Next page visit(s)

Technical stuff on SEO @Rodolphe

06 Apr 2009 01:36 am

Google's Search Based Keyword Tool: Monetize The Long Tail of Search

CurveEvery once in a long while you come across a tool that just gives you goose bumps, you are instantly infatuated.

The Search Based Keyword Tool (SbKT) was that for me. The data it brings together and the transparency it brings is just so…. sexy.

Let's see if you feel that by the time you are done with this post. I can tell you know that you'll never think of Paid Search the same way!

One of the most common phenomenon is the long tail of search. Yet precious few people understand this and even fewer actually are able to use it to their advantage.



Les données de compétition d’Adwords Keyword Tools sont-elles utilisables en SEO?

Publié le 20 mai 2009 par Jean

adwordsAdwords keyword tool est probablement l’outil de recherche de mots clés le plus populaire pour le référencement naturel et payant. C’est un outil puissant montrant des données relativement fiables pour estimer le volume de recherches effectuées chaque mois avec un mot clé. Cet outil donne aussi le niveau de compétition que se livrent les annonceurs du réseau Adwords pour ces mêmes mots clés. Cette dernière donnée peut-elle être utilisée pour estimer la compétition pour un mot clé dans un contexte de référencement naturel ? Je débattais de la question récemment avec mes collègues et j’ai décidé d’en avoir le coeur net.

La compétition pour un mot clé est une donnée très importante qui permet de savoir à quel point il sera difficile d’obtenir une position intéressante dans les moteurs de recherche. Viser un mot clé très populaire peut sembler attractif, mais il faut s’assurer que le site ait ce qu’il faut pour se positionner dans la première page des résultats de recherche pour espérer obtenir du trafic. La démarche d’Adviso est donc de balancer le volume de recherche et la compétition des mots clés en fonction du PageRank du site. Ainsi, seuls des sites à très haut PageRank peuvent espérer se positionner sur des mots clés très compétitifs comme « iPod ». Il serait préférable pour un site à plus faible PageRank de viser des mots clés moins compétitifs tels que « étui rose pour iPod vendu au Canada ».

Plusieurs facteurs expliquent la compétition pour un mot clé. Le premier est simplement sa popularité « naturelle ». Un mot comme « voiture » est très général et il est donc très probable qu’il se retrouve dans des milliers de site internet sans que les auteurs des sites aient vraiment eu l’intention de se positionner sur le mot clé en utilisant le SEO. Ce type de compétition peut être estimé simplement en faisant une recherche sur Google, le nombre de pages possédant le mot clé est inscrit à la mention « Résulats 1-10 sur un total de <nombre> pour <mot clé>. La popularité/compétition naturelle va décroître en fonction du nombre de mots clés présents dans la phrase de recherche (« internet » vs « consultant en marketing internet »).

resultats_21

Le deuxième facteur expliquant la compétitivité d’un mot clé se résume souvent à des facteurs monétaires : plus il y a d’argent à faire avec un mot clé, plus de sites seront optimisés pour ce dernier et plus il sera difficile pour un nouvel entrant de se positionner correctement. Ce deuxième type de compétition est aussi facile à déterminer en utilisant la commande allintitle dans Google. Cette commande permet de savoir combien de sites possèdent le mot clé dans la balise TITLE, l’élément le plus important en SEO. Il est certain que l’ensemble des sites possédant un mot clé dans leur balise TITLE n’ont pas nécessairement une intention SEO, mais c’est un bon point de départ pour estimer le nombre de sites qui ont fait des efforts minimums. C’est la donnée de compétition pour les mots clés la plus fiable à mon avis, et de celle d’une bonne partie de la communauté SEO. Il existe beaucoup d’autres outils permettant d’estimer la compétition pour un mot clé, mais ceux-ci sont généralement basés sur les deux facteurs cités ci-dessus.

allintitle_2
Ayant un bon bagage scientifique et sortant d’un cours de stats au MBA, j’ai décidé d’utiliser une approche statistique pour prouver mon point : une régression multivariée. Certain vont trouver que c’est un peu comme tuer une mouche avec un 12, mais cette approche permet d’éliminer les ambiguïtés et les divergences d’interprétation puisque les critères d’analyse sont généralement reconnus. Pour les non-matheux, cette méthode permet de savoir s’il y a une association entre un facteur et un autre (F-value), la force/confiance de cette association (P-value) et à quel point un facteur explique l’autre (R2). Mon échantillon consiste en 30 mots clés provenant de 3 secteurs ayant des volumes de recherche et une compétitivité très différente : assurance, moniteurs pour bébé et batteries pour cellulaire. Les facteurs explicatifs (X) utilisés sont : 1) La donnée d’ « advertiser competition » provenant de Adwords (sujet principal de l’étude) et  2) le volume de recherche donné par Adwords (pour déterminer si la compétition pour un mot clé serait directement fonction de la demande). Ces variables ont été opposées premièrement contre le nombre de sites compétiteurs déterminés par la commande « allintitle » et deuxièmement contre le nombre total de sites compétiteurs, pour des fins de vérification.

Les résultats sont les suivants : il n’y a pas d’association entre les données de « advertiser competiton » d’ Adwords et la compétitivité pour un mot clé d’un point de vue SEO (p=0.67). L’utilisation des données d’ Adwords pour déterminer la compétition pour un mot clé est donc à proscrire. C’est dommage, parce qu’il serait très simple de pouvoir utiliser les données d’Adwords plutôt qu’utiliser la commande allintitle pour valider des mots clés un par un. Une très forte association a cependant été détectée entre le volume de recherche et la compétition pour un mot clé (p<0.001), ce qui suggère simplement que plus un mot clé est en demande, plus il y aura des sites compétitionnant pour le trafic organique. Rien de vraiment surprenant ici. Le coefficient de détermination (R2=0.95) très élevé obtenu après une régression univariée montre aussi que 95% de la compétition est expliquée par le volume de recherche, ce qui fait que seulement 5% des mots clés peuvent être considérés comme « de niche » (haut volume de recherche/faible compétition). Fait intéressant, le comportement des annonceurs est différent en PPC comparativement au SEO puisqu’il y a peu d’association entre le volume de recherches  et l’« advertiser competition », ce qui suggère que les annonceurs d’Adwords sont davantage intéressés par leur retour sur investissement par clic plutôt qu’obtenir un simple volume de visiteurs.

Une des limites importantes de cette étude est qu’elle utilise un critère d’évaluation général pour développer une méthode pour trouver les aberrations statistiques que représentent par définition les mots clés de niche. Tout en gardant ce fait en perspective, je considère qu’il vaut mieux utiliser la commande allintitle pour mesurer efficacement la compétition des mots clés, même si cela représente davantage de travail. La validité et la qualité des données n’en seront que meilleures.

  • Twitter
  • Facebook
  • del.icio.us
  • LinkedIn
  • StumbleUpon
  • Digg

Tags: ,

Publié le 20 mai 2009 à 17:56 sous Référencement. Vous pouvez laisser une réponse, ou créer un rétrolien de votre site.

The SEO Playbook - Welcome to the Rabbit Hole Alice

Filed under: Search Engine Optimization by Stuntdubl SEO at 9:40 pm, 4/18/2007

If you're new here, you may want to learn more about me, introduce yourself or check out my best posts. Thanks for visiting!

Friday, March 12, 2010 at 7:50 AM

Webmaster Level: Intermediate

Did you know that a majority of users surveyed feel that having information in their own language was more important than a low price? Living in a non-English-speaking country, I've seen friends and family members explicitly look for and use local and localized websites—properly localized sites definitely have an advantage with users. Google works hard to show users the best possible search results. Many times those are going to be pages that are localized, for the user's location and/or in the user's language.

If you're planning to take the time to create and maintain a localized version of your website, making it easy to recognize and find is a logical part of that process. In this blog post series, we'll take a look at what is involved with multi-regional and multi-lingual websites from a search engine point of view. A multi-regional website is one that explicitly targets users in various regions (generally different countries); we call it multilingual when it is available in multiple languages, and sometimes, the website targets both multiple regions and is in multiple languages. Let's start with some general preparations and then look at websites that target multiple regions.

Preparing for global websites

Expanding a website to cover multiple regions and/or languages can be challenging. By creating multiple versions of your website, any issues with the base version will be multiplied; make sure that you have everything working properly before you start. Given that this generally means you'll suddenly be working with a multiplied number of URLs, don't forget that you'll need appropriate infrastructure to support the website.

Planning multi-regional websites

When planning sites for multiple regions (usually countries), don't forget to research legal or administrative requirements that might come into play first. These requirements may determine how you proceed, for instance whether or not you would be eligible to use a country-specific domain name.

All websites start with domain names; when it comes to domain names, Google differentiates between two types of domain names:
  • ccTLDs (country-code top level domain names): These are tied to a specific country (for example .de for Germany, .cn for China). Users and search engines use this as a strong sign that your website is explicitly for a certain country.
  • gTLDs (generic top level domain names): These are not tied to a specific country. Examples of gTLds are .com, .net, .org, .museum. Google sees regional top level domain names such as .eu and .asia as gTLDs, since they cannot be tied to a specific country. We also treat some vanity ccTLDs (such as .tv, .me, etc.) as gTLDs as we've found that users and webmasters frequently see these as being more generic than country-targeted (we don't have a complete list of such vanity ccTLDs that we treat as gTLDs as it may change over time). You can set geotargeting for websites with gTLDs using the Webmaster Tools Geographic Target setting.

Geotargeting factors

Google generally uses the following elements to determine the geotargeting of a website (or a part of a website):
  1. Use of a ccTLD is generally a strong signal for users since it explicitly specifies a single country in an unmistakable way.
    or
    Webmaster Tools' manual geotargeting for gTLDs (this can be on a domain, subdomain or subdirectory level); more information on this can be found in our blog post and in the Help Center. With region tags from geotargeting being shown in search results, this method is also very clear to users. Please keep in mind that it generally does not make sense to set a geographic target if the same pages on your site target more than a single country (say, all German-speaking countries) — just write in that language and do not use the geotargeting setting (more on writing in other languages will follow soon!).
  2. Server location (through the IP address of the server) is frequently near your users. However, some websites use distributed content delivery networks (CDNs) or are hosted in a country with better webserver infrastructure, so we try not to rely on the server location alone.
  3. Other signals can give us hints. This could be from local addresses & phone numbers on the pages, use of local language and currency, links from other local sites, and/or the use of Google's Local Business Center (where available).

Note that we do not use locational meta tags (like "geo.position" or "distribution") or HTML attributes for geotargeting. While these may be useful in other regards, we've found that they are generally not reliable enough to use for geotargeting.

URL structures

The first three elements used for geotargeting are strongly tied to the server and to the URLs used. It's difficult to determine geotargeting on a page by page basis, so it makes sense to consider using a URL structure that makes it easy to segment parts of the website for geotargeting. Here are some of the possible URL structures with pros and cons with regards to geotargeting:

ccTLDs
eg: example.de, example.fr
Subdomains with gTLDs
eg: de.site.com, fr.site.com, etc.
Subdirectories with gTLDs
eg: site.com/de/, site.com/fr/, etc.
URL parameters
eg: site.com?loc=de, ?country=france, etc.
pros (+)
- clear geotargeting
- server location is irrelevant
- easy separation of sites
- legal requirements (sometimes)
pros (+)
- easy to set up
- can use Webmaster Tools geotargeting
- allows different server locations
- easy separation of sites
pros (+)
- easy to set up
- can use Webmaster Tools geotargeting
- low maintenance (same host)
pros (+)
(not recommended)
cons (-)
- expensive (+ availability)
- more infrastructure
- ccTLD requirements (sometimes)
cons (-)
- users might not recognize geotargeting from the URL alone (is "de" the language or country?)
cons (-)
- users might not recognize geotargeting from the URL alone
- single server location
- separation of sites harder
cons (-)
- segmentation based on the URL is difficult
- users might not recognize geotargeting from the URL alone
- geotargeting in Webmaster Tools is not possible

As you can see, geotargeting is not an exact science (even sites using country-code top level domain names can be global in nature), so it's important that you plan for the users from the "wrong" location. One way to do this could be to show links on all pages for users to select their region and language of choice. We'll look at some other possible solutions further on in this blog post series.

Websites that provide content for different regions and in different languages sometimes create content that is the same or similar but available on different URLs. This is generally not a problem as long as the content is for different users in different countries. While we strongly recommend that you provide unique content for each different group of users, we understand that this may not always be possible for all pages and variations from the start. There is generally no need to "hide" the duplicates by disallowing crawling in a or by using a . However, if you're providing the same content to the same users on different URLs (for instance, if both "example.de/" and "example.com/de/" show German language content for users in Germany), it would make sense to choose a preferred version and to (or use the ) appropriately.

Do you already have a website that targets multiple regions or do you have questions about the process of planning one? Come to the Help Forum and . In following posts, we'll take a look at multi-lingual websites and then look at some special situations that can arise with global websites. !

Friday, March 12, 2010 at 7:50 AM

Webmaster Level: Intermediate

Did you know that a majority of users surveyed feel that having information in their own language was more important than a low price? Living in a non-English-speaking country, I've seen friends and family members explicitly look for and use local and localized websites—properly localized sites definitely have an advantage with users. Google works hard to show users the best possible search results. Many times those are going to be pages that are localized, for the user's location and/or in the user's language.

If you're planning to take the time to create and maintain a localized version of your website, making it easy to recognize and find is a logical part of that process. In this blog post series, we'll take a look at what is involved with multi-regional and multi-lingual websites from a search engine point of view. A multi-regional website is one that explicitly targets users in various regions (generally different countries); we call it multilingual when it is available in multiple languages, and sometimes, the website targets both multiple regions and is in multiple languages. Let's start with some general preparations and then look at websites that target multiple regions.

Preparing for global websites

Expanding a website to cover multiple regions and/or languages can be challenging. By creating multiple versions of your website, any issues with the base version will be multiplied; make sure that you have everything working properly before you start. Given that this generally means you'll suddenly be working with a multiplied number of URLs, don't forget that you'll need appropriate infrastructure to support the website.

Planning multi-regional websites

When planning sites for multiple regions (usually countries), don't forget to research legal or administrative requirements that might come into play first. These requirements may determine how you proceed, for instance whether or not you would be eligible to use a country-specific domain name.

All websites start with domain names; when it comes to domain names, Google differentiates between two types of domain names:
  • ccTLDs (country-code top level domain names): These are tied to a specific country (for example .de for Germany, .cn for China). Users and search engines use this as a strong sign that your website is explicitly for a certain country.
  • gTLDs (generic top level domain names): These are not tied to a specific country. Examples of gTLds are .com, .net, .org, .museum. Google sees regional top level domain names such as .eu and .asia as gTLDs, since they cannot be tied to a specific country. We also treat some vanity ccTLDs (such as .tv, .me, etc.) as gTLDs as we've found that users and webmasters frequently see these as being more generic than country-targeted (we don't have a complete list of such vanity ccTLDs that we treat as gTLDs as it may change over time). You can set geotargeting for websites with gTLDs using the Webmaster Tools Geographic Target setting.

Geotargeting factors

Google generally uses the following elements to determine the geotargeting of a website (or a part of a website):
  1. Use of a ccTLD is generally a strong signal for users since it explicitly specifies a single country in an unmistakable way.
    or
    Webmaster Tools' manual geotargeting for gTLDs (this can be on a domain, subdomain or subdirectory level); more information on this can be found in our blog post and in the Help Center. With region tags from geotargeting being shown in search results, this method is also very clear to users. Please keep in mind that it generally does not make sense to set a geographic target if the same pages on your site target more than a single country (say, all German-speaking countries) — just write in that language and do not use the geotargeting setting (more on writing in other languages will follow soon!).
  2. Server location (through the IP address of the server) is frequently near your users. However, some websites use distributed content delivery networks (CDNs) or are hosted in a country with better webserver infrastructure, so we try not to rely on the server location alone.
  3. Other signals can give us hints. This could be from local addresses & phone numbers on the pages, use of local language and currency, links from other local sites, and/or the use of Google's Local Business Center (where available).

Note that we do not use locational meta tags (like "geo.position" or "distribution") or HTML attributes for geotargeting. While these may be useful in other regards, we've found that they are generally not reliable enough to use for geotargeting.

URL structures

The first three elements used for geotargeting are strongly tied to the server and to the URLs used. It's difficult to determine geotargeting on a page by page basis, so it makes sense to consider using a URL structure that makes it easy to segment parts of the website for geotargeting. Here are some of the possible URL structures with pros and cons with regards to geotargeting:

ccTLDs
eg: example.de, example.fr
Subdomains with gTLDs
eg: de.site.com, fr.site.com, etc.
Subdirectories with gTLDs
eg: site.com/de/, site.com/fr/, etc.
URL parameters
eg: site.com?loc=de, ?country=france, etc.
pros (+)
- clear geotargeting
- server location is irrelevant
- easy separation of sites
- legal requirements (sometimes)
pros (+)
- easy to set up
- can use Webmaster Tools geotargeting
- allows different server locations
- easy separation of sites
pros (+)
- easy to set up
- can use Webmaster Tools geotargeting
- low maintenance (same host)
pros (+)
(not recommended)
cons (-)
- expensive (+ availability)
- more infrastructure
- ccTLD requirements (sometimes)
cons (-)
- users might not recognize geotargeting from the URL alone (is "de" the language or country?)
cons (-)
- users might not recognize geotargeting from the URL alone
- single server location
- separation of sites harder
cons (-)
- segmentation based on the URL is difficult
- users might not recognize geotargeting from the URL alone
- geotargeting in Webmaster Tools is not possible

As you can see, geotargeting is not an exact science (even sites using country-code top level domain names can be global in nature), so it's important that you plan for the users from the "wrong" location. One way to do this could be to show links on all pages for users to select their region and language of choice. We'll look at some other possible solutions further on in this blog post series.

Dealing with duplicate content on global websites

Websites that provide content for different regions and in different languages sometimes create content that is the same or similar but available on different URLs. This is generally not a problem as long as the content is for different users in different countries. While we strongly recommend that you provide unique content for each different group of users, we understand that this may not always be possible for all pages and variations from the start. There is generally no need to "hide" the duplicates by disallowing crawling in a robots.txt file or by using a "noindex" robots meta tag. However, if you're providing the same content to the same users on different URLs (for instance, if both "example.de/" and "example.com/de/" show German language content for users in Germany), it would make sense to choose a preferred version and to redirect (or use the "rel=canonical" link element) appropriately.

Do you already have a website that targets multiple regions or do you have questions about the process of planning one? Come to the Help Forum and . In following posts, we'll take a look at multi-lingual websites and then look at some special situations that can arise with global websites. !