5

Let's have some fun shall we??

At the bottom of the content for about 1/3rd of my pages, but not in the footer, I create an internal link for each related IP address to the page on my site for that IP address. This was fine 6 months ago, but now that my database has grown so large, I discovered that it was not uncommon to create 700+ links within this section with one page having 2600. Ooopppsss!! ;-) Houston, we have a problem!

Now this is easy enough to fix. For right now, I put a limit 100 on the query. But this got me thinking and searching for a definitive answer. So far... no luck.

Please understand I do fully know the effects regarding SEO. Most of you know that already. I am not looking for a lesson- just an idea of a range.

Q: How many internal links per page is too many?

There are a lot of answers on the net culled from various nether regions and I have a number in mind (also culled), but I was thinking that there has to be an authoritative answer somewhere. Google says reasonable and MOZ's answer seemed to very arbitrary to me. Some so-called SEO experts say as little as 6 and as much as 200. Yeah. That is useful! One said there is no limit. I know there is a semi-flexible practical limit. I remember reading about this years ago, but cannot remember the mechanics behind it right now. I would have to think on this for a while.

Bonus Q: If given a choice of creating links from a random query (and yes I can do this) or first 100 or so (using limit), which would you prefer/advise and why? Again, I have an idea in mind (again, culled). Is there anything definitive in this regard as well?

My primary concerns are UX, download speed, and lastly, Giigle. (<= please do not edit to change)

  • 1
    What are the IP address pages for? Is it some sort of honeypot detection service? – Anonymous May 13 '15 at 0:47
  • 1
    @Anonymous No. These are content pages either domain name or IP addresses. In this case, just IP addresses since domain names do not present a problem with volume. The links are IP addresses linking to it's content page. These are mostly hack attempts, spammers, stealth bots, and so on. – closetnoc May 13 '15 at 0:50
  • 1
  • Hey what is meaning of "Giigle"? – I am the Most Stupid Person Jan 16 '18 at 4:03
  • 1
    @IamtheMostStupidPerson That was my attempt at humor meaning Google. BTW, I like that you ask questions. It makes me happy! Questions are good. Cheers!! – closetnoc Jan 16 '18 at 4:25
4

Here is a visualization of the number of links on the homepages of top 98 webpages. Very few have less than 100, and many have 500 or more.

Google used to recommend that any page have no more than 100 links per page. However, they removed the "100 links per page" from the webmaster guidelines some time ago. Matt Cutts released a video where he says that the limits for page size and number of links per page are much higher than they used to be.

I've personally worked on a website with 250-400 links on every page. That site did very well in search engines.

I'm not sure what the upper limit today actually is. I'd keep it at no more than 500 if I were you. Above that and I'd think your site would be an outlier compared to other websites.


Unrelated to the number of links per page, your website about IP addresses is likely going to have other ranking problems if you have a page for every IP address. That is a lot of pages and most of them are not going to have much content. You probably only want to submit the pages with the most content to search engines.

I remember that Google was trying to remove phone number sites from its index a few years ago. Partly that was because so many of them had "Be the first to say something about this phone number" style blank pages.

  • 1
    I appreciate your answer as always. These pages are offenders of my website. They are hack attempts, spammers, bad bots, stealth bots, and so on. There is enough on each page with Whois, APNIC, DNS, GEOIP, full IP block data, how to block using... code/instructions, related domain names, related IP addresses, user agents, coming soon- stability scores both IP and domain (using actual SOP algo- not culled), actual attack and what vulnerability sought including CVE record, and so on. So there is something on each page and some are likely too long according to SEO Power Suite. – closetnoc May 13 '15 at 1:38
  • 1
    I also liked the graphic you linked to. HOLY COW!! Now I don't feel like such as dumb @$$! 6 months ago, the numbers approached 40 at the most. How did it get to 2600?? I rather suspect the new stealth bot mechanism has something to do with it. -- If you had your druthers, would you use a limit 100 which returns the first 100 of a result set or go random?? – closetnoc May 13 '15 at 1:52
  • 1
    If you have pages only the problematic IP addresses (as opposed to all IP addresses) you should be fine. – Stephen Ostermiller May 13 '15 at 11:02
  • I have just about 700,000 domain names and IP addresses now. I let the system do what it does and it seems to only wander off the reservation a bit for IP addresses that host spam domains since these change very often. The latest explosion seems to be stealth bots which are institutional in some cases. There are some networks that are famous for hosting bad bots such as AWS, ENom, and so on. Some subscriber networks can also be pervasive with bots though this is less likely these days. It is the growth of the site that I am having to compensate for with code fixes the most. – closetnoc May 13 '15 at 15:20
  • I would like to accepted all answers of course... or up-vote twice at least for all the effort and input. But in the end, I had to make a decision. The one outstanding feature of your answer was the metric-graphic. I was looking for some authority on the subject, but I rather suspected there was not one. The graphic came as close to a factual authoritative answer as anyone could have expected. In the end, I think we are just guessing on the subject and that Google only get curious after a certain point. Otherwise, it is just a matter of practicality. – closetnoc May 13 '15 at 15:26
2

Ideal would probably be 50. 100 works too if you want to go on the higher side.

Taking your points one at a time:

  • UX: Unless I misunderstood how this website is to be used, a search feature would be much more useful than a list of IP addresses. That way, people can look up the domain or IP of any websites that they have had issues with. However, maybe if someone is simply looking for spam IP addresses, this could be useful to them. In that case, it may be more advantageous to simply have a page listing the IP addresses and domains in a table with different "pages" for each category (hack attempts, spammers, and stealth bots). That page could be linked to instead. If you still wanted to stick with the links on the bottom, I would say random works best since seeing only a small, static section every time when visiting the site becomes old. If you were to take the random approach, a maximum of 50 would be needed I would say.

  • Download Speed: I honestly doubt having 50-100 links on a page would affect it barely at all...

  • Google: As with everything, Google seems to answer that it should be optimized for users more than for search engines. They also say that 100 is suggested, but it is not a hard limit. Matt Cutts even says himself that:

    These days, Google will index more than 100K of a page, but there’s still a good reason to recommend keeping to under a hundred links or so: the user experience. If you’re showing well over 100 links per page, you could be overwhelming your users and giving them a bad experience. A page might look good to you until you put on your “user hat” and see what it looks like to a new visitor.

  • I like your point about search in particular. I had a Google SE on the site until I upgraded the templates to my new responsive theme (not a CMS). Prior to that, no one ever used search. I will be putting it back, but I have a technical hurdle to take care of first for which there is no definitive answer. I was not so worried about Google. The site performed well enough given how poor it was, but now a lot of the details are being fixed. I just did an audit and the link numbers are huge. BTW- I have approached over 700,000 domain names and IP addresses with many millions of attacks. Sheeesh! – closetnoc May 13 '15 at 1:46
  • I originally put these links in for users but I am not sure people even notice them. Back about 6 months ago, it made sense since the high number was about 40. I have a new stealth bot mechanism I blame for the jump in related IP addresses. -- If you had your druthers, would you use a limit 100 which returns the first 100 of a result set or go random?? – closetnoc May 13 '15 at 1:54
  • Probably random, it would seem to make more sense than just choosing 100 specific ones to show at all times. – Anonymous May 13 '15 at 2:03
  • Okay. Thanks! I was holding back on my opinion but thought I would let it out eventually. I was thinking either 100, 150, but no more than 200 links and random was best for exactly the same reason you gave. Interesting factoid (if you are a geek otherwise it would be rather dull): I took the byte count of a link and calculated how many bytes 750 and 2600 links would be. Answer: 35.15625 and 121.875 assuming my calculations are right. – closetnoc May 13 '15 at 2:14
  • Ooopppsss! The answers are in k not bytes. Silly silly me!! – closetnoc May 13 '15 at 2:21
1

My primary concerns are UX, download speed, and lastly, Gaggle.

oops... I meant giigle, or maybe googoogaagaagle.

Ok, back to the real issues.

To me, I think having a ridiculously huge number of links on a single page may cause a slight slowdown because when a robot scans a page, every single byte is downloaded from the server, and each minimally-crafted link requires 16 bytes with anchor text excluded. For example:

<a href=" "></a>

If you have 500 links, then that's 8 KB used just for code. You need to try to have a good code/text ratio on your website so search engines think you have content instead of black-hat techniques going on in the site. This means if you went with the 500 links, then you need to come up with almost 2000 characters of text users can see without adding additional HTML code just to meet 15% code/text ratio.

So to me personally, there is no magic number on the actual number of links one should limit themselves to, but if I have to define an absolute minimum, I'd go with two. One for a privacy policy page, and the other for the "next section" page.

The more links you have along with other HTML tags you have on the site, the more bytes one must download.

Bonus Q: If given a choice of creating links from a random query (and yes I can do this) or first 100 or so (using limit), which would you prefer/advise and why? Again, I have an idea in mind (again, culled). Is there anything definitive in this regard as well?

You're better off to go linear and take the first 100 in 99% of the cases because reading the data is straightforward, and the computer doesn't have to use extra clock cycles in figuring out random numbers and piecing random rows of data together. If you must go random, then at least do all database operations in memory without swapping kicking in.

  • I knew you would not disappoint me! Yeah. I knew I had to limit the number of links. One option is just to remove this section. I am not sure it really helps anyone anymore. Most people would go cross-eyed before getting to the bottom. BTW- What is google + google? goooogle. How does Google give the finger? G00100gle. (binary four- count it on your hand) Already my pages are being quickly filled with content so I might consider removing this section in time. I will only keep it for search engines to chew on and SERP impressions. I am concerned about page size somewhat now. Some are huge. – closetnoc May 13 '15 at 3:12
  • BTW- I am not sure you saw the byte/kb calculations I did in another comment. It is significant. – closetnoc May 13 '15 at 3:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.