| This 105 message thread spans 4 pages: 105 (  2 3 4 ) > > || |
|Google, time to define "RELEVANT" or "RELATED" linking and reciprocal linking|
what's the idea of MC and GG about what it is relevant & related in linking
| 10:37 am on May 19, 2006 (gmt 0)|
I'm a bit worried about the semplicistic concept of "relevant" or "related" content used by MC when he talks about linking and reciprocal linking.
I explain what I mean with an example: we are a hotel reservation website and we deal with hotels in various
destinations of the world.
Our "related resources" are the ones that would be _USEFUL_ for a traveller.
As the traveller will book the hotel with us, the rest of the resources are "complementary" resources and not
Example of what we link and what our travellers want us to link (as these are useful things to know if you have already booked or about to book an hotel):
- Car rentals
- Airport transfer services
- Bicycle Rentals
- Art Galleries
- Food Festivals
- Casinos (Yes, if you book an hotel in Las Vegas, you want to know the best casinos if you don't have one inside your hotel)
- Clubs and Discos
- Festivals & events
I also have another 195 categories of resources that we regularly link in order to build a good service for our
As you see, these are all hotel and travel related resources, that makes our websites very visited and one-way-linked just because these are useful info for a traveller than wants to book an hotel and know more about the area.
NOW: I'm worried about what MC says in his blog and about the use and definition that all the SEO world has done about "relevant/related" content.
It should be natural that a website will link COMPLEMENTARY resources, not COMPETITORS. Therefore, the keywords to be inspected on our outgoing links are 100% different from what we sell.
Therefore, I'm deeply worried about the concept of "related" that Google will or is applying in evaluating what type of links you have on your pages.
"another real estate site......I checked out the site. Aha, Poor quality links...mortgages sites...."
Now: is MC aware that mortgages sites are natural and relevant and pertinent to be linked if you are a real estate agent, as you might want to give related services to your visitors telling them how to find the money to buy his services?
Or does MC search for the related content in terms of a semplicistic "real estate words are good, anythign else is bad"? I mean: is Google even thinking about the fact that a real estate site cannot link a competitor but will be more likely to link complementary services?
In short: does Google and MC want us (a hotel reservation service) link Hotels.com as it will be relevant (and a complete nonsense as they are our competitors) or is googe "mapping" the related (complementary) services for every industry?
I doubt that Google will have a map of every complementary service for any given industry: therefore, I'm afraid that "related" for MC means "same topic, same industry... competitors, essentially".
Will MC want Expedia to link Orbits, in order to evaluate Expedia's lik as relevant?
Or will MC and Google better evaluating (or not "worse evaluating" at least) Hotels.com linking Avis or Budget?
[edited by: engine at 2:20 pm (utc) on May 22, 2006]
[edit reason] formatting [/edit]
| 9:14 am on May 20, 2006 (gmt 0)|
Of course Google want's webmasters to only link to relevent sites. Then they essentially will be doing G's work. )
My (simplistic) view is to create my site around what my visitors will want. Search engine flucuations be damned.
| 9:25 am on May 20, 2006 (gmt 0)|
the idea of whats a relevant link is something thats aways stumped me, after all as in your hotel site just about any other site could be viewed as relevant! and really this applies to a lot of other sites.
So when they say relevant are they talking 100% spot on as in hotels only linking with hotels or hotels linking to say bars - after all if your going to hotel you may want to find out about local bars
| 10:22 am on May 20, 2006 (gmt 0)|
All of Google's current problems seem to be rooted in a single inalienable truth:
Google currently believe themselves to be much smarter than they actually are.
As a consequence, there are probably two answers to this question: Google's imaginary one, and the real one.
Google's imaginary answer would go something like: "Don't worry about search engines. Just create useful, original, content-rich websites. Our algorithms will intelligently determine what each site is about (it's "theme") and mark all inbound and outbound links accordingly. Any links likely to be of use to users of a Hotel booking website, for example, would count as relevent to that website."
The reality: "Algorithms just aren't smart enough for this task. On a website that predominantly seems to be about Hotels for example, relevant links will be those that feature anchor text and content that predominantly feature the keyword Hotel/s".
With Big Daddy, Google has switched on a number of new algorithmic filters, none of which are up to the task for which they were designed. Only sites with vast numbers of inbound links, such as large commercial sites and spammy link-farmed sites, seem to be immune to the worst effects of these filters.
| 10:27 am on May 20, 2006 (gmt 0)|
that does seem to be the current situation Clint talk about frustrating & confusing
| 10:52 am on May 20, 2006 (gmt 0)|
|Google currently believe themselves to be much smarter than they actually are. |
That's a nail hit squarely on the head.
It's often said of pop-stars, etc. that the end starts with believing their own publicity. I think Google have been guilty of that mentality for some time now.
Back to the subject of this thread...
Since it is impossible to predict future algo changes or the effects of those changes, design your site for users, apply broad-spectrum SEO (but not not excessively) and keep your fingers crossed. Remember, the so-called "Sandbox" was (according to GG) an unexpected result of algo changes. If the designers/engineers can't predict the results of algo changes, there is very little point anyone else trying to do so (unless you make a living from black-hat SEO).
| 3:52 pm on May 20, 2006 (gmt 0)|
I don't know if Google is as smart as they think, but the hotel example actually shows that google doesn't need to be exceedingly brilliant to spot patterns of relevance. In fact, this example demonstrates how simple it could be for Google to identify the relevance of all these links.
Think for a second: what do all of the sites listed have in common?
Answer: they are all (or mostly) sites sponsored by brick and mortar businesses in the general vicinity of the hotel.
There are at least two ways Google could spot this geographic pattern quite easily.
First, it might notice that all of the sites have city names, street addresses and zip codes that identify them as having content related to a specific geographic area, which just happens to match that of the hotel.
Second, the whois information for many -- perhaps most -- of these sites would indicate that the registrant, or the service they hired to help them maintain the site (the technical contact) is located in a specific state or city, which just happens to match that of the hotel.
P.S. Didn't I read something somewhere that Google had plans to pursue "local" search? Weren't there rumours they were going to launch a beta product that competes with mapquest? Or was that my imagination?
| 4:45 pm on May 20, 2006 (gmt 0)|
Another possibility is that Google develops a kind of semantic network. Concepts are used often together on pages/sites, sometimes close to eachother, sometimes far away from eachother. Only thing Google has to do is to use statistics to measure the amount of relatedness between pages/sites with the same concepts on the web.
Hotels and car rent occur often on travel sites so relatedness is high value, hotels and morgages occur nearly never together so will have a low relatedness value. This is possible to automize but needs a lot of computing power and data.
Maybe the semantic network of Google is will improve over time when it is evaluated with new data.
Anyhow this is how I will interpret relatedness when making links. For the concept Tourism hotels, car rents, musea etc. are relevant. For the concept Vegas Casino's is relevant.
| 5:14 pm on May 20, 2006 (gmt 0)|
|I don't know if Google is as smart as they think, but the hotel example actually shows that google doesn't need to be exceedingly brilliant to spot patterns of relevance. In fact, this example demonstrates how simple it could be for Google to identify the relevance of all these links. |
The assumption a small team could write a program that can detect what is relevant for all people on the planet at all times is really very worrying.
| 5:19 pm on May 20, 2006 (gmt 0)|
|Google currently believe themselves to be much smarter than they actually are. |
There in lies the problem.
Nothing against those with PhD's, but from my experience working with people who possess PhD’s, a lot of them (not all) tend to over think things no matter how simplistic the problem may really be.
Now, put 100’s of those types together at the Googleplex and you end up with stuff like Bid Daddy!
| 5:23 pm on May 20, 2006 (gmt 0)|
|This is possible to automize but needs a lot of computing power and data. |
So if you crack the holy grail of AI, it still means you can't even link to other sites you made.
IE Google Mail can't link to Google Trends as it is irrelevant?
Without more information how they do that, we are in a very dodgy area here.
Relevance, regarding it was made by you or your friend can't be detected by a link list.
Maybe you are just linking to a site you want other people to know about. How you gonna detect that with an algorithm? You can't.
| 5:27 pm on May 20, 2006 (gmt 0)|
|Nothing against those with PhD's, but from my experience working with people who possess PhD’s, a lot of them (not all) tend to over think things no matter how simplistic the problem may really be. |
On the other hand my PhD tells me, what G tries to do is impossible .. ;)
On matters that are as complex as the web, as soon as you patch one hole the next one pops open.. :)
Bit like state budgets, really ..
So you just fiddle about and pretend you are doing something worthwhile, while fully knowing it's impossible to solve.
| 7:05 pm on May 20, 2006 (gmt 0)|
<<<Maybe you are just linking to a site you want other people to know about. How you gonna detect that with an algorithm? You can't. >>>
Not without losing a little relevancy or relatedness for the keywords you optimized the page for I would think.
| 8:06 pm on May 20, 2006 (gmt 0)|
I would love to hear how google determines if links are relevant. I guess some smart engineer made up some magnificient algo for that.....
What ever happened to content?
| 8:09 pm on May 20, 2006 (gmt 0)|
It's their job to determine relevant. This isn't news. Making it a more (or less) important scoring factor is just a tweak.
| 11:45 pm on May 20, 2006 (gmt 0)|
|It's their job to determine relevant. This isn't news. Making it a more (or less) important scoring factor is just a tweak. |
I don't want to shatter the mystique or anything but "determining relevence" for a simple keyword-based query is easy-peasy: all pages that contain all of the keywords are relevent. Job done. Now, ranking them perfectly, i.e. determining which pages are likely to be the most relevent is considerably more complicated.
Now, can we apply the same easy-peasy logic to determine if one website is relevent to another? Maybe. But what you certainly cannot do, is use any algorithmic logic to reliably determine that one website is not relevent to another.
Now, if only someone would tell Google this, we could all get on with our lives.
| 12:43 am on May 21, 2006 (gmt 0)|
|My (simplistic) view is to create my site around what my visitors will want. Search engine flucuations be damned. |
I wouldn't suggest spending too much time worrying about what any search engine specifically views as "relevant."
Best to re-read our Quality Guidelines [google.com], particularly this excerpt:
|Don't participate in link schemes designed to increase your site's ranking or PageRank. In particular, avoid links to web spammers or "bad neighborhoods" on the web, as your own ranking may be affected adversely by those links. |
There are some obvious humorously-bad examples of what links AREN'T going to really be of interest to your visitors.
- A link on a orinthology-info site to a gambling site.
- A link on a clothing-store site to a low-interest-loan site (though I suppose if the clothes were REALLY expensive... heh!)
Basically, doing any link stuff for money or related gain that abuses your visitors, is specifically done to trick search engines, or in particular is done very clearly for Page Rank... that's a Bad Idea(tm).
On the other hand, if there's a site you think your visitors would like to know about, I'd say that is relevant.
Look, I know, you want bright lines. I understand. But that's not always going to happen. There is a fine (and admittedly in some contexts, frustrating) balance search engines must maintain to serve users and Webmasters and avoid giving bright lines to those who would abuse users. As you can imagine, too, we can't possibly enumerate every possible "good thing to do" and "practice both your Mom and Google would frown upon."
| 1:15 am on May 21, 2006 (gmt 0)|
Look, I know, you want bright lines. I understand. But that's not always going to happen.
1. No offense, but this is the same mantra we've heard from G since day 1. Nothing new here. Nor is it "any" type of line because...(see 2)
2. Every webmaster here can point out sites on the top of their SERPS that DO get or buy links from rather irrelevant sites. Big name sites that don't have 10% of their links from even remotely "relevant" sites. Which leads to ....(see 3)
3. Webmasters not giving any "trustrank"(tm) into what Google preaches and what actually works.
Care to expound?
| 1:37 am on May 21, 2006 (gmt 0)|
"'determining relevence' for a simple keyword-based query is easy-peasy"
Of course. Basic relevance is mostly a 1990s concept, but that isn't the point.
"But what you certainly cannot do, is use any algorithmic logic to reliably determine that one website is NOT relevent to another."
A strawman, and a bad strawman at that. An algorithm can and should determine what is relevant, and what likely is not relevant. As always, desiring black and white answers is just not helpful. They aren't going to be perfect, and they will sometimes make poor judgments, but they should make judgments. Pages and websites as a whole that link to many other sites/pages that appear non-relevant to the content of the original page should hoist a red flag and get a scoring demerit, whereas pages that link to relevant, QUALITY content should get a scoring boost. This wouldn't be remotely controversial is there weren't yarn websites out there exchanging links with motor oil sites strictly to decieve the engines.
Relevant and related should be extremely easy for a site owner to judge with reasonable accuracy. I link to a friend of mine's wildly offtopic site because he is a friend of mine and that makes it relevant, but I don't expect a search engine to figure that out. I do expect that they judge this link an exception to the rule of me linking to high quality, relevant sites.
| 2:07 am on May 21, 2006 (gmt 0)|
I too am worried about this "relevancy" issue. Lets say I have a page that links to the following: a banking site, a weight loss site, an ecommerce site selling widgets, a sports site listing events, a non profit childrens program. Looking at that your first thought is there is no way these are relevant/related item.
But lets say I am a webdesigner and these links are items in my portfolio. Live examples of my work are extremely relevent to potential customers. Do I now have to be concerned with linking to examples of my own work.
If you provide a service/product to a varied customer base are you no longer "relevant" and deserve to be penalized?
| 2:37 am on May 21, 2006 (gmt 0)|
Good post, Adam.
I took the time to read through Matt's blog earlier, at least his posts and some of the replies that seemed pertinent. Adam, Matt, and GG (should be GGG for Generic Google Guy), have now made it very clear that linking schemes that worked in the past don't work now. Specifically:
Having the vast majority of your IBL's as recips is suspect. This filters out a lot of sites that went missing in BD (affiliates, grab and dash link farms, etc). Sites that have natural, un-reciped inbounds, especially to pages deeper in the site tree, are seen as more legit (as they should be).
It's a theming thing much more than before. They're looking at clusters, and a site that's all over the board, desperately sucking traffic from wherever, and linking to whomever, looks dodgy, especially if there's limited original content (as it should).
Along with that, it's been made clear that there is a greater value in being "unique" now. Matt suggested blogs as a source of unique content that others might want to link to (without having been asked to do so), although I think that's a lowest-denominator suggestion. [Some sites (mine) are getting un-reciped inbound links from blogs, so what the hec... fine with me, the more blogs the better, even if they don't get links to them.] So, despite some posts in the various BD threads that scoff at content still being "king", it seems that it's as important as ever. It wasn't just a suggestion on how to get natural links, it was a pointer. Of course, that's how you get unrequested links in the first place, so it goes hand in hand.
Imho, we're still looking at Brett's 26 steps to success, and some of the casualties of BD should rethink their approach. Not meaning to be a troll here, but at some point people are going to have to move on. Maybe the future of the internet is to not make money solely on the basis of having a network of sites, but to actually supply something, such as an original product, or idea. Unfortunately, it seems there are many webmasters who think that scooping traffic en route is a legitimate business model.
| 2:55 am on May 21, 2006 (gmt 0)|
New links that Google recognizes [ spiders ] are given a default value.
Google later evaluates the "weight" based upon the strength of the linking "page" and assigns a new value.
Relevancy comes into play after several iterations and only on the highest weighed inbounds.
A site that has low weighed inbounds will not experience the affects of this phase.
| 3:20 am on May 21, 2006 (gmt 0)|
Keep in mind that when the search engines are assessing relevance, they're looking at huge numbers of sites, not just yours.
If yours was the only hotel site in the universe that linked to a car rental agency, the algos might not recognize the good sense in that. But if enough hotel sites linked to car rental agencies, the algos would start to recognize the pattern. Whey they saw your hotel site linking to a car rental agency, they'd think,"Yep, that's a pretty standard thing for a hotel site to link to." So that car rental link would indeed support the "hotelness" of your site ... because it's something that hotel sites do.
Assorted links to bicycle rentals, art galleries, and so on that someone mentioned earlier might not be such a clear support for the "hotel-ness" of that site, but they'd certainly help to reinforce its "Las-Vegas-ness". It would all add up to support you for "Las Vegas hotels".
I wouldn't be surprised if the algos recognize that it's common for web designers' portfolios to have extremely varied links. They'd look at that site and think, "Hmmm ... she says she's a web designer, and her links are all over the place just like the last bunch we looked at who said they were web designers. What other clues do we have?"
(I have a very anthropomorphic view of the search engines!)
| 4:53 am on May 21, 2006 (gmt 0)|
A link is relevant when people use it. You might think a link to 'Bicycle Hire' is relevant on your hotel site, but if nobody clicks on it then its not relevant. User habits surely determine relevancy?
And if this is the case, the more links you place on your page side-by-side, you're in effect diluting their usefullness as peoiple are less likely to click any particular one.
| 5:03 am on May 21, 2006 (gmt 0)|
I wouldn't worry too much if I were you. Despite the statements that Google can't possibly determine relevance, they're well on the way: Google Sets [labs.google.com] for hotels+car rental+airlports+Restaurants+las vegas
They consider those results highly relevant, yet they vary more widely than your examples do. The common element, of course, is "travel."
Google purchased a company called "Applied Semantics" several years ago (ca. 2002). They are almost certainly using the ideas they bought with it.
| 5:38 am on May 21, 2006 (gmt 0)|
This is all well & good ... But there is still something tremendously wrong when the following occurs:
3 well established Wikipedia pages (Wikified quite a while ago), "dead on" topic links from those pages to a website that is 4 years old (same subject, but no duplicate content), Google page caches for the Wikipedia pages updated many times over ... Still no visits from Googlebot (in any form) to the linked site.
Just the Wikipedia links alone bring in quite a bit of visitor traffic from Wikipedia visitors. That in itself says quite a bit. Everyone knows how fast off topic links are killed in Wikipedia. Mine have stayed in because they are directly on topic, use on topic anchor text, and the site they point to contains more in depth information on the same subject.
How can Google NOT consider this relationship relevant?
| 6:07 am on May 21, 2006 (gmt 0)|
|Despite the statements that Google can't possibly determine relevance, they're well on the way |
Google naturally cannot determine relevance in many cases. They could not even if they had solved the artificial intelligence problem and they are far from it.
Just one extreme example.
Suppose that people don't know that the Earth is round. Then a scientist discovers that the Earth is round. The page will be out of the Google index until the idea is widely accepted.
Fortunately, Google team probably understands the problem and will preserves the root page of the scientist site, so people will have the chance to know that the Earth is round if follow the links in the scientist site to the not indexed page.
| 6:37 am on May 21, 2006 (gmt 0)|
I have always assumed relevancy related to the page not the whole site.
|But lets say I am a webdesigner and these links are items in my portfolio |
I'd think if you write a bit about each sit you designed including their topic along with the link you would be covered.
Along the same line if a person has a travel page about transportation in the area the links to car rentals would be relevant.
I don't know what is causing pages to go missing in the serps and all but I don't think it is imperfect relevancy of a few links.
PS - Could someone give the date of the MC info so it would be easier to find.
| 7:49 am on May 21, 2006 (gmt 0)|
I found the MC bit and waded trough it responses and all.
On one hand he points out unrelated links as being the reasan a site might not have been completly crawled. But then he implies in some discussion posts that unrelated links and off topic links neither help nor hurt a site.
| 1:39 am on May 22, 2006 (gmt 0)|
There is no difinitive answer to what Google may consider relevant or related as far as linking goes, but if we use our own logic, Google will follow. As jdmorgan pointed out, Google sets is a pretty good indicator of how well Google is able to determine relevance and valid relationships.
If I have a flower shop and link to sites about:
garden photo sites
flower pot suppliers
A casino in nevada
All would easily be considered both relevant and related by Google ... with the possible exception of the casino in Nevada.
However, If I linked to that casino because I supplied all the flowers for a wedding which took place there, then presumably, I would be linking to a page which displays the floral arrangements I was responsible for providing and not their home page.
Somewhere on that page, my company might be mentioned ... floral arrangements by Gail's Flowers ... and there may even be a link to my site on that page.
I see no problem at all with that sort of link or even a link exchange. However, Google may not value that particular link too highly ... but it shouldn't hurt you either.
Now if I were to link to the casino home page and there was no mention of flowers on that page at all, I would imagine that sort of link might raise a red flag if there were also ten more links to casinos ... and all of them to their home pages.
Now, if my business was the official supplier of floral arrangements to those ten casinos ... they would be legitimate links. As casinos buy a lot of floral arrangements on a daily basis, it would likely figure strongly within your website copy that you specialize in supplying casinos! In that case, links to their home pages may be considered legitimate.
However, if I also were linking to a site about car stereos, computer printers, a site selling knives and a ladies underwear site ... those sites may be considered highly suspect and the link values may be considered nil. They may even count against me ... I don't know.
If any of those sites were also linking back to mine and also showed no particular relevance in their copy anywhere on the page and if they happened to link to hundreds or even thousands of other sites not related to what they sell ... then you can bet those suspect links might then be considered a bad neighbourhood. All of a sudden, your site begins to drop in the rankings or perhaps those pages disappear from the index by way of a penalty.
I liken outgoing links to having safe sex. Check out your partners before getting into bed with them. That is not to say you should be paranoid ... just cautious. If you know you are linking for valid reasons ... then go ahead and link. If you have any doubts about the validity, then Google likely will too!
As for webmasters linking to sites they created ... that is a legitimate practice and nobody should be worried about that. You body of work or portfolio is a legitimate way to display your work. To be on the safe side, be sure your copy on that page of links explains what you are doing. There's nothing wrong at all in linking to your own work.
| This 105 message thread spans 4 pages: 105 (  2 3 4 ) > > |