| This 45 message thread spans 2 pages: < < 45 ( 1  ) || |
|Theming: Is it a buzz word or is it real?|
What is it?
| 3:20 pm on Sep 30, 2002 (gmt 0)|
Is there any evidence that Google (or any other search engine) actually uses themes? I like the concept, but would like to know its practical implications, if any.
| 10:53 pm on Oct 5, 2002 (gmt 0)|
|While a search engine based on themes will find well the sites that best deal with the content, this could result in missing important pages because they are off theme to the site they are on. |
If I have a "potpurri" site with over 100 seemingly unrelated topics I could still potentially rank #1 for one of those topics even once the SE's take theming into account. (depending of course on how they do so)
If I have a small cluster of 10 pages devoted to the subject extensively interlinked, with only a single vertical link to the subsection from one or two of my highest PR pages, the page's theme would not be significantly diluted.
Furthermore, if the page was truly authoratative and filled with outstanding quality content, it could also attract inbound links from other on theme pages which would further strengthen it's "ThemeRank".
Add to this outbound links to other topical specific sites and pages and you have a mix that could still do very well even in a themed environment.
It would make sense for a given page's "ThemeRank" to be based on the contextual relationship between it's on page content and it's off page relationships... whether those relationships are coming from in site linkage or off site linkage could (and should imo) be irrelevant.
The inclusion of Themes into the mix does not necessarily imply a move from page authority to site authority... it simply adds context by defining that authority as being topic specific which could potentially improve SERP relevancy a great deal if done properly.
| 11:10 pm on Oct 5, 2002 (gmt 0)|
|It's easy to stay on topic if you are selling what people are looking for. But what about the guy who sells, say vans. He might know that there are 100 times more searches for cars than vans, and he might have much more success selling to the car audience than the van audience, purely because of the search volume. |
sorry, but IMHO this is spam! If you want to get qualified traffic then target jeep, cherokee, scenic or whatwever is relevant to vans ... not just simply cars. "Cars" could be -one- starting point of a theme pyramide related to vans - it could be targeted - but the sum of the whole site should be vans - other things are spam!
IMHO the problem is that people are trying to get traffic and loose the focus on their theme. If you want traffic (so, for the commercial sites - customers) you should target every possible word that is related to your theme - not the words "above" your theme!
Maybe there are 100 times more searches for cars than for vans - but maybe there are more than 100 times more searches for cherokee, scenic, minivan, familycar ... than searches just for cars!
| 11:19 pm on Oct 5, 2002 (gmt 0)|
>>links from sites outside your theme might hurt you more than help you.<<
I understand the logic behind this, but if I've got a light bulb web site ( best I could come up with ) and I've got links from the whole spectrum of sites ( any conceivable site ) then wouldn't that show Google that mine transcends category, and put me at the top of the pyramid? ( This is the first I've heard of the pyramid.)
| 11:25 pm on Oct 5, 2002 (gmt 0)|
Guess I'm a spammer then. :(
Must have gone over to the Dark Side while I was sleeping.
| 7:49 pm on Oct 6, 2002 (gmt 0)|
Does Google use themes in its algo, or not?
| 7:57 pm on Oct 6, 2002 (gmt 0)|
I think we've been discussing it here [webmasterworld.com]
| 8:35 pm on Oct 6, 2002 (gmt 0)|
Thanks. Been up to long and should have remembered that. :( The problem is that article by Brett says flat out that Google uses themes. But then we have Marcia, an admin here, saying about Google not being a theme search engine: "Maybe not in the strict sense of the word, but it's not conclusive that some of the factors that contribute to the concept of theming don't come into play." Brett's comments are so murky to me I really can't figure out what he says about this. Somebody on Usenet just slammed me for saying I didn't think Google was a theme based engine citing Brett saying absolutely it was; although that was 2 years old. Thus, what is the simple answer: yes, no, or maybe?
| 7:29 am on Oct 8, 2002 (gmt 0)|
I remember 2 or 3 years ago (or was it 4?) we decided to split our site up into 4 different domains. Partly due to Brett's early article on theming and other's backing it up, we felt our site was too broad (lets say our site is about "Business in Europe". In an evolutionary manner we had developed directories and sections relating to business travel reviews and tips in Europe, marketing in europe, and business culture and etituette in Europe. At the same time AV moved to generally listing only one page from each site in the SERP, so we felt our material on travel, or culture, for example would be less visible.
A couple of years later crosslinking and the effect on Google became the big topic. Though our sites were independent by then, they still crosslinked because there were all to do with business in europe, but each site looked at a different aspect, ie travel, culture, marketing, and management/organizational development etc. We quickly removed a lot of our interlinking and still today it is a major decision on what to link to and what not to, not because of off-topic relevance, but fear of getting caught in a Google trap.
We have never had a penalty however, we have become less worried about interlinking as long as it is done with the user in mind, and we are glad we applied Brett's theories on theming. His latest post is of extremely high quality. As search engines mature from 100% text on-page driven, that list of criteria he outlines is important. It also explains why people who use just a few technqiues that worked last month to boost exposure in Google rankings were so upset this time when i think Google made a significant step to doing just what Brett is talking about.
When Google themselves say they use over 100 diff criteria to determine ranking, many of these which are driven by complex foumlaes themselves, and many of which the webmaster has no contril over as they are controlled by the community of other webmasters, it must be obvious that SEO is no longer a cottage industry, that only the real smarties can even get close to getting a long term, non spammed, prescence in Google, and that SE ranking technology is getting closer to the point where ranking will be failry impervious to SE), and predominatly based on content, themes, and link patterns. Yes. Google sets is a great illustration on how themes can be built algorthmically. I use it a lot for my own "non Web research". Know only a few clothing brands and someone has asked you to develop a list of all major brands? List them in the google sets page, and it will return a big list of most clothing brands worldwide. You dont even have to say they are "clothing brands". The relationship between the elements do it for you. You can do the same for countries, political leaders etc. its like a thesaurus.
To take it a step forward, it can also establish a basis for determining "authority" - which is another key concept that needs to be considered in how search engines expose useful resources in the future.
Other tools like the Google java interlationships thingo and Kartoo can also show "themes" by the relationships between key phrases on a website or two.
For long term exposure, I agree, themes have to be embraced.
| 8:13 am on Oct 8, 2002 (gmt 0)|
Therefore, the shopping malls, the department stores, the Wal-Marts, etc. with hundreds or even thousands of different products have no meaningful future on the internet.
The search engines are going to decide to offer us Louie's Paint Store rather than Wal-Mart for a gallon of paint. We are better to be directed to Vinny's Hardware for tools rather than the Craftsman tools sold exclusively at Sears. Our designer perfume is best found at Aunt Ruby's Second-Hand Perfume Kiosk than Neiman-Marcus.
The assumption is being made that human behavior is based on themes. In contrast, I believe people make decisions based on exposure. So does the whole advertising world. One may go to the convience store for just a loaf of bread but, like me, walk out with a carton of cigarettes, lottery ticket, bottle of beer and a pack of rolling papers. Only to get home and have to go back for the bread.
If one could define and predict human behavior with numeric equasions alone they would rule the world. Much like the Irish would if it weren't for whiskey.
[edited by: nell at 10:26 am (utc) on Oct. 8, 2002]
| 8:25 am on Oct 8, 2002 (gmt 0)|
|Therefore, shopping malls with hundreds of different products have no future |
you are right, but theoretically you can still be very "themed" per product, whilst having several different topics on one site.
Take the example of a university site with 100 different not directly related programs.
Each faculty would be a subdirectory of the main university site (but still the same site).
If each faculty has hundreds of really topical, authoritative, theme related information, they will earn their own theme related incoming links.
| 10:20 am on Oct 8, 2002 (gmt 0)|
|If each faculty has hundreds of really topical, authoritative, theme related information, they will earn their own theme related incoming links |
The University itself loses identity. The overall educational experience is not exposed. Why would I want to attend a University for Physics that had no Math Department? Engineering with no Engineering Library? I've got to interogate "hundreds of really topical, authoritative, theme related information" sources to determine if I want to attend that University?
| 10:51 am on Oct 8, 2002 (gmt 0)|
I am not sure if we are meaning to say the same or not:
|I've got to interogate "hundreds of really topical, authoritative, theme related information" sources to determine if I want to attend that University? |
What I tried to put forward is:
- in real life you can have many highly topical, but not really subject related themes within one configuration (the different faculties within one university)
- in real web-life these different themes can be hosted under one configuration (the main university url, i.e. one site)
- in real Google life these different, individual themes could still be rewarded as being a strong theme, even though they are hosted within the same site.
Why? because Google will (sooner or later) look at the clusters of pages that link together and relate their theme according to the internal links and the incoming external links and their theme related and weighted subjects.
Or in other words, it may be more difficult to have several different themes within one site, but it is still possible and perfectly respectible.
| 12:52 pm on Oct 8, 2002 (gmt 0)|
|The University itself loses identity. The overall educational experience is not exposed. |
I think we're losing sight of the trees for the forest here. (inverse reference intended)
We're talking about specific keyword driven topical searches here.
The example cited wasn't in reference to a search engine user looking for an overall educational experience.
It was pointing out that a multi-faceted university site could very well come up highly ranked for a very specific keyword search... even in an algorhythmic environment that weighs themes heavily.
If someone is searching for a specific advanced mathematical formula, and M.I.T.'s math dept. is renownded for it's research papers on the subject, and hence, has a mini-portal devoted to it with hundreds of inbound links from related sites that recognize it's authority on the subject... shouldn't it come up highly ranked for that search?
In this scenario the theme is defined by a grouping of related pages... some on site, some off site.
Themes will come into play as a way of mapping virtual geographies of related pages... whether they are on the same site or spread out across many domains will not be the single defining factor. Relevance based on keyword specific authority will.
|I've got to interogate "hundreds of really topical, authoritative, theme related information" sources to determine if I want to attend that University? |
If you want a sense of the "overall educational experience"... you won't be *doing* highly specific topical searches to get to their pages.
If you're interested in overall educational experience you're likely to be doing general searches like "University Comparisons" and "College Reviews" which will give you third party editorial pages... and when you want to investigate the Universities themselves you'd be visiting their admissions and home pages.
OTOH if you have a highly refined idea of what you want to learn you may well do a search for specific phrases... in which case you'd likely be far more impressed by a University that has an entire niche portal dedicated to your interest of choice.
A site shouldn't come up first for 10,000 different products just because they stock them. Though they should come up first for a product or service for which they are the single most renowned source... and in a well executed theme based environment they will.
Because the strength of a given page's "ThemeRank" will not be limited by a single domain... it will be defined by the shared context of all pages that are linking to it.
If I create a 1000 page site devoted solely to Red Sparkling Doodads (RSDs) that will certainly help my chances of ranking well for them... but if the most popular Red Sparkling Doodad distributor in the world has only one single page offering them and has 10,000 fan and affinity sites linking to it I will never top it in the rankings in a well executed theme based environment.
Likewise if my page has 10,000 topically random inbound links because I offer a hit counter that is widely used, I shouldn't expect to beat out another RSD page that only has 1000 topically specific inbound links exclusively from sites devoted to RSDs.
Understanding the potential of Themes, like most things, is all about Context. (pun intended)
Dante (not to be confused with The Inferno... different Theme. ;))
| 8:54 pm on Oct 8, 2002 (gmt 0)|
I am not positive that I am interpreting the concept of themes properly, but I think I have seen evidence of an increase in themes in the latest Google algorithm.
Last month, a site themed around widgets was considered very relevant by Google for phrases like: widget services, online widget, widget lawyer, widget advice etc. After this month, the site is viewed as relevant for all except widget lawyer which took a severe nose dive. The widget lawyer SERPs are now dominated by sites around the general lawyer theme and the widget part of the query it is much less apparent in the SERPs. The lawyer theme seems to overpower close proximity between widget and lawyer.
I originally interpreted these results as if Google was just down-playing word proximity, but now I think Google might also be up-playing themes.
I am wondering if people who saw similar nose dives for some of their rankings can isolate the nose dives to phrases that contain words with well defined and competitive themes.
| 7:59 pm on Oct 10, 2002 (gmt 0)|
>Thus, what is the simple answer: yes, no, or maybe?
How about "sort-of" or "not-really" for an answer. Google is not a theme search engine because Google is not using a Term Vector database.
However, Google is taking into account some contextual elements, such as link text relevancy (and perhaps others such as Brett & Marcia alluded to) that leverage a themed website.
A themed website is simply pages of related content linking to each other. Google doesn't have to analyze the your entire site as a group of pages for this to be valuable (i.e as a Term Vector database would). Google merely needs to look at the how related the content of the interlinked pages actually are.
I think at the current time, where Google is/is-going could be called a *context-search-engine*. Themes really applies more to what can be done within a website.
| This 45 message thread spans 2 pages: < < 45 ( 1  ) |