The True Potential of Anchor Links (#)

Inspired by this cool experiment on how google ignores the first anchor text by Errioxa I decided to set up an experiment of my own, to further look into the issue of how google treats links that include anchors (e.g. site.com/index.html#paragraph2). I will continue calling them links with #, so one doesn’t confuse the linking anchor text with the #-text anchor within the content of the site.

This was actually very relevant to me from a practical standpoint, let me explain:

We had redesigned our website to feature the different games we offer in different tabs (sorry, it’s in German, but I think you’ll understand). The idea was to be able to direct people directly to their favorite game and show them the relevant information, i.e. either the tab for Schnapsen or Bauernschnapsen, kind of like different landing pages. Obviously it was relevant, whether these links that included # would pass juice or not- otherwise the whole thing could have been a bit counterproductive from a SEO-perspective:

Either someone interested in one game might be directed to the “wrong” game, or the #-link directing to the relevant game would not pass any value. Another practical example would be the Doppelkopf rules section, that we would sometimes link to, when people need more information. Actually, this should apply for any link that includes an #.

So I started experimenting a bit: A couple of months ago, fabioricotta showed that he could use canonical tags to get more than one anchor text valued by google. My approach is similar but might be a bit easier to implement, I believe. I started my experiment a while before I read about his findings.

I put up four links on a page to another page on another domain in the following way:

• Anchor Text –> URL linked to

1. DCBAwertfkgjdkrelsdgjsdfk –> test.html#ABCDvWVRAWVC
2. HGFEasmcisdlmOsdmsnvklefs –> test.html#EFGHaverawvtaevraervae
3. LKJIsacneksaenlksdcknscdl –> test.html#IJKLsdfsfvssdvavvaasva
4. AKJFassdfsdffskkascsjcdsd –> test.html

None of the anchor texts or the letters behind the hash tags were on the page or domain. There weren’t any in text anchors set within the site.

What happened was that I could find the site using any of the anchor texts which had # in the URL, just not the fourth one that merely pointed to test.html:

Conclusion:

1. If on a site there is a #-link and a regular link pointing to the same site, google ignores the link that doesn’t include the #.
2. If there is more than one #-link to another site, all the anchor texts of these links are attributed to that site.
3. The letters behind the “#” (e.g. #paragraph2) are ignored completely, at least when they aren’t set on the page.

In Errioxas article the first link was a regular one (without a #), and it was ignored, which is in accordance with my findings. In this recent article by Zen2Seo, however, google does not ignore the regular link, although there is a #-link as well. This contradicts my findings, maybe it has to do with the fact that there was only one #-link. Any ideas?

Knowing I could attribute several linking anchor texts from a page, I went one step further: I tried to evaluate how powerful the links were and whether I could influence how much juice is passed to any given page.

Note: While I am fairly certain that part 1 works, this is really a first experiment. In order to actually prove this, tests on more sites/domains would be necessary.

I put two links with the same link text to different new pages on the same domain.

• 231HGFEasmcisdlmOsdmsnvklefs –> page1.html#EFGHaverawvtaevraervae
• 231HGFEasmcisdlmOsdmsnvklefs –> page2.html#IJKLsdfsfvssdvavvaasva

Both pages ranked for the anchor Text in google, I had sort of expected that. What was a bit strange was that page2 that had the second link pointing to it, ranked higher than the one that the first link pointed to. (Again, both pages were new, on the same domain and did not have anchor text or letters behind the hash tag on them).

So I tried to change the weight of the links a bit, by adding a couple more to the first page, using the same anchor text but a different part after the hash tag, thus “simulating” links to different pages:

• 231HGFEasmcisdlmOsdmsnvklefs –> page1.html#EFGHaverawvtaevraervae
• 231HGFEasmcisdlmOsdmsnvklefs –> page1.html##IJKLsdfsfvssdvavvaasva
• 231HGFEasmcisdlmOsdmsnvklefs –> page1.html#IJKLsdfkzukhkjhkjgjghj
• 231HGFEasmcisdlmOsdmsnvklefs –> page2.html#IJKLsdfsfvssdvavvaasva
• DCBAwertfkgjdkrelsdgjsdfk –> test.html#ABCDvWVRAWVCASDACSS
• 231HGFEasmcisdlmOsdmsnvklefs –> franz.html#IJKLsdfsfvssdvavvaasva
• 231HGFEasmcisdlmOsdmsnvklefs –> page1.html#IJKLsjdtghhdfvavvaasva

Result: Nothing. Although four links now pointed to page 1, and just one to page 2, page 2 still ranked higher.

I did this months ago, even kept adding more links to page 1 over time, but could never get the ranking of the pages to change. To be very honest, this was a bit disappointing. I must admit though, juice which flows to this test domain was very low, maybe trying this over a number of domains would bring different results. Or maybe one of you guys finds an error in how I conducted this experiment, I wanted to share these results in any case.

Conclusions and potential real-life use of this experiment:

1. Most obviously: Not wasting or wondering about wasting juice, because you aren’t sure which links count.
2. The long Tail: I have read this somewhere in the comments of the other articles and it’s pretty obvious: You could now attribute several long tail comments from one blog post. This might be interesting for niche, very long tail keywords. I don’t know how useful this is, by itself, though. Especially since it might be extremely irritating from a usability standpoint to see five links to the same page. Not very useful or to be recommended, I guess.
3. Domain diversity: This could be a bit more interesting. As we all know or just learned from Rand’s quiz, it’s usually better to have 10 links from 10 different domains than 10 links from one domain. So instead of having five domains link to my site using keyword A and five domains using keyword B, it would probably be better to have 10 links with both keywords, right? At least considering domain diversity.
4. More juice: Let’s assume the juice passed by a link to a given page roughly equals “links to this page” divided by “total external links from the linking page”. Assuming that this article has about 15 external links (data from mozbar for a random youmoz article), being able to have three links instead of one link to my page in it could transfer maybe 20% of the juice to my site, instead of less than 7%.

One might argue, that putting a higher number of links on a page as suggested in 3. and 4. is probably not an option in most cases. I agree, it probably isn’t and shouldn’t be, I certainly don’t want to suggest any kind of manipulation or behavior, that hurts the user experience or is spammy in any way. But when you have a site structure like in the image below, these links might actually be useful to your users and make sense to put in- in addition to transferring lots of juice.

Source: Seomoz

Or I could imagine a widget that has some sort of tabs, in our case maybe showing player statistics for the two different games mentioned at the very beginning of this article, linking back to both games from one page.

Although these aren’t straight forward suggestions on what you should do, I hope these findings help some you with their SEO efforts and maybe lead to further investigation of this issue. I hope the article wasn’t too complicated, I will try to answer any questions you guys might have.

Thanks for reading!