth: 7.8x.
  • freshvideos — 4.4%. YouTube 94%. Appears +15h after creatorcontent. Near-pure video.
  • neoncluster — 4.5%. 100% YouTube. Reach 13%. Growth 18x. The video broadcast pipeline, absent from France.

The logic is sequential: creatorcontent is the intake (mixed content), freshvideos filters toward pure video (+15h), neoncluster amplifies the best into a massive broadcast (+23h). At each step, content purifies and reach increases.

Video % × Reach — the EN video spectrum Video percentage (X) vs reach (Y) for each EN pipeline. neoncluster at 100% video / 13% reach, freshvideos at 94%, creatorcontent mixed. The cascade filters and amplifies.

The Video Cascade — creatorcontent → freshvideos → neoncluster The three-stage video cascade: creatorcontent (intake, 1.9h) → freshvideos (amplifier, 8.6h) → neoncluster (broadcast, 17.3h, 13% reach). At each stage, content purifies toward pure video and reach increases.

Layer 6 — Commercial

  • shoppinginspiration — 2.9%. Reach 13.1%, lifespan 2.5 days. TechRadar (9.9%), What Hi-Fi (5.9%), Tom's Hardware (5.6%).
  • feedads — 11.1%. 58.4% reach — the highest of all pipelines, across all markets. YouTube (53.7% of ads). Campaigns of 57-day median. Growth 4.1x.

The hidden dimension: position in the feed

As in French, pipelines control placement in the feed, not just selection.

Position in the feed by pipeline — EN Median position and interquartile range of each pipeline. Breaking news and related content at positions 2–4 (top of feed). Engagement and shopping at positions 6–8 (deeper).

Premium placement (positions 2–4) goes to breaking news (newsstoriesheadlines) and related content (relatedcontentruby). Engagement content (moonstone, shopping) sits deeper. Same logic as in French — urgent pipelines capture attention at the top, browse pipelines reward scrolling.

Pipeline by pipeline — EN data

For each pipeline, an identical structured profile. Data is exclusively English — for French data, see the FR reference article.

content — the generalist baseline

  • EN reach: 8.8%
  • Volume: 34.2% of EN feed — the largest pipeline
  • Median age: ~11 hours
  • Top EN domains: YouTube (29.8%), Guardian (10.4%), BBC UK (7.9%), BBC.com (7.7%), NYT (3.7%)
  • Distinguishing signal: the common trunk, as in FR. But YouTube at 29.8% — three times the French proportion — changes the nature of the feed. In English, video is already in the baseline.

moonstone — the engagement broadcast, reduced version

  • EN reach: 9.4% — half of France's (19.3%)
  • Volume: 7.8% of EN feed
  • Top EN domains: YouTube (23%), x.com (20.4%), Guardian (9%), BBC.com (6.5%)
  • Distinguishing signal: the gap with France is the largest structural difference between the two markets. In English, moonstone competes with other high-reach pipelines — neoncluster (13%), feedads (58.4%) — that don't exist or are less powerful in FR. The hypothesis: in France, moonstone is the dominant broadcast because it has no direct competitor.

The other difference: x.com at 20.4% in moonstone EN. In France, x.com doesn't appear in moonstone's top domains. In English, viral X posts are an engagement source for the broadcast pipeline.

aura — the long-tail diversifier

  • EN reach: 4.8%
  • Volume: 8.7% of EN feed
  • Median age: ~1.5 days
  • Top EN domains: YouTube (27.4%), BBC UK (12.2%), x.com (7.8%), BBC.com (7.8%), Guardian (6.1%)
  • Distinguishing signal: science (2–2.4x), tech hardware, space over-represented. YouTube at 27.4% — a video presence absent from French aura. EN aura is a diversification pipeline with a strong video component.

paginationpanoptic — the scroll infrastructure

  • EN reach: 6.2%
  • Volume: 5.5% of EN feed
  • Top EN domains: Guardian (12.2%), BBC UK (11.8%), YouTube (10.6%), BBC.com (9.3%), NYT (6.4%)
  • Distinguishing signal: 7.8% AIO content — paginationpanoptic is the second pipeline in AIO penetration after mustntmiss. AIO isn't confined to dedicated pipelines — it infiltrates the scroll infrastructure. Growth: 4x in 3 months.

relatedcontentruby — related content

  • EN reach: 4.8%
  • Volume: 6.7% of EN feed
  • Top EN domains: BBC UK (14.4%), Guardian (13%), BBC.com (8.7%), NYT (4.5%), Telegraph (3.7%)
  • Distinguishing signal: triggered by user click, as in FR. The BBC + Guardian duo concentrates 36% of the pipeline — quality UK press dominates related content.

mustntmiss — editorial importance and the AIO gateway

  • EN reach: 7.3%
  • Volume: 0.5% of EN feed — low volume, high reach and AIO
  • Top EN domains: YouTube (9.3%), CNN (8.8%), NYT (8.1%), Guardian (7.3%), BBC.com (6.5%)
  • Distinguishing signal: 29% AIO content — the highest rate of any pipeline. The ~2x priority boost is the same as in FR. But the nature changes: in EN, mustntmiss is the gateway for AI Overview into Discover. Publishers selected for AIO (Reuters, NYT, CNBC, FT) see their content summarized and amplified. Growth: 2.6x.

This is a paradigm shift. In FR, mustntmiss rewards pure editorial importance. In EN, it does the same — and adds the AI Overview layer on top.

newsstoriesheadlines — breaking news

  • EN reach: 10.6%
  • Volume: 0.4% of EN feed
  • Top EN domains: Guardian (14.3%), BBC UK (13.1%), BBC.com (10.6%), NYT (5.1%), Independent (5%)
  • Distinguishing signal: Google News story clusters, as in FR. Quality UK and American press dominate. EN reach of 10.6% is higher than FR (8.8%) — breaking news reaches more devices in the English feed.

deeptrendsfable — the trend scanner

  • EN reach: 3.4%
  • Volume: 0.5% of EN feed
  • Top EN domains: x.com (21.4%), BBC UK (19.8%), Telegraph (10.5%), Guardian (7.2%), Express (5.8%)
  • Distinguishing signal: x.com as the top source — the pipeline detects trends circulating on the social network. British press dominates the rest. YouTube is not a trend source for deeptrendsfable, even in English.

deeptrends — trend persistence

  • EN reach: 3.2%
  • Volume: 0.2% of EN feed — contracting (0.5x)
  • Top EN domains: BBC UK (24.7%), Telegraph (13.4%), Guardian (11.6%), Express (5.9%)
  • Distinguishing signal: the second stage, as in FR. The deeptrendsfable → deeptrends pass-through rate is 27%. But EN deeptrends is declining — a possible signal that the system is evolving toward other trend persistence mechanisms.

geotargetingstories — geo-filtered mainstream, dominated by x.com

  • EN reach: 4.9%
  • Volume: 0.7% of EN feed
  • Top EN domains: x.com (43.2%), YouTube (8.5%), Guardian (4.8%), BBC UK (4%), CNN (3.5%)
  • Distinguishing signal: x.com at 43.2% — the most surprising fact about this pipeline. In EN, geolocated social content (X posts with local context) is the top source for geo-targeting. In FR, distribution is balanced among traditional publishers. The social network plays a structurally different role in the English local layer.

webkicklocalstories — pure hyperlocal

  • EN reach: 2.8%
  • Volume: 0.3% of EN feed
  • Top EN domains: EDP24 (4.9%), Oxford Mail (4.5%), Dorset Echo (4.5%), Oregon Live (4%), Liverpool Echo (3.8%)
  • Distinguishing signal: British and American local press. No national title. The same 67% exclusive URLs as in FR — a dedicated channel feeding Discover content you wouldn't find anywhere else.

astria — local authority and lifestyle

  • EN reach: 5.7%
  • Volume: 1.7% of EN feed
  • Top EN domains: BBC UK (29.3%), Guardian (15.7%), Telegraph (7.9%), BBC.com (7.1%), Independent (4.5%)
  • Distinguishing signal: BBC at 29.3% — a remarkable concentration. Horse racing, astrology, Google Showcase (38% of Mirror and Express hits come from astria/Showcase). A "day two" pipeline in EN as in FR.

creatorcontent — the social intake, YouTube version

  • EN reach: 6.7%
  • Volume: 4.0% of EN feed
  • Top EN domains: YouTube (72.4%), x.com (23.2%), BBC.com (1.1%), BBC UK (1%)
  • Distinguishing signal: the perfect inversion. In FR, x.com dominates at 75%. In EN, YouTube dominates at 72.4%. The same pipeline, two completely different social content sources. Growth: 7.8x in 3 months.

freshvideos — the video amplifier

  • EN reach: 7.1%
  • Volume: 4.4% of EN feed
  • Top EN domains: YouTube (94%), x.com (4.1%), ESPN (0.6%)
  • Distinguishing signal: 94% video, 94% YouTube. Appears +15h after creatorcontent. In EN, freshvideos is a near-pure video pipeline — unlike FR where 47% of content is articles (TF1, L'Equipe). The cascade works as designed in English: content purifies at each step.

neoncluster — the YouTube broadcast

  • EN reach: 13.0% — the highest editorial reach of any EN pipeline
  • Volume: 4.5% of EN feed
  • Top EN domains: YouTube (100%)
  • Distinguishing signal: 100% YouTube, 100% English. In France: 36 hits in 3 months — the pipeline doesn't exist. In EN, neoncluster is the third stage of the video cascade, with 13% reach — massive broadcast level. Dominant content: international news (WION, NBC), politics, news. Growth: 18x in 3 months.

Neoncluster is not an entertainment pipeline. It's a YouTube news content amplifier. News YouTube channels (WION 46% international news, NBC) are over-represented. For an EN YouTube creator focused on news/politics, neoncluster is the most powerful visibility window in Discover.

discoverviewerrelatedcontent — viewing recommendation

  • EN reach: 2.6%
  • Volume: 0.8% of EN feed
  • Distinguishing signal: triggered by a click on video content or an AIO card. Post-click exploration content. EN data not in cat_domains (pipeline appeared late). Estimated growth: 50x in EN.

userpersonascontent — user personas

  • EN reach: 2.4%
  • Volume: 0.1% of EN feed — marginal and declining (0.4x)
  • Top EN domains: BBC.com (9.6%), Guardian (9.2%), x.com (8.8%), Telegraph (6.8%), BBC UK (6.1%)
  • Distinguishing signal: hard news, science/nature. The MUR framework (Multiple User Representations) — each user has N interest profiles. Declining in both markets: the system seems to be migrating toward other personalization mechanisms.

shoppinginspiration — the product broadcast

  • EN reach: 13.1%
  • Volume: 2.9% of EN feed
  • Median age: 2.5 days — 5x longer than a news article
  • Top EN domains: TechRadar (9.9%), What Hi-Fi (5.9%), Digital Camera World (5.7%), Tom's Hardware (5.6%), Tom's Guide (4.6%)
  • Distinguishing signal: same silo as in FR — low co-occurrence with other pipelines. The difference: in EN, specialized tech review sites dominate (TechRadar, Tom's Hardware). In FR, it's generalist tech outlets (Frandroid, Les Numériques). Lifespan is shorter in EN (2.5 days vs 3.7 days FR) — product content recycles faster.

feedads — the advertising machine

  • EN reach: 58.4% — the highest of all pipelines, across all markets
  • Volume: 11.1% of EN feed
  • Median age: 57 days — campaigns lasting months
  • Top EN domains: YouTube (53.7% of ads), share.google (2%), Johnstons of Elgin (1.9%), Radisson Hotels (1.8%), Oman Air (1.4%)
  • Distinguishing signal: 99.8% exclusive URLs. Completely closed ecosystem, as in FR. But the scale changes: 58.4% EN reach vs 24% FR. The English advertising market is 2.4x more aggressive in Discover. YouTube accounts for 53.7% of ads — video advertising dominates.

Feedads is the most massive fact of the EN feed. More than half of devices see each ad. It's an ad impression channel of considerable power — and it's hermetically sealed from the editorial feed.

EN pipeline reach hierarchy Reach of each EN pipeline (% of devices touched). feedads dominates at 58.4%, neoncluster at 13%, moonstone at 9.4%.

Article lifespan EN Median article age by pipeline EN. From creatorcontent (1.9h) to feedads (month-long campaigns). Product content lives longer than news.

The three EN-specific stories

Beyond differences in proportions, the English feed contains three phenomena that don't exist (or barely exist) in French.

AIO: the English revolution

The AI Overview — the AI-generated summary at the top of certain Google results — has entered Discover. But only in English.

  • discover_ai_summary: 1.1% of EN volume, 99.997% AIO content. Reach 3.5%. Sources: Reuters (12.3%), NYT (7.5%), CNBC (7.3%), Yahoo Finance (6.3%), Guardian (5.4%). Finance, space, US sports over-represented.
  • mustntmiss EN: 29% AIO content — the highest rate of any non-dedicated pipeline
  • paginationpanoptic EN: 7.8% AIO — AIO infiltrates the scroll infrastructure
  • In France: 72 AIO hits in 3 months — effectively zero

AIO penetration by pipeline EN AIO content rate by pipeline. discover_ai_summary at 99.997%, mustntmiss at 29%, paginationpanoptic at 7.8%. AIO only exists in English.

The publishers selected for AIO form an exclusive club: Reuters, NYT, CNBC, Financial Times, Guardian. It's quality anglophone press — factual, structured, financial. Our scoring analysis suggests that AIO in mustntmiss provides an additional visibility boost.

For French publishers: AIO is not a priority today. But the signal is clear — what's in English today will likely be in French tomorrow.

The EPL exclusion: the most surprising finding

Our data reveals a unique phenomenon: Premier League-related content is systematically under-represented in 7+ EN pipelines. This is not an artifact — it's a coherent, statistically significant pattern.

Affected terms: Premier League, football, Arsenal, Liverpool, Chelsea, Manchester United, Tottenham. Each shows a negative signal in aura, deeptrendsfable, deeptrends, geotargetingstories, shoppinginspiration, astria, freshvideos — and others.

Terms not affected: NFL, NBA, Olympics, rugby, cricket, Formula 1. The exclusion is specific to the EPL.

EPL exclusion heatmap Under-representation of EPL terms in EN pipelines. Arsenal, Liverpool, Chelsea, Premier League — systematically suppressed in 7+ pipelines. Other sports (NFL, NBA, F1) are unaffected.

The most likely hypothesis: Premier League broadcasting rights and licensing constraints create editorial restrictions that propagate through the selection system. But we cannot confirm this — it's an observation, not an explanation.

For EN sports publishers: sport works in content and moonstone, but the EPL is a blind spot in diversification, trend, and local pipelines.

The video cascade: from intake to broadcast

The creatorcontent → freshvideos → neoncluster cascade is the most fascinating mechanism in the English feed. Three pipelines, one content journey:

Stage Pipeline Content Reach Timing
1. Intake creatorcontent Mixed (72% YouTube, 23% x.com) 6.7% T₀
2. Filter freshvideos Near-pure video (94% YouTube) 7.1% T₀ + 15h
3. Broadcast neoncluster 100% YouTube 13.0% T₀ + 23h

At each stage, content purifies (from mixed to 100% video) and reach increases (from 6.7% to 13%). It's a selective amplifier: the best YouTube video content is filtered then projected to 13% of devices.

Growth is explosive across all three stages: creatorcontent 7.8x, freshvideos 7.2x, neoncluster 18x in 3 months. The video cascade is the fastest-growing pipeline in Discover EN.

In France, this cascade doesn't exist — or rather, it's vestigial. Neoncluster totals 36 hits in 3 months. FR freshvideos is 47% articles (not video). FR creatorcontent is 75% x.com (not YouTube). The three conditions for the cascade — YouTube dominance, pure video content, broadcast audience — are only met in English.

The EN domain landscape

Who dominates which pipelines? Each publisher's fingerprint tells a strategy.

Pipeline DNA — top 30 EN domains Each row = a domain, each column = a pipeline family, color = percentage of hits. YouTube dominates the social/video column (49.9%). Quality publishers (Guardian, BBC, NYT) show a balanced editorial spread. x.com concentrates the social column.

Topic Heatmap — Over/under-representation by pipeline EN Topic representation by pipeline (EN). neoncluster: news/politics dominant. mustntmiss: 29% AIO, news-oriented. aura: science/tech 2–2.4x. Finance concentrated in content + discover_ai_summary.

Domain Dominance by Pipeline — Top domains per pipeline EN For each pipeline, the dominant domains. neoncluster: 100% YouTube. feedads: 53.7% YouTube ads. content: YouTube 29.8%, Guardian, BBC. shopping: TechRadar, Tom's Hardware.

YouTube — the colossus. Dominant in content (29.8%), creatorcontent (72.4%), freshvideos (94%), neoncluster (100%), feedads (53.7%). YouTube is present in 5 of the 6 functional layers. No other domain has this footprint.

Guardian — the editorial spread. Present in the top 5 of 12 pipelines. Content, aura, ruby, pagpan, nsh, astria, deeptrendsfable, deeptrends — one of the most diversified profiles. It's the EN equivalent of Ouest-France: the multi-pipeline publisher par excellence.

BBC (bbc.co.uk + bbc.com) — authority. Dominant in newsstoriesheadlines (23.7% combined), deeptrends (28.9%), astria (36.4%). Less diversified than the Guardian — more concentrated on news and authority.

x.com — the social network. Dominant in geotargetingstories (43.2%), deeptrendsfable (21.4%), creatorcontent (23.2%). In EN as in FR, x.com feeds trend and social pipelines — but in EN, YouTube takes the lead role in social.

TechRadar / Tom's Hardware — the tech silo. Dominant in shoppinginspiration (9.9% and 5.6%). Near-absent from other pipelines. The same pattern as in FR with Frandroid/Les Numériques: product reviews remain confined to the shopping silo.

EN multi-pipeline: 37% of URLs in 2+ pipelines

Multi-labeling exists in English, but it's structurally different from French.

  • 63% of URLs appear in only one pipeline
  • 19% in two pipelines
  • 9% in three
  • 9% in four or more
  • Maximum observed: 12 pipelines (vs 14 in FR)

Distribution of pipeline count per URL Number of distinct pipelines per URL (log scale). The FR tail is longer (14 pipelines) than the EN tail (12). Multi-pipeline is more common in French.

Jaccard similarity matrix between EN pipelines The content-aura and content-pagpan pairs form a hot block in EN as in FR. But neoncluster-freshvideos-creatorcontent form a video block specific to the English feed.

The EN multi-labeling rate (37%) is lower than FR (58%). The likely explanation: the video cascade (creatorcontent → freshvideos → neoncluster) is itself a multi-labeling vector — YouTube URLs traverse three pipelines by design. In FR, this mechanism doesn't exist, but press URLs traverse more editorial pipelines (content, aura, moonstone, deeptrendsfable, mustntmiss) thanks to the more "textual" structure of the feed.

A living system

The EN feed evolves rapidly — and differently from French.

Growth Dynamics × Volume Growth dynamics (Y, log) vs total volume (X, log). Above the stability line = growth. neoncluster EN at 18x, freshvideos and creatorcontent at ~8x. userpersonascontent and deeptrends declining.

Monthly Growth Heatmap — Pipelines EN Monthly dynamics by pipeline (EN). neoncluster explodes at 18x (YouTube cascade), feedads at 4.1x (advertising acceleration). The English feed is restructuring toward video, advertising, and AI.

Exploding pipelines:

  • neoncluster: 18x in 3 months — YouTube broadcast is being built in real time
  • freshvideos: 7.2x — the video amplifier is ramping up
  • creatorcontent: 7.8x — the social intake is opening
  • feedads: 4.1x — advertising is accelerating
  • paginationpanoptic: 4x — scrolling is intensifying

Declining pipelines:

  • userpersonascontent: 0.4x — the persona system seems to be retreating
  • deeptrends: 0.5x — trend persistence is contracting
  • discover_ai_summary: 0.9x — stable but not growing

The EN direction: the video cascade (3 pipelines growing simultaneously), advertising intensification (feedads 4x), and AIO (present in 3 pipelines) paint an English feed increasingly dominated by YouTube video, advertising, and artificial intelligence. A very different profile from the French feed, oriented toward text, engagement, and traditional press.

What's next?

The English and French Discover feeds share the same pipelines — but not the same product.

English has a three-stage video cascade (neoncluster at 13% reach, absent from FR), a nascent AIO ecosystem (29% of mustntmiss, zero in FR), a moonstone halved (9.4% vs 19.3%), 2.4x more aggressive advertising, and structurally different multi-labeling.

For an EN publisher, the levers aren't the same as in FR:

  • Video-first: the YouTube cascade is the most explosive route to broadcast (neoncluster 13%)
  • AIO readiness: mustntmiss EN is already at 29% AIO — AI Overview is a real competitive advantage for quality press
  • EPL awareness: sport, and EPL in particular, is a blind spot in diversification pipelines
  • Shopping isolation: the same silo as in FR — the challenge is getting out

For publisher-specific details and targeted recommendations, follow our Substack newsletter. For a structured FR vs EN comparison, see the dedicated comparison article.

The system evolves. This data is a snapshot from December 2025 to February 2026. The video cascade that didn't exist in December already represents 13% of the feed in February. At this pace, today's snapshot is tomorrow's past.

Data: 42 million Discover cards, December 2025 to February 2026. Analysis: 1492.vision. Internal mechanisms are presented as our interpretations based on observed data and available public research.

Authors

Posted on 2026-03-28

R&D articles from 1492.vision