Link Removal Case Study

[FranticFish]

Just wanted to share something that I found interesting.

Last year, a UK SME site I work on suffered severe drops in rankings for its trophy term (household niche, non-geographic term, fell about 7-10 pages from top of page 2 / bottom of page 1).

At the time I was slowly building new links to the site (all on the brand name) and adding new content to the site. These new links are medium quality - not stellar, but OK. Small numbers, carefully chosen, less than 20 domains in all.

When I started my work the site had not had any links to it for years. It had very few genuine or reputable links, about 5 domains. It did also have some crappy links from about 40 SEO directories that dated back to 2003/4. These had optimised anchor text mostly, and about 4 different phrases used. The trophy term mentioned above was used in less than 15% of those links (it is also in the title tag of the home page).

It's always seemed to me (at least until very recently) that Google had a sort of amnesty of sorts WRT to old links, so I hadn't bothered to get these removed. That was a real mistake as it turns out.

The client's in-house SEO worked very hard to get these all removed, and at the start of this year daily traffic doubled, because the client is now top 2/3 for their main term across the UK i.e. jumped back to where they were before and then up another page.

Content on site remained completely untouched throughout this period. No PR and no link building done either.

There's a lot of things I found very interesting:
- such small numbers of bad links, only slightly more in number than the reputable links
- bad links had existed untouched for 10 years
- only 15% of bad links contained trophy term, so ONLY 5 OR 6 LINKS was al it took in this instance
- WMT and Analytics continued to show 'unfiltered' ranking for the trophy term throughout entire period

So, if you haven't cleaned up your link profile because you haven't been penalised, those links could still be holding you back.

Long-standing advice has often been 'build new quality to outweigh old crud'. Perhaps that's true if you can build stellar new links.

But the next time I start work on a new site, I will remove EVERY LAST crud link.

[martinibuster]

I've been wondering about this phenomenom. I've heard similar stories. That a site with old crap links can chug along ranking like normal. But if they start a link building initiative all of a sudden those old links become a liability. As if... the link building activity triggered renewed scrutiny of those old links.

[Webwork]

What "before and after data" do you have on the other sites in the relevant SERPs - such as changes in their link profiles or content or architecture AND/OR changes in the ranking of the surrounding sites in the SERPs - to support your conclusion or inference that your actions were causative?

[FranticFish]

@MB: I hadn't considered that, but it's an interesting thought.

Both the fall from grace and the return were overnight affairs.

The good new links, which are now all the remain, were built over a period of 4 months. No improvement was noted whilst they were being added.

Not sure what to take from the timing.

It's established that there are refresh dates for Penguin, so perhaps there are periodic audits for questionable links too. Even if the link graph is monitored constantly, updates of a certain type are scheduled.

OR... perhaps there is a very tight threshold and Google had to be satisfied that every single bad link had gone.

OR... Google doesn't crawl the sort of sites these links are on that often, so the changes that occurred gradually were picked up all at once.

[FranticFish]

@Webwork: this is a fairly stable SERP at the top (far less so further down), and today's SERP is practically identical to one from 1/2/3/4/5 months ago, with a Wiki result, 2 UK household names and the established players in that market. There's always a 'new kid on the block' thrown in too but that changes every week/fortnight/month.

I recognise almost site there - except for the client.

Remember, we're talking a 8-10 page jump here, so IMO it's too much of a stretch to think that almost every other site on page 1 also had some radical reassessment.

[Planet13]

@ FranticFish:

Lots of questions for you, my friend.

"At the time I was slowly building new links to the site... The good new links, which are now all the remain, were built over a period of 4 months."

How exactly were these links "built" and why do you assess them as "good" when compared to the previous links that were removed?

"Both the fall from grace and the return were overnight affairs. "

Would you mind terribly giving us those dates for the drop and the resurfacing?

[martinibuster]

How exactly were these links "built" and why do you assess them as "good"...

Perhaps it's arguable, but it seems it's self-evident that the links are good because they haven't triggered an unnatural links warning.

[FranticFish]

@ Planet13 - sorry for slow response, not been online for a while.

The new links were the best I could find of what I call 'low hanging fruit': some were pay for a review web directories, some were niche portals, some were household finder/media sites, some local directories.

All of them were found after going through competitors' IBL profiles. In my view they were the sites that we should approach first, so I emailed / called to ask if they'd feature my client: nearly all of the time money changed hands.

I consider them good because the sites looked like quality. Some old, some new, but OK to good content. Like I said, nothing stellar, but not poor quality either. I didn't use any metrics to judge them other than checking that the page we'd end up on was indexed. I do use metrics now and most of them have fairly good ones according to my data sources.

Drop date is hard to pinpoint as I didn't watch rankings that closely; I was more concerned with traffic, and as the trophy term was on page 2 it drove very, very little overall compared to the rest of the site. I was only monitoring the term as a phrase/broad match term.

Estimated date (from looking back at number of referrals on the exact term and WMT data) was mid January 2013. Traffic was unaffected - in fact it increased overall. Only the 'exact match' term was affected; the group term continued to show up in WMT data.

Their in-house SEO decided to remove the old links in August; it took a long time to get them all deleted. Traffic improved overnight from 1st/2nd January 2014.

One other thing that might be relevant: the client is a brand in their niche, and do get searched on (6.5% of impressions and 8% of visits last month are on brand name).

[Shai]

I've been wondering about this phenomenom. I've heard similar stories. That a site with old crap links can chug along ranking like normal. But if they start a link building initiative all of a sudden those old links become a liability. As if... the link building activity triggered renewed scrutiny of those old links.

We have seen his numerous times and the way we explain it is that its the traffic increase that the link building causes that triggers the extra checks and therefore the penalty. We then carry out a clean up of those old links, and the penalty is removed and traffic is back to normal.

[londrum]

That a site with old crap links can chug along ranking like normal. But if they start a link building initiative all of a sudden those old links become a liability. As if... the link building activity triggered renewed scrutiny of those old links.

maybe its the link building itself that gave you the penalty.

if you've got an established site with some links to it (good or bad, it doesnt matter), google will know the usual pattern for link building. one new link every month, or whatever. if you suddenly come along and obtain a load of new links to it, even if they are good ones, google might recognise that as being part of a link building campaign and give you a penalty. thats what you got demoted for.

you then got rid of some of the old bad links, which gave you a boost again. but that was a totally separate solution, to a different problem.