By checking the 'exclude on-site links' option, only external (off-site) links will be processed by OptiLink. Links from pages on the same domain as the target will be skipped.

Take a look at the details shown in the log and summary views and you will see where the reported internal links are dropped and how many links (all external) were actually found and processed.

There is basicly no rule about what links get reported by the engines, so generally not all your 'real' external links will show. Of the ones that do not show, some will actually exist at the search engines and not be reported, while others will not have been crawled yet at all.

OptiLink does not look at robots.txt because it is not a spider. It will take whatever you give it in the additional pages list and/or whatever is returned from the search engines you tell it to access.
By way of contrast, OptiSpider is a spider and does obey the robots exclusion file.

The combination of OptiLink and OptiSpider used together can almost do what you want, with one caveat.
Here goes:
First run OptiSpider on the domain where you are considering links. This will spider all the pages on the site, which should be all and only the pages Google will find.

Doing that analysis doesn't really make much sense -- it is never meaningful to include Google and Google cache in the same
run as this is classic "apples and oranges". These two interfaces will always analyze precisely the same set of links, the only difference being what versions of the pages are analyzed. The GoogleCache looks at Google's snapshot of the

You can save and edit the report html files and insert CSS like the following.

table {display: table;table-layout: fixed;}
tr {}
td {overflow: scroll;width: 20%;}

This works in IE, it does not work correctly in Netscape/Mozilla :-((. For those browsers, change the overflow: scroll to overflow: hidden -- the full URL is still available by coping the cell contents with copy/paste.

OptiLink analyzes a single page, whereas OptiSpider analyzes all the pages of a given domain.

OptiLink depends primarily on the search engines to discover linking pages, both internal and external, while OptiSite discovers all internal links automatically.

The filtering of link results by google is mostly about the page rank of the linking pages. Internal links are just as good as external links for building both link reputation and Page Rank.

To see if your pages are included in the google index, type the URL of the page into the Google search box and see if it recognizes it as a URL in its index. Also check your server logs to see what pages googlebot has visited.

Optilink incorporates 'stealth features' so your IP looks just like any heterogeneous computer network to the outside world. This makes it very hard to find a reliable signature of what is or is not OptiLink in operation. In two and half years there has never been a case of an OptiLink user being shutoff by Google.

The IOException error indicates that OptiTools was not able to load a webpage. It automatically retries several times, so you may see several of these errors together. These errors are caused by any of:

It all depends on what you are trying to do. For Ranking Analysis, it is off-page factors that are key and only OptiLink can help there. But if you are instead wanting to manage a Reciprocal Linking project, Arelis is probably very useful (I do not know from my own experience).
OptiLink will be helpful as well by providing data about potential linking partners so you can pick and choose where to apply your effort for best results.

Optilink avoids doing DNS lookups as a speed improvement, so what it does instead is to match the text of the URL. It attempts to find multiple forms of the target URL, but it is not perfect. For the target url xyz.com, Optilink will test www.xyz.com to see if it is the same IP address and looks like the same page. All internal pages on the target URL will also be loaded and checksumed and compared to the target page so we can usually figure out that www.xyz.com/index.html is actually the same link as xyz.com.
But there are some cases the program can not handle, which is what the 'aliases' list is for. You might have noticed that OptiLink automatically fills in the aliases it does find. You can add aliases as well for those cases where OptiLink would not be able to discover it.

Look at the log view and you will probably see that optilink is failing when parsing the target page. Just select the 'skip on-page measurements' option and rerun -- you will loose the on-page kw %, but that is of little importance in ranking anyway. Also, run your pages through the validator at validator.w3.org to ensure they are error free.

The search with the space between link: is a search for pages including the domain that link anywhere -- not what you want

The combination of link quantity and link quality is the real target, so what I typically try to do is get better quality and the same number of links. I will generally use the link counts reported by yahoo instead of google.

They are both 'right', they just mean two different things. The number returned from Google is not the real link count -- which there is no way to get --but it is a relative measure for comparison between two pages in the rankings. Moreover, since the spider always lags the real web, there will often be pages in the SE's index that no longer link to the target page, but since they are in the index, they do count in the ranking algorithm.

On the other hand, the OptiLink number is the measure right now of the real web, so it could be considered a prospective measure of what the SE index will show once it catches up with changes.

Total links and total pages with links are both important.

Since there may be more than one link per page, this number will be equal to or greater than the number of pages found with (at least one) link.

The reported count is just an estimate so they can provide quick results without actually counting. This is but one example of a heuristic, or short-cut, taken by search engines so that they can deal with huge datasets and still provide rapid response to us users.

The alt text is not counted. This decision is made based on experiments I ran and measurements I made that showed Google in particular was not counting the alt text in the Reputation analysis. Image links are therefore 'empty' of text so the total of words in the numerator of the computation can indeed be smaller than the total of links used as the denominator -- hence less than 100%.
Also, there is one other trick in the numerator that is well founded in observation, but is still open to some argument: a given word repeated in the same link counts as a single occurrance, not multiple occurrances. For example, the link 'blah blah blah' will credit the word 'blah' with a single link, not three.

Seperate. the primary purpose of OptiLink is ranking analysis,
so a comparison of SE-reported links as not a feature, but it is
easy to copy to a spreadsheet or just put multiple OptiLink's
side-by-side

Primarily one of two ways.
1. The search engines are by nature out of date, so it may really be that the search engine is reporting a link that is now gone.
2. OptiLink, to be fast, compares URL strings rather than doing DNS lookups, so it will miss legitimate links where the site is known by two different domains. For example, acws.com and academywebspecialist.com are the same site and treated the same by the engines, but OptiLink will not automatically do so -- you have to manually provide the alias on the command view.
To tell which problem you might be seeing, select the offending page in the pages view and use the right mouse button to look at all the links on the page (view links). From this you should be able to find either an alias for the target page, or verify that the link is in fact missing.

Yes, OptiLink works great for that application. The on-page analysis will be all zeros since the page can not be loaded, but all the linking information will be collected as for non-expired pages.

First, whatever Google has is what it is using to do ranking so that is what we need to use in doing ranking analysis and prediction. where the results are being used for another purpose, you may be concerned about the currency of the data.
But actually, it is much worse than that...
Google does not return all links when doing the link: query (same as backlinks on the toolbar) but instead filters out lower PR pages from the set of linking pages in its database. This incomplete picture introduces a number of complications, but again, for ranking analysis it turns out not to matter since the same filters are in place for all pages.

Yes. That is how OptiLink does the analysis we call Link Reputation. OptiLink must find the actual text use in the link. All the pages returned by the search engine are listed on the Pages view, along with
the number of links to the target page that were actually found and processed -- zero indicates a missing link.

The link command is the only method for getting linking data out of the search engines. Google does not return all linking pages in response to the link: command because it excludes low PR pages. You can expand the list of pages by including the other engines in the query as well since Google will likely have these other pages in its own index too.
If you are not seeing links where you think you should there are four things to do.
First, look at the log view. you might find error messages that account for the problem.
Second, is the page served by more than a single domain name at the same ip? For example, acws.com and academywebspecialists.com are two DNS entries for the same pages. Since OptiLink does not do DNS lookups (too slow) you should list these known aliases on the command view.

Totally impossible to do.

Pages in google does not equal links to your page.

None of the search engines will return more than 1000 results for any query, which is why the OptiLink limit is 1000 as well.

There is _almost_ never a real need to fetch 1000 links. Ranking analysis never requires the analysis of this many links. Reciprocal linking partner discovery is the only case where it makes sense to fetch 1000 links -- see the app secrets guide for comments.

Since the different engines do return results in different orders, even when the underlying link database is shared, fetching 1000 links from all the different search engines should gather more than 1000 links total.

This shows that OptiLink discovered a form of the target URL different than the one entered one the command view. For example, the domain www.nowhere.com is actually a different domain than simply nowhere.com and resolves to the same IP only because of widespread convention. The only way to be sure that these domains are the same is to do an DNS lookup which is very time consuming. Instead of DNS lookups, OptiLink uses a set of heuristics to find "aliases" of the target url. When it succeeds, these are written to the log view as "alternate paths".

Just run OptiLink twice and compare the results side-by-side (you can open any number of copies simultaneously without problem). See Leslie's articles at http://www.windrosesoftware.com/pub for real examples of ranking analysis with OptiLink.