J'ai retrouvé la source originale dans le cache de Google : Google Uses The Meta Keyword Tag, But Not for What You'd Think.
Je me permets de recopier le contenu du message de Jill Whalen (Fondatrice de High Rankings, un des premières entreprises spécialisées SEO aux US, si j'en crois leur présentation) :
Citation
Some months back I realized that the internal search engine that works in conjunction with my content management system (on the main pages of HighRankings.com not the forum) used the Meta keyword tag to show the most relevant pages. Oops, I thought, I guess I probably should have been using the keyword tag on my posts as it would likely improve my site visitors' chances of finding what they were looking for.
So for many of the more recent posts I've made in the newsletter section, if there were words that weren't really being used within the content itself, but were relevant to the article, I'd stick them in the Meta keyword tag (when I remembered).
Here's where the story gets interesting.
The other day I got a Google Alert for High Rankings which looked odd. It was a search result page from my site.
My first thought was...wth? I thought I blocked all search results from appearing in Google.
I don't know if you've noticed, but for the past few years Google puts some pretty weird stuff into search boxes of websites and indexes the resulting pages. And by weird, I mean they'll put single words such as "the" or "bed" or whatever. It can lead to some really silly pages being indexed, which I'm not typically a fan of.
The other problem with search results pages being indexed is that for most sites, their search result template has the same Title tag of something like "Search Results" which when constantly indexed with different words leads to there being thousands of pages of nearly duplicate content. I typically recommend to my clients that they block all search results pages from being indexed, for this very reason.
Turns out that for the High Rankings site, I had in fact blocked the main search results pages, but I had missed the fact that we had a separate search function for forum pages (still within the main pages of the site, not the search function for the forum pages themselves). So that mystery was solved, and it wasn't a case of Google ignoring my robots.txt file.
But when I looked at the weird search result page they had indexed, I saw that the search result was for this search query:
"conversions, rewriting content, 301-redirects, website development, content management"
Huh?
I thought about it a bit as those words somehow seemed familiar to me.
AHA! I remembered that those were the very keywords I had put into the Meta keyword tag of my last posted newsletter article: 3 SEO Traps to Avoid During Your Redesign.
So what does this mean?
For one thing, it helps solve the mystery of where Google gets some of the weird stuff they stick in website's search boxes. It's certainly not the only thing they use, but they obviously use it as one way of making searches. I'm kind of surprised they put the whole string of keywords into the search box though. Why not use the comma separation and put the individual phrases in instead? Seems like that would be more useful. It's possible that they do that as well, of course.
The other interesting thing is that this means that the Meta keyword tag is not completely ignored by Google as most of us have believed. While it's still most definitely ignored in terms of it having any effect on whether your site will show up in a Google search, they do appear to extract the information for other reasons such as this.
Now I have to decide if I should block those forum search pages from being indexed or maybe let it ride so I can see what other interesting queries Google uses and try to figure out where they're getting them from.
Sa version me semble un peu différente de celle postée sur le JDN :
En gros, pour faire des tests, Google tente de valider des formulaires de recherche pour accéder à des pages potentiellement masquées (Web profond, ou web invisible) qui ne seraient accessibles que depuis une requête dans un moteur interne.
Son expérience démontre « seulement » que Google se sert du contenu de ces meta keywords pour remplir ces formulaires (permettant également de vérifier la pertinence de ces keywords), pas pour classer les pages (Ce que prétend pourtant le JDN).
Donc en substance, Google n'utilise pas la meta keywords pour positionner les pages (Et aucun moteur actuel ne le fait : selon ces sites (VO ou VF), seuls Altavista, Inktomi, Lycos et d'autres le faisaient, puis ont progressivement arrêté lorsqu'ils se sont rendus compte que la balise était trop facilement spammable. De tous les moteurs cités, seul Altavista subsiste. Peut-être peut-on lier ces diverses chutes en partie au manque de pertinence induit par la pondération des mots-clés dans les résultats.
Des trois principaux mastodontes, seul Yahoo l'a véritablement utilisée, jusqu'à 2009. Après cette date, sa pondération semble avoir été grandement diminuée, même s'il est toujours possible de positionner une page avec un mot-clé inexistant au préalable (Sorry Yahoo, you DO index the meta keywords tag).
De manière générale, un moteur ne doit pas baser ses résultats de recherche sur des contenus non visibles par l'utilisateur, qui sont beaucoup trop faciles à spammer (meta keywords, meta description …) au risque de se tirer une balle dans le pied. C'est également pour ça qu'ils font tous la chasse aux textes cachés, aux liens invisibles etc. De plus, les algorithmes utilisés maintenant sont parfaitement capables d'extraire les mots-clés importants du contenu du texte.
Par contre, je serai Google (ou tout autre), je me servirai quand même du contenu de la meta keywords pour détecter du spam : si ceux-ci ne correspondent pas au thème de la page, un flag « spam potentiel » pourrait déclencher une vérification manuelle du site pour déterminer s'il y a lieu de lui apposer une pénalité.
Qu'en pensez-vous ?




Haut















