🏪 Les internautes font des recherches commerciales 16x +sur #Google que sur ChatGPT, mm ratio #Bing / Google
📊 L'étude des prompts de 12K utilisateurs de #ChatGPT révèle que 7% sont des rech. commerciales
🛠️ QQchose qui sait ê un moteur de recherche n'en est pas un à 100%🙂
Interesting data from @randfish -> We Analyzed 118,000 ChatGPT Prompts from 12,000+ Users to Answer: How Do Americans Use AI Tools in 2026?
"ChatGPT is a heavily-consumer app, and anyone who tells you otherwise is wrong. People use it all the time for a massive variety of tasks. We can’t see the API requests, but I’d guess that makes up 50-75% of all usage, and it’s probably sending very different prompts than what consumers do via their phones and desktops."
Regarding purchase intent:
"Only 6.9% of prompts showed anything related to researching a potential brand to buy from. For comparison, when we looked at 332 million Google queries two years ago, 15%+ matched this definition. Google’s not only 5.5X bigger than ChatGPT in terms of total searches to prompts, it’s also ~2.8X bigger (probably more) in the percent of searches with purchase intent. Conservatively, I’d guess that people ask Google purchase-journey-related queries at 16X+ the rate of ChatGPT. That market share number isn’t dissimilar to the Google:Bing ratio."
Regarding the comparison to Google:
"The competition with Google has been massively overblown by marketing press and influencers. Going down the list of prompts and conversations will make you realize that this isn’t Google or Bing. It’s not a search engine. People don’t treat it that way. It’s something else: a confidant, a creator tool, an entertainment product, an assistant. Comparing how people use ChatGPT and how they use Google finds far less overlap than my LinkedIn feed suggests."
But -> "That said, 12% of ChatGPT usage is part of the buyer journey. That’s not nothing. It’s just close to the size of Bing than Google."
sparktoro.com/blog/we-analyz…
27
Olivier Perbet retweeted
Soyons clair : cette loi fera des morts. Ne serait-ce qu'en relevant le seuil psychologique de décision de tir. Y compris des morts innocents, c'est inéluctable, mais qui ne pourront pas prouver qu'ils l'étaient. Est-ce un prix acceptable à payer pour complaire aux syndicats de policiers ?
Mesdames, Messieurs du Sénat,
Vous examinez aujourd'hui un texte qui présume légal tout tir policier.
Chaque tir. Y compris celui qui tue.
Ce sera aux victimes, ou à leurs familles en deuil, de prouver que le tir ne l'était pas. La charge de la preuve s'inverse, et la présomption de culpabilité change de camp.
Le policier qui a tiré n’ira plus automatiquement en garde à vue. Parfait, pour prendre le temps d'accorder sa version avec celle de ses collègues.
Accorder les versions, ils savent faire. L'audience #Zecler, où aucun d'entre vous n'est venu, l'a montré.
Des coups, des insultes racistes, puis des procès-verbaux mensongers, écrits pour transformer la victime en coupable. Sans les images, Michel Zecler serait en prison. Ses agresseurs, en service, avec des dommages-intérêts.
La Justice aurait cru les policiers, comme toujours. Le mensonge policier n'est pas une dérive : c'est une méthode. Et l'institution judiciaire en est trop souvent le premier relais.
Voilà la police que votre texte veut présumer innocente avant même l'enquête.
Les armes de la police tueront encore. Votre loi ne les en dissuadera pas. Elle les y encouragera.
Mesdames, Messieurs les sénateurs, ne votez pas ce texte.
Olivier Perbet retweeted
Interesting data from @randfish -> We Analyzed 118,000 ChatGPT Prompts from 12,000+ Users to Answer: How Do Americans Use AI Tools in 2026?
"ChatGPT is a heavily-consumer app, and anyone who tells you otherwise is wrong. People use it all the time for a massive variety of tasks. We can’t see the API requests, but I’d guess that makes up 50-75% of all usage, and it’s probably sending very different prompts than what consumers do via their phones and desktops."
Regarding purchase intent:
"Only 6.9% of prompts showed anything related to researching a potential brand to buy from. For comparison, when we looked at 332 million Google queries two years ago, 15%+ matched this definition. Google’s not only 5.5X bigger than ChatGPT in terms of total searches to prompts, it’s also ~2.8X bigger (probably more) in the percent of searches with purchase intent. Conservatively, I’d guess that people ask Google purchase-journey-related queries at 16X+ the rate of ChatGPT. That market share number isn’t dissimilar to the Google:Bing ratio."
Regarding the comparison to Google:
"The competition with Google has been massively overblown by marketing press and influencers. Going down the list of prompts and conversations will make you realize that this isn’t Google or Bing. It’s not a search engine. People don’t treat it that way. It’s something else: a confidant, a creator tool, an entertainment product, an assistant. Comparing how people use ChatGPT and how they use Google finds far less overlap than my LinkedIn feed suggests."
But -> "That said, 12% of ChatGPT usage is part of the buyer journey. That’s not nothing. It’s just close to the size of Bing than Google."
sparktoro.com/blog/we-analyz…
Olivier Perbet retweeted
Remember it’s only a true Outage if it’s from the Outage region of France. Otherwise it’s just sparkling downtime.
Olivier Perbet retweeted
This is one big reason why the web is dying
Professional, ethical SEO, was about surfacing legitimate information from domain experts.
This is like throwing trash all over your house and wondering why you can't find anything anymore.
Please, don't. Save the web.
ALRIGHT LOOK AT THIS
And then do this…
1. Go to chatgpt
2. Submit a search that's important to your niche
3. right click
4. Open Inspect Element
5. copy the url id after /c/
6. go to network
7. paste it
8. click refresh
9. click the orange brackets here
10. go to the response tab
11. search on “queries”
12. now you can literally see what searches chatgpt did in the background
13. copy all of this
14. go to a site like rankdesk(.ai)
15. go to blog posts
16. and literally ask it to write articles for all of these queries
The AI is going to auto-write and auto-publish them in the upcoming days and weeks
Infinite SEO hack!
📹 Time-lapse x60 du vent poussant les nuages vers l'est à #Samoens dans les #Alpes
⛰️
Début de soirée d'un jour d'août 2026
☁️
@Samoens_Village – at Samoëns, France
Je confirme !
Grosse chute pour un site pur SEO, malgré la masse de visites et de conversions que ça générait.
Mes autres sites plus variés n'ont pas bougé.
On va analyser ça un peu plus en détail mais voici l'aperçu de la récente Spam Update (terminée le 21 août). 🚨
C'est sans doute l'une des pires mises à jour depuis Panda, Penguin et la HCU. 📉
Les sites les plus impactés :
• Plus d'un an de visibilité (avec VI >17 sur @haloscancom )
• Trafic reposant essentiellement sur le SEO
• Marque faible ou peu connue face aux concurrents
• Très peu de trafic issu des réseaux sociaux
• Pas ou peu de budget publicité Ads
En France, la baisse risque de s'intensifier avec l'arrivée des AI Overviews. 🇫🇷
Le problème est mécanique : baisse de trafic direct + analyse du comportement via Chrome / android = perte de positions. Moins de visites entraîne une baisse de ranking, qui entraîne encore moins de visites. 🔄
💡 La leçon à en tirer : investissez dans la notoriété de votre marque, développez vos réseaux sociaux (notamment YouTube), visibilité sur Google map, Ads et ne comptez plus uniquement sur SEO pur.
On analyse cela en détail il y a aussi une autre corrélation liés aux indexations mais c'est plus long à analyser car il faut crawler le site + backlinks.
Bref, comme d'habitude, ce sont les gros acteurs qui raflent tout : meilleure visibilité en SEO et sur les IA. Le terrain devient de plus en plus difficile pour les petits entrepreneurs. 🛑
Bonne rentrée à tous ! 💼
1
73
Olivier Perbet retweeted
Yep, @thetafferboy is awesome. His case study is awesome. You must read this today. And then set up cats.txt to make sure your GEO strategy is complete. 😂😹 -> How cats.txt showed llms.txt evidence is GEO astrology
"Google indexes text files. It has done so, enthusiastically, since before most of the people currently selling llms.txt owned a smartphone. Being in the index is a statement that a URL exists and contains words. It is not a verdict on truth, usefulness or sanity."
The case study: markwilliamscook.substack.co…
🫠 Il va encore faire +30° à #Paris cette semaine.
🍲On a pas toujours la climatisation chez soi, mais on peut aller manger au frais !
❄️ Voilà une liste de restaurants climatisés sur Paris, dans les 2e, 6e, 9e, 10e, 11e et 12e arrondissements
👇
peroliv.blogspot.com/2019/07…
1
32
Dites les #SEO,
J'ai cette personne qui m'écrit depuis plusieurs années, tjrs avec le même sujet de mail
"RE: Réintroduction de la publication"
pour les conditions d'achat de lien de mon site.
Je lui ai répondu 20 fois, donc 1 où elle a validé.
Le mail générique revient tjrs
72
Olivier Perbet retweeted
Oof, that's bad for business. Also, prompting AIOs isn't why this changed. I'm sure Google saw the press and removed the AIO -> Owner of popular Millbank, Ont., bakery ‘panicked’ after Google's AI overview mistakenly said it was closing
"Over the weekend, someone Googled Anna Mae’s Bakery and the AI overview, or summary, popped up. It said the Millbank business, just northwest of Kitchener, was officially closing on July 31. Anna Mae's owner Amanda Herrfort found out when she was on her way to a wedding and someone sent her a screenshot of the search result. "That’s when I panicked," Herrfort told CBC News on Thursday."
cbc.ca/news/canada/kitchener…
🔎 Aberration majeure de #Google sur les AI Overviews : pour des recherches locales, les sources sont TOUT sauf Google Maps.
Qui appartient à Google et que les commerces mettent à jour en priorité (voire uniquement).
On est exposé à des infos potentiellement fausses.
🫠
4
3
1,196
Olivier Perbet retweeted
Ecommerce SEO: the more I work with eCommerce stores the more I realize how different robots.txt strategies can be.
My client received a notification for an error in Search Console and they were recommended to exclude this file path from being crawled from the team (which seemed like reasonable advice to me).
Hours later: product pages start to be removed from free listing eligibility at scale, with hundreds of products now no longer able to appear within organic shopping results... not good.
This was surprising to me, because the notice in Merchant Center said that they were "unable to check product pages", but there was nothing related to the file that was being blocked – it was just a seemingly unrelated file path.
Turns out the path was actually related to theme utility file as an internal endpoint for the Storefront Renderer. So while the product pages themselves weren't being blocked by robots.txt (in a traditional sense), but the file was necessary to be crawled for product rendering.
Thankfully, we were able to get on top of this (self-inflicted) issue quickly by removing the newly added lines to the file, but it just goes to show how different the system requirements for Googlebot and StoreBot can be.
Also, the notifications that Merchant Center sends for store issues need a ton of work! There are barely any details that are provided in the notifications (compared to Search Console), so this could have easily been missed for some time if I wasn't in there every day reviewing my client accounts.
The notice within Merchant Center also directs you to use the URL Inspection Tool in order to diagnose the issue, which was not of use at all in this instance. The only reason we knew it was related to the newly added line was because the exclusion appeared almost immediately after the change was made.




























