Removing disavow file test = 37.31% increase in organic traffic

You may remember, but following a number of recommendations at Pubcon 2016, I decided to run an experiment to see what would happen if I removed a disavow file from a previously penalised domain:

At the time, it seemed to me that there were a lot of opinions on this, but no-one had actually tried it – or at least they hadn’t publicly shared the results…

So that was my goal, to run a controlled experiment where everyone could learn from the findings.

I picked a website I’ve had for a long time:

  • The domain had previously received a Google penguin penalty in May 2012
  • This recovered in October 2013 following disavowing links and reconsideration requests
  • The site generated a reasonable amount of traffic ~40,000 organic visits per month
  • And from the penalty removal, still had 977 domains actively disavowed

Early learnings:

Digging deeper into this, we found that because a lot of the links were historical, 28% of the links were now dead, with a further 3% as domains listed for sale – that meant that 69% were still active (674 unique domains in total):

Of course, some of the link penalties could have expired too (after-all, the penalty had been removed for 3 years by this point).

In which case because of the time period involved, that would indicate that perhaps the disavow was now unnecessary.

The early results saw positive, but inconclusive signs…

After 1 week, this showed that the average position had dropped slightly, then increased slightly – but nothing out of the ordinary and overall across the week has dropped from an average of 10.2 to 10.4.

In addition to this, the daily clicks has increased by 1,982 to 2,181:

disavow-update

(Google Search Console report)

After 2 weeks, the average position of rankings has dropped after my last update, but quickly rose back up – from an initial starting point of 10.2 (13th October), the average position dropped to a low of 11.6 (19th October) and then rose to a high of 9.8 (25th Oct) before returning to 10.2 (26th Oct):

disavow-update

I kept a close eye on this throughout, but the early conclusion was potentially that these are positive signs, but overall, it was too early and inconclusive at this stage to be recommending that removing your disavow file is a sensible move.

Fast forward to month 4… and results still seemed relatively static in Search Console:

In which case it’s becoming safer to say that perhaps removing the disavow file had no impact.

There’s still other factors to consider, like the age/potential expiry of links, but in this particular case it’s starting to look like removing the disavow wasn’t as crazy an idea as it may have initially sounded.

Month 5 organic traffic has increased by 37.31%!

Having now given this 5 months since the disavow file was removed, I decided to take another look back at progress.

I want to make it clear that no other activity has taken place on this site, in terms of on-site optimisation, publishing new content, PPC/other channels, or the active addition (or removal) of links…

Obviously there may have been other factors at play, Google’s algorithm updates being the big one, or competitor activity. But of the factors in our control, the disavow file removal was the only thing that changed.

Firstly looking at Google Search Console data for the last 90 days vs the start of the test, you can see that;

  1. The average position is now up to 8.3, from 10.3 (which is a -19.4% improvement, if that makes sense!)
  2. Average CTR has increased from 6.75% to 8.26% (22.3% increase)

Both of these are significant performance improvements – but what does this mean to traffic?

Here’s where we are today (highlighting the week before the disavow file removal):

Looking back at this, although I didn’t want to get too carried away with an early uplift – it’s pretty clear to see that the impact to organic traffic was fairly immediate and it has been sustained since.

The peak of the traffic in mid Feb, saw 16,151 sessions (44.7% increase vs the Oct benchmark) – and last week saw 13,911 sessions (24.6% increase).

As you can see, seasonality is very flat – and when viewed as a year-on-year comparison, you can see that there has been a 37.31% increase in organic traffic:

Why did rankings/traffic increase?

I think the biggest impact from this would be that within the disavow file, there are likely some positive links as well as negative. So in removing the disavow, bear in mind that a large number of links have been turned back on.

When disavowing, you want to get out of the penalty and maybe need to disavow a wider net of links to get this removed. But in time, and once out of the penalty, it’s interesting to see that in this case the net effect is a positive gain.

I would still be cautious in removing your disavow file completely, especially if coming off the back of a recent penalty, or freshly disavowed links – but based on this we’ve certainly seen good reason to re-visit which domains are disavowed and looking to at least prune down the list to a limited number in stages.

Although this was quite an extreme test to remove the disavow in one go – overall, I would class this as a very positive result!

Kevin Gibbons is UK managing director of BlueGlass - he has been involved in digital marketing since 2003 and frequently speaks/writes at industry leading events and publications. Connect on LinkedIn and Twitter.

by Kevin Gibbons
  • Wes Bradley

    Interesting case study. Even though I agree with a lot of the comments here that it’s not enough to draw solid conclusions, I enjoy reading about these ‘anomalies’ or ‘edge cases’. Makes for a much more interesting read rather than your standard SEO dribble these days.

    They also give me good inspiration to think outside the box on things to try with SEO – once you scratch the surface there really is a plethora of activities you can do to increase your rankings. I’d be interested to hear of any other unique strategies you’ve tried and had good results with?

    • Kevin Gibbons

      Thanks Wes – always be testing 🙂

      I’ve got a couple of others I’m running at the moment, so plan on sharing once there’s some interesting results to share…

  • David

    hey Kevin, this is a great case study. i think there are quite a lot of learning points in this case which may get consolidated by running the same or similar experiment with other clients.

    In my humble experience of a large shot of negative SEO attack had in some of the sites in the network I manage inhouse, I wonder if:

    1. The no of additional backlinks gained in the post penguin-driven 1st disavow (years ago) and now may have allowed your client to earn additional referrals and backlinks. This in turn could have been increasing so the overall total count of backlinks/referring domains for that site, therefore making the percentage of those (theoretical) toxic ones in the disavow file a much lower number, not enough to warrant a penalty any more. In my experience unless a site has a certain percentage of toxic links, it doesn’t get under the filter or penalty for the matter.

    2. the fact that current Penguin algorithm being more granular, the previous (perhaps unfair) penalty could have been lifted as the genuine penalisation was on a fraction, segment of the site?

    3. perhaps some of the disavowed links didn’t quite need/merit the radical decision of disavowing and goodies fell into the disavow file as baddies, or perhaps some of the offending sites with toxic links have got better in the last few years, solve their spam issues, stopped being hacked, done disavows on their end, etc… (TOX2, SUSP 4 and SUSP 5 in LRT kind of thing).

    I think it would be really hard to conclude on just one single variable being the reason for the uplift in traffic based on knocking off the disavow, but all in all, is a very interesting case study. I would rewrite it and post it in Moz, or at the least in Medium.

    thanks again and all the best 😉

    • Kevin Gibbons

      Thanks David!

      1) In this case, we didn’t do any other marketing activity, so the link profile should have remained very similar – but I agree overall that the % mix is really important.

      2) I remember a Google engineer saying a while back, that they either class links as negative, positive or neutral – meaning that some have an impact (either way), but a large % are actually ignored – I think that’s the case here… At the time they wanted to make an example of sites with negative links, but now the penalty is removed + expiration of time, the positive is outweighing the negative

      3) I think that’s fair – in order to get out of a penalty, sometimes it’s more about showing effort to remove/disavow – which means some good links can end up in there too. So remove the disavow and it’s turned back on some positive signals.

  • Frushi

    Google said that they are able to not count bad links instead hurt a website for keeping that links point to it. So maybe in your disavow were good links which may impact positively and those bad links were just not count into your link profile. So that’s why you got up.

    • Kevin Gibbons

      I think that’s exactly what happened – in combination with some of the bad links disappearing/penalty expiring…

  • Interesting Kevin, thanks for sharing!

  • The check I would do before (or even now if you kept the file would be to check how many websites from your disavow file are still active, how many converted the links to nofollow etc – the test as it is does’t bring any safe conclusions.

    Also on the figures, looks like you had some gains aligned with the Phantom 5 update in February as well – which should be irrelevant to the disavow file.

    Lot’s of data is missing as well from the article – e.g. content you added, if you switched on AMP etc that could all contribute to the gains and might make the removal of the disavow file irrelevant

    • Scott Milsom

      “I want to make it clear that no other activity has taken place on this site, in terms of on-site optimisation, publishing new content, PPC/other channels, or the active addition (or removal) of links…”

      • How about revealing the website then? 🙂

        • Kevin Gibbons

          Thanks both – we did check the active links, there’s a pie chart in the post to show that 28% are dead, with a further 3% domains listed for sale.

          There could definitely have been external factors at play, although the impact to organic traffic was fairly immediate – which indicates to me that this was caused by essentially turning back on previously disavowed links, which are now having a positive impact.

          This is an interesting one in itself – someone Google penalty removals = rankings/traffic going back to exactly where they were before the penalty, which indicates that the links didn’t have an impact in the first place. In other situations, once the penalty is lifted organic performance improves, but not to the previous level – in which case those links likely did “work” before – and once removed from the disavow they have impact again…

          Revealing the website – if I did that people would say Google know about the test and are messing with us 😉



45 Leather Lane
London
EC1N 7TJ
UK