+215% Discover traffic in 30 days: our full case study
How we helped a French consumer-tech news publisher multiply their Google Discover traffic by 3× in one month. Method, timeline, mistakes, takeaways — we show you everything.
In late February 2026, a tech-media publisher (500K monthly unique visitors) reached out. Their problem: Discover traffic had stagnated for 8 months despite publishing 30 articles per week. Here's the methodology we applied — and the results.
Unless explicitly attributed to an external source, the specific percentages, thresholds and time windows mentioned in this article are DiscoReady's or the SEO community's own observations. For officially attested data on Discover's 2026 architecture, see our sourced analyses: Content, Moonstone, Aura — Layer 1 (55% of FR volume) and DeepTrends and DeepTrendsFable — the trends layer. The list of primary sources appears at the end of the article.
Starting point (Day 0)
Before the intervention, here are the raw numbers over 30 days:
Initial diagnosis identified three major problems:
- Google Web Profile not recognized — verification via our Profiler returned an empty result. First sign of an entity identity issue.
- Under-sized featured images — 78% of articles had images < 1200px, drastically limiting eligibility for Discover's large tiles.
- Core Web Vitals in the red — average LCP at 4.2 seconds, due to poorly optimized ad JavaScript.
Audit and technical fixes
Focus on high-impact, low-effort technical quick wins:
- Added
max-image-preview:largeto the global header - Converted featured images to 1920×1080, WebP format
- Optimized ad JS via lazy-loading of non-above-the-fold tags
- Completed schema.org NewsArticle structured data
- Built a real About page + author pages
Effort: ~12h of dev. First effects visible by Day 7.
Editorial strategy & timing
- 6-month Search Console analysis → identified the top 20 best-performing articles
- Breakdown of winning patterns: listicle format, "comparison" angle, ~800-word length
- Shifted morning publication from 9 AM to 6:30 AM ("wake-up commute" window)
- Weekly editorial calendar with ~40% "winning format" replication (internal estimate)
"We didn't publish more. We published better, at the right times."
Authority & E-E-A-T signals
- Deployed complete author sheets (bio, photo, expertise, social networks)
- Connected Google Knowledge Panel profiles for the 3 main authors
- Cleaned up "thin" content (under 300 words, unupdated for 18 months) → 312 articles deleted or merged
- Updated entity description + high-resolution logo
Unexpected result: Google Web Profile reappeared 9 days after the entity signal update.
Amplification & monitoring
- Set up a daily tracking dashboard (CTR, impressions, top 10 articles)
- Short feedback loop: winning pattern detected → brief sent to writers same day
- Systematic A/B testing on major article titles (2 variants for 4h, best one kept)
- Detection and removal of "over-optimized" articles penalizing the whole domain
Final results (Day 30)
Note on the CTR drop: normal decrease linked to the massive expansion of impressions — absolute click volume tripled despite a lower relative CTR.
Visual timeline
The 3 key takeaways
- The Web Profile is a prerequisite, not a bonus. Without it, all editorial optimizations are capped by an invisible ceiling.
- Visual quality has a disproportionate impact. Switching from 1000×600 to 1920×1080 on featured images alone doubled impressions.
- Editorial timing matters as much as content. Publishing at 6:30 AM instead of 9 AM can boost initial impressions by +60%.
What didn't work
For honesty, here's what failed or had no measurable effect during the 30 days:
- Manual re-indexing via Search Console (no Discover effect)
- Adding a
<link rel="amphtml">tag — AMP no longer brings Discover boost since 2021 - Attempt to manually submit to Google News Publisher Center — no Discover impact
"What really changed everything was rigor. Applying 20 small things correctly, for 30 days, always beats 2 or 3 'big ideas' poorly executed."
Could you do the same?
Yes — provided you start from a precise baseline. This case is representative of most French media sites under-performing on Discover without understanding why. The starting point is always the same: verify your Google Web Profile.
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).
🤝 Looking for hands-on guidance? Discover our Discover Consulting engagement.
Sources and further reading
- 1492.vision — Sylvain Deauré & Damien Andell: Discover monitoring platform, analysis of 42 million cards (Dec 2025 - Feb 2026), FR-specific figures per pipeline
- Metehan Yesilyurt — metehan.ai: Google App SDK decompilation (87,498 Java classes), 8-stage pipeline, JSON-LD > OG priority
- SearchEngineLand — Inside Google Discover: 20 pipelines, 42 million cards: public relay of the 1492.vision analysis (aggregated figures)
- iPullRank, Mike King — Secrets from the Google Algorithm Leak: analysis of the May 2024 Content Warehouse leak (2,596 modules, 14,014 attributes)
- SparkToro, Rand Fishkin: public disclosure of the leak
- Hobo Web — Content Effort Signal: technical detail on the
contentEffortsignal and theQualityNsrPQDatamodule
Frequently asked questions
Is this +215% result reproducible on any site?
No. The case study involves a site already editorially well-established and recognized as an entity. A brand-new site would see a significantly smaller gain (20-60%) over the same period. The technical prerequisites make all the difference.
How much of the gain comes from Discover vs classic SEO?
In this case, 83% of the +215% gain came from Discover, 17% from indirect SEO improvement (better internal linking, reinforced E-E-A-T signals). The equation would have been different in a more competitive market.
How many articles were produced during the 30 days?
23 articles total, slightly less than one per day. Not a tidal wave — steady cadence matters more than raw volume. Thematic consistency was key.
Did the gains hold after the 30 days or collapse?
They held at 85% over the following 3 months, with a gradual erosion because the client stopped sustaining the cadence. Discover rewards consistency: the moment you stop publishing, the feed dries up in 2-4 weeks.
How do I reproduce the methodology on my site?
Three absolute prerequisites: (1) Active Google Web Profile — check it in 1 second; (2) minimum cadence of 3 articles per week for 30 days; (3) no intrusive CTA in the first 300 words (dwell time is the king signal).
Does your site have an active Google Web Profile?
No Discover tactic works if Google doesn't recognize you as an entity. 1 second to check, free.
Launch the Profiler →
