The journey of a viral article through Discover

When you publish an article, it doesn’t appear once in Discover and vanish. If it performs, it travels through multiple pipelines over the following hours and days : each opening a new visibility window, at a different timing, for a partially different audience. Here’s that journey.

The lifecycle of an article in Discover, based on median ages per pipeline:

0-2h — The breaking window

• newsstoriesheadlines (1.8h EN): Google News clusters : if your article is part of a breaking story, it enters here first

• creatorcontent (1.9h EN): social intake : if the content circulates on x.com or YouTube. This is EN’s fastest entry point alongside news : far quicker than in FR markets where social intake is slower (7.4h).

2-12h — The baseline pickup

• mustntmiss (3.4h EN): editorial importance flag : if the subject is major, ~2x priority boost. Slower than in FR (2.6h) : the EN pipeline treats editorial urgency with a slightly wider evaluation window.

• freshvideos (~8.6h): video amplification : arrives much earlier in the EN lifecycle and with 7.1% reach (vs 2.9% FR). Video is a first-class citizen in the EN ecosystem, not an afterthought.

• discover_ai_summary (~9.8h): AI-generated overview cards : EN-only pipeline, 3.5% reach. A new pipeline surfacing AI summaries of trending topics. Does not exist in the FR market.

• content (~11.6h): the general pipeline : this is where most articles first appear

12-24h — The broadcast and amplification

• deeptrendsfable (~14.2h): the trend scanner : later than FR (10.1h), but still within the first day

• moonstone (~15.7h): engagement broadcast : if the article triggers engagement signals. 9.4% reach : significantly lower than FR’s 19.3%. The EN engagement broadcast is narrower.

• neoncluster (~17.3h): YouTube video clusters : 13% reach, EN’s third-largest pipeline. Near-zero in FR. This is the biggest EN-specific pathway: YouTube content gets its own broadcast pipeline.

• relatedcontentruby (~10.9h EN): triggered by user click : a click on a related article → seed→expansion at next refresh

• paginationpanoptic (~14.9h): scroll continuation : users who scroll deep

1-3 days — Persistence

• aura (1.0 day EN vs 1.5 day FR): long-tail diversifier : the article reaches new segments via cross-user profiles. Faster in EN : diversification kicks in earlier.

• astria (1.2 days): local/lifestyle authority

• deeptrends (1.7 days): trend persistence : if the subject held, +26h after deeptrendsfable

3-7 days — The long-tail

• shoppinginspiration (2.5 days EN vs 3.7 days FR): product reviews live the longest : 13.1% reach. Faster turnover in EN, but still the longest-lived editorial pipeline.

The X-axis (median age, log) IS the freshness ladder. From left (nsh, creatorcontent = breaking) to right (shopping, feedads = persistent). The Y-axis adds reach : neoncluster and shopping are the EN broadcasts.

The lifecycle of an article in Discover: from breaking (0-2h) to long-tail (3-7 days). Each pipeline has its own time window.

It’s not just about timing. Breaking pipelines also place content higher in the feed.

newsstoriesheadlines and relatedcontentruby in positions 2-4 (top of feed). moonstone and shopping in positions 6-8 (deep). Speed AND prominence go together.

The first two hours deliver a double advantage: fast selection AND premium placement.

The ladder is the best case. Reality: a large share of EN URLs appear in only one pipeline : typically content.

How many pipelines per URL? The drop is exponential : from single-pipeline majority down to a few outliers at 14.

To climb, an article needs signals: engagement (→ moonstone), trend (→ deeptrendsfable), editorial importance (→ mustntmiss), local relevance (→ geo/wklocal). Without a signal, the article stays in content : the baseline. With a signal, it traverses the system.

Why articles move from one pipeline to another:

• Engagement-triggered: content → moonstone (engagement signals detected)

• Time-triggered: deeptrendsfable → deeptrends (+26h if the subject holds)

• Video cascade: creatorcontent → freshvideos (+7h) → neoncluster (+15h) — the EN video cascade is a major pathway, carrying 13% reach at its peak

• User-action: any article → relatedcontentruby (user click → seed→expansion at next refresh)

• Google News: breaking story → newsstoriesheadlines (Google News cluster)

• Editorial boost: major subject → mustntmiss (priority multiplier ~2x)

• AI summary: trending topic → discover_ai_summary (EN only : AI-generated overview card)

The freshness ladder exists in both markets, but the EN ecosystem has distinct characteristics:

Video is a real pathway, not a niche. The creatorcontent → freshvideos → neoncluster cascade carries massive volume in EN. Neoncluster alone has 13% reach : it’s the third-largest pipeline. In FR, it’s effectively zero. If you produce video content, the EN Discover ecosystem has a dedicated three-stage amplification path for it.

AI Overviews are live. discover_ai_summary is an EN-only pipeline with 3.5% reach and ~10h median age. It surfaces AI-generated summary cards for trending topics. This pipeline doesn’t exist in FR.

The engagement broadcast is narrower. Moonstone reaches 9.4% in EN vs 19.3% in FR. The EN market’s engagement amplification is roughly half of what FR publishers see. This may reflect a more competitive, more fragmented EN content landscape.

Diversification is faster. Aura kicks in at ~24h in EN vs ~35h in FR. EN articles reach new audience segments through cross-user profiles about 10 hours earlier.

Shopping turns over faster. Product reviews peak at 2.5 days EN vs 3.7 days FR, but with lower reach (13.1% vs 19.7%). The EN shopping pipeline is faster but shallower.

13 EN-exclusive pipelines exist : sports stacks (home_stack_sports), entertainment drops (entertainmenttrailerdrop), weather (wx, air_quality), followed creator content. These are small-volume but show that the EN ecosystem is broader and more specialized.

• Publish when the topic is rising, not when it’s peaked : deeptrendsfable detects trends early

• The first 2 hours matter most for news: mustntmiss and newsstoriesheadlines decide fast

• The first 12 hours of engagement determine moonstone pickup, but the EN moonstone bar is higher (9.4% reach vs 19.3% FR)

• Video content has its own clock: creatorcontent picks up in under 2 hours, neoncluster broadcasts at ~17h : plan your YouTube/video publishing for the EN cascade

• Product content is the exception: shoppinginspiration doesn’t care about timing : it lives 2-3 days regardless of publication date

• AI summary cards (EN only) surface at ~10h : if your topic is trending, there’s a new pipeline carrying it

Next newsletter: shopping, trends, local : the specialized pipelines that each operate by their own rules.

Full data: 1492.vision | Interactive: explorer

Data: 42 million Discover cards, Dec 2025 – Feb 2026. Analysis: 1492.vision.