Paul Madden, formerly of Automica, recently teamed up with Gareth Hoyle nad became a partner of Manual Link Building. I really enjoyed Paul’s talk at SearchLove earlier this year, where he outlined how he scaled automated SEO processes using Odesk.
With Penguin, Google sold us the idea that using the disavow tool and simultaneously reporting on spurious agencies was a good idea. They targeted sites with poor link profiles and penalized thousands of sites overnight.
What is Penguin?
A filter or a penalty designed to target SEOs and anyone manipulating Google. Penalties are much more formal than they used to be and Google are definitely still sending unnatural link warnings.
People will tell you that content is king, but links are still what Google is based on, and will be for the foreseeable future. Links are the signal and Google needs them; no one has ever linked to the history of ladders just because they like ladders, so there will always be sites and pages that will need help in getting links.
It’s fine to use Infographics, guest posts and fashionable techniques, but strike a balance!
What types of links are bad?
1. Paid links. Although they are algorithmically the hardest links to spot.
2. Gifts. Eg: Cambridge bags as shown in Chrome ad where vendor sends in free samples
Types of Link Profile
If you’re lucky enough to have brand presence then you’ll always be indexed, you might get slapped and disappear for a while but Google will always have to put you back in. In fact, Interflora are a good example because if users search for flowers and Interflora don’t rank they won’t think “Interflora must have bad SEO,” they’ll think “God Google’s got bad hasn’t it!?” – so Google will always have to index large brands like that.
So there are two real types of link profile: major brands which have some immunity, or you’re like the rest of us and you have no immunity and you have to be much more careful. Every link that you have has risk, and you must manage each risk.
Validate every link opportunity based on its risk, and you have to make sure that if you’re going to outsource link-building work to a third party then you have to make sure you know what the risk is and make them demonstrate how they mitigate that risk. You’ve also got to be more realistic, so the more provisos and conditions you make on the links you build the higher the cost of the link.
So as a volume link building service – what was our response to this development? Well we started gathering data, 60 million data points to try to understand what’s good and what’s bad.
We discovered loads of hidden signals, so to manage the risk:
1. Audit what you have – include everything (even ancient links, because they will harm you just as much)
2. Have you got adult links
3. Use experts to audit – not juniors
Unfortunately, removing is almost as bad as not removing – let’s say you find 100 dodgy links, Google will only know about some of those links, and the rest will be the reason why you rank at all. So effectively by removing those links, you’ll be penalizing yourself! So it’s a balancing act.
Majestic is the best source of link data – download your link profile and start evaluating it in Excel. So what signals should we look for, well there’s no real revelation here, bad links have…
- Far higher % of on-anchor links (i.e. keywords Vs brand names)
- Higher % of sitewide links
- Higher % of suspect cctlds
Use screaming frog on sites linking to you – see if they have links that have been disavowed. What text are those sites using to link to other sites and do you think they exist for any other reason than to host a link?
What makes a link problematic?
- Poor link metrics
- Wrong ccltd
Paul has built a tool which does this for you: LinkRisk.
Matt cutts’ website – lots of different incoming anchor terms. Paul looked at a payday loans site – almost half if the incoming anchor terms are the same anchor text. Also, remember that all sites have some bad links.
For EVERY link you remove, you should think about what kind of link you need to build to replace that link.
After doing a reconsideration request, it’s ok to try again if you aren’t successful (sweet-spot 4 and 7 reinclusion requests).
301ing your existing website to a new one isn’t going to help, sometimes you’re better off just starting again.
NB: Huge thanks to my colleague Fraser for writing up this session, as I was unable to make it.