For businesses like ours which derive most of their traffic through organic means, Google is like GOD and we are always on its mercy. Everyday we live under a constant fear of what might happen the next day; is our traffic in place? And I am sure that even those people who have got all the things right would agree to that since Google is constantly changing the parameters. We at its mercy can only make things correct (as per our understanding) and hope that we would not lose but win.
2013 has been very tumultuous for us. We got hit pretty badly on couple of occasions to the tune of around 40% last year. Here are some traffic graphs.
Above traffic chart is from May. Penguin 2.0 came on 22 May and graphs tells everything. Wait the misery doesn’t end here. Penguin graced us all with its presence once more i.e. in October (4 October). For Google algorithm updates, refer here. Here is the traffic chart from October.
So overall we lost around 40% traffic. Now our task was cut out. If we wanted to bring business back on track, we had to deal with this monster called Penguin (which till now I used to like).
Penguin is concerned about your link profile. It tries to judge you on what sort of company (links) do you keep. Therefore one thing was pretty clear – Our company was terribly bad since GOD punished us and GOD would not do anything wrong. But we were confused about one thing. You see there are 2 ways of improving the kind of company you have -
- Get reliable people in your company i.e. get links from good relevant domains
- Get rid of bad people i.e. remove bad links
And since we didn’t know how does Google sees things, we had to do both the things. Getting good links is terribly difficult which is time consuming and requires lots and lots of patience and buy-in from upper management. Thankfully we had both and we (our outreach team is brilliant) did some fantastic work in this area. Will probably have a post later on some of the good practices of link building and outreach.
So we had to remove our bad links. But we were still a confused lot and had many questions. Some of questions are -
- How to determine whether a link is bad or good? How would Google determine it?
- Would Google look at individual link or overall link profile?
- If Google looks at link profile, then what all does it look?
- What ratio of total link profile be good links?
- How would we remove bad links? And what-if the webmaster who created the link in the first place refuse to remove it? And if he charges us, then that would be an extra cost to the business which is already in doldrums.
- Is disavowing links good enough for Google?
- Does Google has different barometer for different verticals to access link profile?
- How to determine whether a domain is bad?
- How would Google see a no-follow link from a good domain? And would it count towards making our link profile a bit better?
Questions galore but nevertheless we started. We still are not very confident about answers to these questions but some of the things which we did helped us recover back some of our traffic. The first thing we did was to understand our overall link profile and benchmark it against different players in the same vertical. Below are the steps which we took-
- We dug out all the links from GWT and charted it out. Here is the slide. Notice the problem here -> Links/Domain is increasing rapidly which is not natural at all. We dug out more to find out that all these links were emanating from directory networks. And these were not created by us. Therefore action point was to contact these directory networks and get our links removed.
- Next we wanted to see how does our referring domains vs Pagerank/DA (Domain Authority) chart look like. Not surprisingly enough, this had to look bad. Approximately half of our links were coming from PR 0 domains. And we knew here that this is quite bad in absolute. We referenced to this article for comparative analysis.
- Next we wanted to see how do we fare on anchor text vs links distribution. We learnt that it is important to have this distribution look very natural. We referred here for the distribution. Here is a snapshot of how the distribution should look like -
- Exact match means – Your business keywords like used car reviews, assignment help etc.
- Phrase match means – business keywords + some text like online assignment help
- Brand Name – Your business name for example Long Tail Read
- Branded – Business name + Some text i.e Long Tail Read is a great blog.
- KW Branded – Business name + Business Keywords i.e. learn seo at Long Tail Read
- URL - means URL
- Our distribution sucked. In almost 50% of links, our anchor text was our business keywords which is not natural. Therefore Action Point was to get rid of those links which had such anchor texts and were coming from PR 0 websites. Not surprisingly enough majority of these links were from directory networks. Tools used: Open Site Explorer and Ahref. Our outreach team simultaneously started building links for branded anchor text as the ratio was abysmally low.
- Till now we have been talking about links, now we focused on domains. Here is the spreadsheet which we made to decide which one to keep and which one to remove. Domain_file. Brief about each field -
- Source: Which tool helped us to discover this domain. We have used GWT, Bing, Cognitive SEO, Link Detox beside others.
- Indexed Pages: How many indexed pages does the domain has on Google? We found that quite a lot of domains from where we were getting links didn’t have a single page indexed. Alarming situation.
- Outbound Links: How many links from the domain
- Theme: This is our way of categorizing the domain for example social networking site, directory, Forum etc.
- DA: Domain authority as per MOZ
- Alexa: Alexa rank
- PR: Pagerank
- IP: IP address. We wanted to make sure that there aren’t too many domains from the same IP.
- Now one may add more parameters to the list.If you have any more parameter, do tell me – I would be happy to edit ours list too.
So by now we had figured out which domains to remove and which links to remove, but now we were stuck on whether we should just disavow them or actually make an effort to remove them. We didn’t take any chances. Though we disavowed all of them but also followed up with each webmaster to remove them. Some webmasters tried to squeeze us hard for money and frankly some of them succeeded as well as we were shit scared. In total we invested around $800 to get links removed. But we think it was worth it.
The worst thing about traffic fluctuations due to such algorithms is that you have to wait for a long time to know whether you have done it right this time or screwed up again. And we had to wait for full one year this time. Penguin graced us again with its presence in October 2014 (October 17 to be precise). And here is what our traffic chart looks like -
We have been able to recover 25% of the traffic which we lost during Penguin attacks and am pretty confident that with further updates, we will come out even stronger.
So coming back to the questions, here is what I think -
- Google looks at overall link profile and not individual link per say.
- We looked at different parameters (shared above) to determine the health of link profile. There might be other parameters as well. If you know more, please do share with me.
- Be prepared to shell out some bucks to get links removed. I am not 100% sure whether this is absolutely required or not but we didn’t take any chance. We negotiated hard and brought people to an amount with which we were comfortable with.
- People say that disavowing is sufficient but again we didn’t take any chance.
- We did benchmark some of the parameters against other players in our vertical to get a sense of how we are doing. And my sense says that yes Google should also have different barometers for different verticals.
This has been our journey so far in fight against Penguin (I hate myself for saying that). I would be happy to learn from others’ experiences therefore do share with me.