Chief Marketing Officer at Wordtracker.com
12 February 2003 14:47pm
Link popularity is increasingly recognised as an important part of any marketing plan. Search engines and speciality sites such as www.linkpopularity.com allow you to measure the link popularity of your site.
However, from my experience the ‘scores’ produced by these measurements need to be taken with a very large pinch of salt.
In this post, I want to give two examples of dodgy measurement, outline the real benefits of a linking strategy and stimulate some debate and feedback.
First, two examples from my own work of measurement gone wrong:
1. A major pharmaceutical site has over 5,500 inbound links to its site. But by analysing the nature of these links it turns out that corporate links (i.e. internal links from their own brand sites) made up 63% of these, pharmaceutical industry directories (mostly dull, low-traffic sites) made up another 14%, academic sites 12%, charity sites 5%, suppliers, 3% and others 13%. Indeed, just 1% of sites from our sample matched the type of link they were after.
2. A major entertainment site in the UK raised my eyebrows when it scored over 18,000 inbound links – 6 times more than its nearest rival. Such a disparity suggests something fishy is going on. Some basic detective work quickly uncovered the trick they were using and I was able to revise their score to a ‘genuine’ 900 or so.
Such examples probably arise from the need to boost search engine rankings and link popularity’s undoubted efficacy in doing so. These tactics, while they may be successful for a time, miss the point and demean the practice of building link popularity.
Second, the benefits of a linking strategy:
An effective linking strategy will take many days of effort over a period of months, but the benefits are well worth it. These include:
(i) Increases industry knowledge - finding out who links to you, who links to your competitors and identifying the major portals and ezines for your industry will not only identify linking opportunities but give you a tremendous view of your industry online.
(ii) Builds traffic - following a link is one of the most popular ways to find new sites. Research from AltaVista, Compaq and IBM shows that sites that are well-linked share high volumes of traffic.
(iii) Improves search engine rankings - most search engines now include link popularity in their algorithms and a proper linking strategies is one of the most legitimate ways of boosting your rankings.
(iv) Creates value for customers – site users love useful links. Sharing some or all of the information you collect while doing link research will provide great value for your site users.
(v) Builds Trust - links from respected industry sites will help small or new companies build trust.
(vi) Links attract links – sites that already have a high level inbound links of are most likely to attract further links. Growth can be exponential.
(vii) Builds community - most importantly, a well linked site becomes part of the informal but well-trafficked community around their industry.
Third, the stages involved:
(i) Analyse your current inbound links
(ii) Analyse your competitor’s inbound links – this will give you your first target list.
(iii) Find further sites that could link to you – this will give you your second target list
(iv) Make sure you have content that potential link targets will find attractive
(v) Set business objectives for your linking strategy
(vi) Make sure your site is link-friendly
(vii) Publish appropriate outbound links on your own site
(viii) Ask for inbound links
(ix) Monitor results and build on them.
I’ve conducted linking campaigns for major clients over the past two years and have sometimes been frustrated by their inability to grasp the real marketing potential of an effective linking strategy. So I’ve created a free site at www.linkingmatters.com to share knowledge and point to extensive resources on linking strategy.
E-consultancy has kindly agreed to host the 33-page report, Linking Matters.
I am working on a number of case studies and would be delighted to hear about other people’s experience in building inbound links – be they positive or negative.
Senior SEO at Weboptimiser
14 February 2003 12:21pm
Ever since the advent of Google and the PR based algo, some sites have lived or died by their ability to attract quality incoming links. Since the hardcore SEOs figured out that pure-play link pop is vunerable to having the life spammed out of it (farm those domains, boy!), Google had to do something a bit cleverer than that. We'll get there in a minute.
First, lets look at the original PageRank set-up, and roughly how that works.
Any domain that includes a link to another domain, can be considered to have cast a "vote" for the other domain. The underlying assumption is that every site is managed by an independent webmaster, who will only link to sites that he/she considers will benefit their users. Some sites will therefore have a higher measure of "popularity" (PageRank), and votes from that site are worth more in terms of boosting the PR of the target site. The overall concept is that PR should be a measure of the likelihood that a surfer will find your site during a random surf.
Sites that have many links, and or links from other popular sites are clearly at an advantage here (high PR).
Right, now we've established the reason to acquire incoming links, lets examine the examples
>1. A major pharmaceutical site has over 5,500 inbound
>links to its site
Most of them were internal, which makes sense. PR is calculated on a per page basis, not a per site basis (which leads to a couple of tactics to "pump" PR around a give site), and large sites, by the nature of the PR calculation will have more "natural" PR, and can generate a positive feedback loop effect. Add in a few good links from industry-speficic directories, suppliers and academia, and you're there
>2. A major entertainment site in the UK raised my eyebrows
>when it scored over 18,000 inbound links
From your further comments, I assume they were employing a tactic known as "domain farming", the practice of registering hundreds or thousands of domains, putting up some sort of content, and ensuring that each domain points one or many times to your "target" domain.
It's the second site that is, from the search engines point of view, the problem. The site is artificially inflating it's apparent popularity (and hence quality) by exploiting a flaw in the underlying logic of the algorithm.
So, if the logic is flawed, you need a new algorithm. Welcome to the Hilltop, people
You can see the paper for the Hilltop algorithm here,
Note that Krishna Bharat now works for Google, and is considered by many far better mathematicians than me to be at the cutting edge of information retrieval technology. To save you some reading, and a lot of brain ache, one of the key bits is here :
" A key phrase is a piece of text that qualifies one or more URLs in the page. Every key phrase has a scope within the document text. URLs located within the scope of a phrase are said to be "qualified" by it. For example, the title, headings (e.g., text within a pair of <H1> </H1> tags) and anchor text within the expert page are considered key phrases."
So, lets say your site is about widgets. You get 2 links, one from www.spamcore.com, and one from www.widgetworld.com. The <title> of the WdigetWorld page reads "Widget World - widget-tastic widget friends", the <h1> reads "Widget links" and the actual HTML for the link to your site goes <a href="http://www.blue-widget.com/fuzzy">blue fuzzy widgets</a>.
Thus the <title>, <body> content (which includes the <h1>), and the anchor text (blue fuzzy widgets) are all related to widgets. That link will significantly boost the apparent quality of your site on the topic of widgets.
The <title> of the SpamCore page is "SpamCore - for all your spamming needs!!!!", with no <h1>, and link code that reads <a href="http://www.blue-widget.com/">Click here!</a>. No mention of widgets anywhere (and SpamCore is likely to be a penalised site anyway, and so not have a lot of PR to "vote" to you in the first place), so not very much benefit, BUT if you can put thousands of them up, it just might do the job
In essence then, Hilltop is a ranking method that combines the straight "one site, one vote" type approach of the standard PR algos, and starts to introduce one of the search engineers Holy Grails, theming. There has been persistent speculation that Google has implemented (or at least partially implemented) Hilltop, or something very like it, at least for certain sectors of their database. It is possible to find results pages where a PR5 site sits at #1, whilst PR7 pages, that on the face of things should beat it, languish in 2nd page obscurity.
I'll finish up with a couple of tips :
1) Google supports a special search syntax, link:url, which will return a list of links that Google has recorded to that Url, with the following caveats : only pages with PR4+ return any results at all, only links from PR4+ pages will be returned, only half of all links will be displayed (this is intentionally reducing the utility of the tool, as when word first got out, the Google servers got hammered by every SEO in the world checking their own links, their clients links, their clients competitor links etc etc)
2) Alltheweb (www.alltheweb.com) supports a link: search with similar characteristics to the Google one (but no halving of results. Quality minima are in place though), BUT they also have a second link.all: syntax, which does return a complete list of links in their database, without restriction. Get it while its hot, if it becomes a strain of their resources they will likely dump it. The Alltheweb database is of a similar size and quality to the Google one, and so provides a useful second opinion (ATW also has the best advanced search features anywhere, its the webmasters choice)
Incoming links, love 'em or loathe 'em, you just can't ignore 'em. Did that all make sense, or have I run on too much?
Self Employed at Self Employed
14 February 2003 15:27pm
I've got just a couple of additions to your well informed comments...
>>1. A major pharmaceutical site has over 5,500 inbound
>>links to its site
>Most of them were internal, which makes sense. PR is
>calculated on a per page basis, not a per site basis
>(which leads to a couple of tactics to "pump" PR
>around a give site), and large sites, by the nature of the
>PR calculation will have more "natural" PR, and
>can generate a positive feedback loop effect. Add in a few
>good links from industry-speficic directories, suppliers
>and academia, and you're there
You're only "there" in terms of generating PR (Page Rank), and playing at the same games as SEO (search engine optimisers). The kind of links you want aren't only for PR, but for clickthrough traffic. It's the high-traffic health portals that the company really wants linking to them.
>Thus the <title>, <body> content (which
>includes the <h1>), and the anchor text (blue fuzzy
>widgets) are all related to widgets. That link will
>significantly boost the apparent quality of your site on
>the topic of widgets.
Increasingly this is the way search engines will go, not only looking at the anchor text, but also the title text, headings and finally, the entire body of the html page it finds the link on. If you look at http://www.linkingmatters.com you'll find all the relevent headings on a page are contained within <h1> through <h3> tags. The entire site is XHTML 1.1 Strict for the same reason. If the site is well coded it will be easier for search engines to correctly parse and understand.
Ultimately it all boils down to what hypertext is and should be. If search engine algorithms such as the one you describe are correctly implemented, and content providers and aggregators follow the spirit of hypertext and XML, search engines would be a lot better and content would be easier to find. All links (on site and off) should have relevent text describing their target contained inside them. For a good page on the topic of link text, look at http://www.w3.org/2001/06tips/noClickHere.
There are numerous technically orientated changes that can be made on a web page itself to help search engines and Page Rank. You can go as far as changing the order of elements within the html file, so relevent text sections come near the top, while navigation is near the bottom (though of course still keeping important internal links at the top).
But the strategy behind Linking Matters isn't related to either the quick fixes of SEO (gateway pages, meta-tag spamming and link farms), or to the common-sense standards compliance described above. These are all secondary to the idea of getting good links on high-traffic sites, and getting visitors through that, not through search engines. If you happen to get a high Page Rank by pursuing this strategy, all the better, and it shows search engines are doing what they are being designed to do, gather relevent content, but ultimately few businesses can rely purely on search engine referals to go about day to day business.
Marketing Director at Nemisys
18 February 2003 09:47am
I'm still wet behind the ears in terms of experience, but wondered if this site I have been working with casts any light on your discussion.
I guess I'm happy to hold it up as an anomally - we have done almost no search engine registration or inbound link promotion, the site has been 'inactive' for 6 months (they went bust, we hosted for nothing, but there was no new content during that time) and yet www.autismjobs.org has a Google ranking of 5 and plenty of relevant traffic as a result.
My assumptions are:
1) We've got the name 'just right'?
2) Charity sites score better than commercials?
This isn't a site plug (it's irrelevant to this forum!), and please be as brutal as you like about the site - I'm just trying to figure out more about how it ranks 5 with so inbound few links?
18 February 2003 13:25pm
Well, there are couple of things that may well have contributed to this.
Firstly, and probably most importantly, I don't believe the number of sites relating to jobs in the autistic field is that large, so Google's PageRank will be increased; this generally applies to any niche topic, it's quite possible to have an extremely high PageRank in something that very few people deal cover on the web.
Looking at DMOZ, Autism Jobs has a good listing on http://dmoz.org/Health/Mental_Health/Disorders/Neurodevelopmental/Autism_Spectrum/Services/. Google definately puts some weight towards DMOZ listings, as well as the fact a lot of people use DMOZ data in other search engines and directories.
Backward links do look especially low for a page with such high PageRank, but at least two of the pages that link to it have a high PageRank themselves.
The domain name matches the topic extremely well, but as for whether the .org suffix helps at all, I'm not sure, but wouldn't rule it out.
As for the actual pages, having a good "about autismjobs" link on all pages, as well as the constant footer with it's internal links should help.
19 February 2003 09:51am
As Archie has said, the dmoz link is very important, because not only do you get PageRank from it, but dmoz-listed sites also seem to get a slight ranking boost from Google anyway (which makes sense; to pass human review, the site must of reasonable quality)
Also, bear in mind that "toolbar PR" is a variable thing, not all PR5 sites are equal. Raw PR is calculated by Google during their big monthly updates, and is then expressed to toolbar requests as a number from 0 - 10, using a logarithmic scale of unknown base (but its around 6 - 6.5). You may be teetering on the brink of a PR6, or just barely over the line for PR5.
You don't have many links, but they ARE quality links, with the dmoz listing and a link from what I suspect is a major hub/authority on the subject
>> 1) We've got the name 'just right'?
The domain name won't contribute to your PR. It MAY have a small effect on your overall ranking, but I doubt it in this case (PR is only one of over 100 factors used by Google to rank sites, although it is an important one)
>> 2) Charity sites score better than commercials?
There is some limited evidence that .org, .edu etc sites recieve a slight boost.
Of course there is also the slim possibilty that a Google search tech has visited the site personally, and "awarded" you an increased ranking/PR just because he likes the site. Its rare, but it happens
Operations Director at Sitelynx Ltd
19 February 2003 16:25pm
This is of course one of the most hotly contested areas presently.
The difficulty with links is not only the Google PR of the site from whom one has a link, but the actual link reference that is provided. How one is described in the link itself.
There are issues concerning IPR matters with linking. One cannot indiscriminately link as one wishes fherefore it is to be recommended that a clearly defined link policy be issued, preferably on ones site so that links accrued describe one's own site in the best manner, rather than merely company name.
This should be reflected in the title of ones own site, so that the circle may be squared.
19 February 2003 16:47pm
Just wanted to say this is great stuff. First time I've posted here, and a pleasure to find a forum where people are knowledgeable.
Think I'm distilling things down to:
1) dmoz is very good
2) correct, relevant URL makes life easier (for PR and plain keyword search popularity, and I guess click-through in general search results, probably?)
3) providing the code/words for others to link back to you will over time give a better result for the same number of links
I've plenty to learn, but that's all a great help - thanks!!
Free market research on digital marketing
Daily Pulse: award winning newsletter
It takes 30 seconds to register