Seg

Mon premier helfie avec IBOU ! Pas besoin de filtres, on est des oiseaux rares hibougeniques 😬 #seosummit #ibouexplorer

I think I’ve found a real champion here 🤪

@bison_seo Pas encore eu le temps de me pencher sur 5.3 et 5.4… mais a priori ça leur coûte plus cher 😅 le premier turn est générique, ensuite ils creusent par NDD. je suis pas convaincu de la pertinence du process…

Qu'est-ce que tu en penses ? @5eg

Google is quietly building publisher profile pages in Discover. We've been tracking the rollout across a sample of 47,000 publishers in 7 languages. Today we're opening our Profile Monitor to all 1492.vision users — free accounts included. Here's what we're seeing🧵

Il va me coûter cher en tokens celui là 😅

Très fiers et honorés d'avoir été choisi par le jury du #smx comme prix de l'innovation pour notre outil, merci a tous ! linkedin.com/posts/smxparis…

On popular demand, you can now set the target URL and its queries to see which parts will represent it in Google's AI search: gs.dejan.ai

@techskunkworks Tu en as trop dit là 😬

@5eg Y a un ÉNORME hack au niveau du repost. Tu peux faire des scores de dingues rien qu'avec du repost bien pensé.

We reverse engineered TikTok's algorithm. 4 ranking engines decoded: Feed, Search, Ads, Commerce. Scoring formulas, ML model names, production thresholds. Full breakdown: linkedin.com/feed/update/ur…

@eldoranext Yes, TikTok tente de prédire notamment si tu vas commenter, mais aussi le wach time probable, et de nombreuses autres metrics. Mais ensuite il y a un gros reranking directement opéré par sur le téléphone, et celui la joue bcp

Super interessant ! Si j’ai tout compris c’est déjà joué à 90% lors de la publication par les predict ?

Ils remettent ça avec TikTok... @resoneo publie une étude super poussée sur le fonctionnement du célèbre réseau et j'en ai appris une tripotée sur le fonctionnement de leur moteur de recherche interne. C'est ici : think.resoneo.com/tiktok/ #TikTok #SEO

@scharbitjrs @resoneo Merci Sylvain pour la mention ! ☺️

L’infographie complète (EN & FR) : think.resoneo.com/tiktok/

damn, @methode going berserk and pushing non stop on Google's official robots.txt repo 💪

🛒 "Discover c'est pour les gros médias" Faux. Notre data montre des fiches produit de tricot, de plantes, de Hifi niche qui poppent régulièrement. 63 likes FB suffisent. Sérieusement. On a tout détaillé 👇 1492vision.substack.com/p/discover-nes…

We talked about it during a webinar in French with @Cariboo_seo : I wrote an article on the subject on Linkedin, I will cover certain concepts explained in an article in the days to come, don't miss it ⬇️Link in the comments 🧐It may be more useful to think of Google Discover as a recommendation pipeline made of multiple systems, not just one algorithm. What we actually see looks much more like a multi-engine recommendation pipeline: Related (interaction-driven) Followed creator / followed web creator High affinity (web creator content) Item-Item Collaborative Filtering Clusters & personas Entertainment trailer drop (event-driven) Tailor your feed (prompt / LLM-driven) In my latest LinkedIn article, I break down these recommendation layers and how they appear to coexist inside Discover. One key point I cover: ➡️ “High affinity” does not seem to mean strong affinity for a topic It appears to be a strong affinity for a web creator / publisher (domain), inferred from implicit signals such as clicks, dwell time, repeat engagement, feedback, etc. — even without an explicit follow. I also explore how Discover seems to be evolving with: stronger collaborative filtering (item-item) clusters/personas for profile orchestration and LLM/prompt-based tuning via “Tailor your feed” If you work in GOogle Discover, recommendation systems, feed personalization, or content strategy, this may be useful. CC @gaganghotra_ @lilyraynyc @rustybrick @VorticonCmdr @ClaraSoteras #googlediscover

Google Lens Deep Dive - V2 : think.resoneo.com/google-lens/ cc @metehan777 ;)

L’Ads Advisor de Google promet des analyses et recos SEA instantanées. Mais pouvez-vous vraiment tout lui déléguer ?🤔 Découvrez les retours de Willy Monniez, Directeur Associé Acquisition dans le @JDNebusiness . Merci @bruno_poncet pour cet article : journaldunet.com/adtech/1548125…

@onlinestratfr Ah mais oui merci pour la correction !

@5eg de la Moldavie

Avec tout ces autolinks en *.md qui débarquent, le ccTld de la Macédoine va reprendre du poil de la bête 😅 Bon ça vaut pas les .es de l’écriture inclusive ^^

@metehan777 wow!

Everyone talks about Google Discover like it's a black box. Post great content. Use big images. Hope for the best. That's the advice. That's been the advice for years. I decided to take a different approach. Instead of guessing, I went into Google's own SDK and read what it actually exposes during normal operation. Event constants. Telemetry counters. Configuration values. Things that are sitting right there if you know where to look. What came out was a picture of a 9-stage content pipeline, and it changed how I think about Discover entirely. Knowledge Graph entities? I need to look carefully! Here is the part that surprised me most: the collection-level filter runs before interest matching. Before the pCTR model. Before any ranking happens at all. If a publisher is blocked at that stage, Google never even evaluates whether their content matches what the user wants. It is binary. One article generates enough negative feedback, and the entire domain gets suppressed. There is no equivalent blanket boost. The penalty surface is wider than the reward surface. Some other things the telemetry confirms: og:title is not just a display label, it feeds directly into the predicted click-through rate model. The first 7 days carry the highest freshness weight. After 30 days, continuous staleness decay begins, tracked in hours. And dismissed content never comes back, tombstones are permanent records stored on-device. There are also 276 event constants, 150 concurrent A/B experiments running on a single session, and a system called NAIADES that most publishers have never heard of. The full breakdown, with every finding traced to a specific string or event, is on my blog. I also created a dashboard with all specifications with a FREE pCTR tool. Link is in the first thread!