Negative & Positive Link Spikes – Tripping Anchor Text Filters

The term ‘link spikes’ has been around a long time, usually to describe unnatural link patterns that Google uses to put the smack down on overly aggressive link builders.

Using the term in this way is correct and yes, creating crazy link spikes in your link profile will put you in trouble.

What people don’t realise is that it isn’t just positive link spikes that can cause issues but also negative ones. Spikes from lost links can be as damaging as spikes from link growth.

Link Spikes

Of course losing links can mean lost rankings, however in my experience lost links can also trip filters and mean a specific page on your site will just never rank, no matter what you do.

Negative Link Spike Case Study

I had the unfortunate opportunity of experiencing a filter on an affiliate site of mine. After renting a few links for 6 months I decided to pull them down, knowing I had plenty in reserve to keep my rankings.

These links were site wide and all had the same anchor text, when they were removed I lost 52,000 links from 3 domains. Technically this should have been a good thing as the links were completely unnatural and obviously paid.

Link Spike Link Loss

Two days after the removal of these links my site dropped rankings, not just for the target keyword but for any other search terms containg the target term. Elsewhere organic traffic was growing, and growing well, but this keyword and variations weren’t even in the top 100, you can clearly see the effect on traffic below for the page where the links were pointing.

traffic link spike

The first thing I did was try and replace the lost links, I didn’t think this would work but wanted to see what happened, the answer? Rankings dropped even further. Now I had created 2 unnatural link spikes in a row and new my rankings would never recover.

Sometimes when you begin working on a new site you may come across ugly site wide links that you think need cleaning up, however do this with extreme caution, negative link spikes are as dangerous as positive ones.

22 comments on “Negative & Positive Link Spikes – Tripping Anchor Text Filters

  1. Wow, great insight. If you took off the sitewides only 1 site at a time (instead of all 3 at the same time), do you think the effects would have been the same?

  2. A few months ago I had some experience with negative linking patterns… I agree that you definitely need to be careful when removing links to your site!

  3. I would be careful with using the word “never” because it seems that your subpage for example would never be ranking doesn’t matter what you do in the future. I don’t think so. If you build natural links time to time, the site will definately rank in the near future. But when is the question and that depends on the filter. But no filter is for ever, right?

  4. Maybe not, my recommendation would be to change the anchor text of the site to brand, one at a time then access the impact. They’re less likely to be considered paid and get devalued if they have a more natural anchor text.

  5. Yeah it’s a nightmare. We have inherited a few clients from other companies, once they’re contract ends the company removes all the links, can be a headache to fix, SEO companies need to be 100% transparent when renting links.

  6. Hi Markus,

    You’re right ‘never’ was just a general statement, ‘very difficult’ would have been more appropriate. I have known these filters last anywhere between 3 months and 2 years, even with a natural ‘brand’ link building program going on in the background.

  7. Hi Tim
    I have experienced similar situations on 2 of my sites. Thanks for verifying this..
    A site of mine getting 20K uv a day went down the drain because of this..
    Anyways wanted to ask have you ever been able to recover from such situations.. is it only constant and slow link building along with lots of patience that will cure the falling rankings??
    Thankyou so much for sharing this report :)

    Regards
    Ishan

  8. also is it only the negative and positive link spikes triggers or even the anchor text matters..as far as Google filters are concerned ?
    Like loads of links with different anchor text wont hurt but loads of links with exact anchor text will put us in trouble.
    Hope I made sense :)

  9. Hi Ishan,

    Yes we have managed to recover a few sites, on all occasions we have carried out aggressive in content link building campaigns but using only brand anchor text for the first 3 months at least. So lots of in content links, all brand anchor text and noise links.

  10. My rankings dropped after a link surge, I have now removed two sitewide links, well changed them to homepage links instead and waiting to see what happens. I have started building contextual links with http://www.domain.com as the link text (as they site name is an EMD for the main keyword.

    Looks like you have to wait between 3 months and 2 years to reappear? That is longer than for a full on penalty?

  11. hi Tim,

    I got a question – I have a online shop with over 30 keywords in top1-10 and one main keyword is overlinked – dropped from top2 to top15 – how can I fix it ?
    should I link to home page with “domain.com” anchor ?
    why it should help in removing filter from one of main keywords ?
    or maybe I get it wrong ;)

  12. Hi Jakub,

    Filters are tricky and there is no definite time scale or technique for removing them, however we find it best practice to start an aggressive in content link building campaign using as many different anchor text variations as you see fit, usually around 10. You also need to include brand based and noise terms, e.g. domain name + variations, and click here etc…

    It works by diluting the impact of the over targeted links, we believe a filter is usually based on a threshold and resolving it is about addressing the balance.

  13. Hi Tim,

    My site has come back slightly, two keywords are at the bottom of page one now. I have checked my link profile again and it still seems odd that a filter is in place. This is because the highest number of anchor texts is the name of the site, which is also the main keyword.

    I got the domain initially as it had the two main keywords in it, but it is hyphenated like this..

    http://www.keyword-keyword.org or a real example might be http://www.payday-loans.org

    Do you think this still counts as an EMD as I have done loads of directory submissions using the keywords as the anchor, or is it that Google does not count this as a brand link. I know overall hyphenated domains are worth less, but do you think they would impact on a filter?

  14. Hi Simon,

    Hyphenated domains are closely linked with spam, so it is possible for this to be the issue, especially if you have built hundreds of links to the exact keyword. You have to remember, your link tool will not be showing all of your links, there will no doubt be hundreds of low cost directories that Google have found that your link tool can’t.

    Try building some domain match anchor text links into the profile http://www.keyword-keyword.com , and use at least 10 variations of the target term. Trying to get out of a filter is tough, you have to just keep going building a natural looking link profile as possible. There are no short cuts, buying up 1000 directory submissions will do you know good, keep everything in content. It won’t all be counted but it won’t make things worse either.

  15. Thanks for the advice Tim. I am just wondering, what would happen is someone pointed hundreds of spammy links at your site, tripped an anchor text filter and you dropped for your main keyterms. Filters are automated right? So would a reincusion request even work?

  16. Good question.

    Technically, competitors aren’t able to sabotage your website, however any crap links will eventually be devalued by Google.

    I think if you rank highly, and a competitor comes along and spams your profile you won’t get hurt. However, if your high rankings are based on spam links whether built by you or a competitor then these will eventually be devalued and you will drop. Going aggressively after the anchor text with low value links won’t work, as well as they did before the devaluation took place, you will need to build your profile from scratch and add diversity to your profile until the signals from the stronger links out weigh the bad ones.

    I know it’s confusing, the best way to look at is like this, rank your site smarter and there is little your competitors can do to bring you down, and lets face it a competitor isn’t going to spam you until you are in a great position in Google.

  17. I read an old SEOMoz post about over optimising internal links and that this can also cause a filter?

    http://www.seomoz.org/blog/testing-the-value-of-anchor-text-optimized-internal-links

    Instead of using the word ‘home’ for the home button in the nav menu I was using KEYWORD. This would have created thousands of internal links for this keyword. I have taken this off now.

    Also I would say 95% of posts link to the homepage with the main keyword contextually as I have been using a wordpress plugin called text replace. I am considering taking this off as well? Sometimes there are 2 links in each post for the same keyword pointing to the homepage.

  18. Since making the above change the rankings have come back quite a lot, although still not first page, jumped about 20 places. I have seen another big branded site come under keyword filter recently. Each blog post has about 3/4 internal links, most of which are keyword rich.

    I am almost certain that you can be filtered for over-linking internally or using too much rich anchor text internally.

Comments are closed.

1 Pings/Trackbacks for "Negative & Positive Link Spikes – Tripping Anchor Text Filters"