- Google relies far more on content on a site than a domain’s link profile to determine relevancy
- Even for legitimate, natural websites, a large proportion of links do not come from niche relevant sites
- Real websites do not solely link to sites in the same niche
- Most links across the internet are from, and to, domains unrelated to each other
- A natural link profile includes many links from a variety of random niches.
- How we ranked 2 websites in highly competitive niches with less than 10% relevant linking domains
For many SEO experts, the relevancy of a linking domain is one of the most crucial metrics to consider when link building.
The SEO blogosphere is full of articles espousing the importance of building niche-relevant links, while SEO forums are overflowing with people agonising whether or not to disavow non-relevant links to remain safe from penalty.
However the question many of these people fail to ask is: what exactly is domain relevancy and how big of a ranking factor is it?
From an algorithmic perspective, we need to look at how Google determines it, and how this affects how you prospect for, and build, links.
We’re going to be looking at the problems Google faces when determining domain relevancy, and find ways to exploit these difficulties in a manner that allows us to build powerful links even if, at first glance, they’d appear irrelevant to your niche.
An Algorithmic Approach to Relevancy
For this section, let’s disregard the huge influence that anchor text and on-page content play in deciding relevancy and instead concentrate on nothing but linking domains and their effect on relevancy.
Let’s first take a minute to visualise how GoogleBot crawls the web, and how this information is sorted and integrated into any of the algorithms that Google runs.
Gbot finds and follows hyperlinks from one site to the hundreds of external sites that the site might link to.
Let’s say, for example, we were to crawl just 10 separate, individual domains. Of course, this crawl would be a tiny, infinitesimally small sliver of the whole internet; by some approximations less than 0.00000005% of Google’s existing index.
However, even a tiny sample of just 10 websites creates an enormous, brain-meltingly complex net, linking each and every of these 10 sites to one another, as well to untold thousands of other sites receiving links from these first 10 sites.
Extrapolate this network, and the amount of data required to track connections between the sites, across the whole of the existing Google index, and you are faced with an unfathomably large set of data to make sense of.
Due to the nature of links, to decide on domain relevancy based on linking domains alone, Google would need to not only look at the direct links to our site, but also at the links pointing at the domains linking to our site – a net of links spanning thousands and thousands of disparate sites.
Theory aside, by looking at a real world example, we can get an idea of how feasible it could be to work out domain relevancy from links alone:
Case Study 1: Mansons Guitars
Manson Guitars is not a client, or affiliated to us in anyway – just a guitar manufacturer based in Devon, who make custom guitars for artists such as Muse and Led Zeppelin.
From looking at their backlink profile, we can see they have powerful, authoritative links from other musical instrument companies such as Yamaha.com, fender.com and guitar sites such as ultimate-guitar.com and premier-guitar.com. Therefore it should be easy to decide that yes, this domain is relevant to guitars/musical instruments.
However, alongside these hyper-relevant links, the majority of links come from sites with no relevancy whatsoever to guitars, musical instruments or even music.
Having audited their whole backlink profile, only 37% of links came from sites related to music, musical instruments or guitars. If we look strictly at guitars/musical instruments, this number drops to 24%.
So even with a 100% legit, world famous business, with a completely natural link profile, only 37% of their profile comes from niche relevant sites.
So, if we are to expect relevancy to be based on what sites link to a domain, we can understand how hard it must be for an algorithm to figure this out. While 37% of links were from sites in the same niche, 10% of sites were fashion sites, 13% came from electronic/gadget related niches.
Does this therefore mean that, based solely on the sites linking to Mansons Guitars, that the site should be associated with fashion and gadgets?
It’s also important to remember that we’re looking at Mansons in complete isolation, whereas the algorithm is also looking at the links all of the sites that link to Mansons have as well.
This means that each and every site is as much of a topical boiling pot as Mansons is. Therefore, when we say ultimate-guitar.com is hyper-relevant to Mansons, we’re ignoring that ultimate-guitar.com also has a huge number of irrelevant domains linking to it. So what is easily defined to us as a “musical instrument website” will have a similarly wide spread of links from seemingly random sites, from an algorithmic perspective.
This therefore dilutes the overall “domain relevance” from the relevant sites linking to Mansons in the first place.
Therefore, for an algorithm to decide, with any degree of accuracy, the niche of a site based on referring domains and links alone is impractical.
As a result, it is logical to conclude that anchor text and the content surrounding links play the most significant part in deciding how relevant a link is, rather than the relevance of the domains linking to it.
Natural Linking Patterns and Relevancy
So, now we understand a little more about how random the link profile of any given site is, let’s apply this to a link building strategy.
A great illustration of how normal sites link to each other can be seen with another example of a real site: LDNfashion.com. Again, we are not associated with this site in any way – I just took the first editorial site I could find after Googling a random keyword “fashion sites UK”.
LDNfashion, as you can probably guess, is a blog/webzine dedicated to fashion in London. We analysed all of their outbound links to see what sites they were linking to. Surely an informational site to do with fashion should link almost exclusively to fashion brands, cosmetics and other tangentially relevant sites?
In reality, we found that 47% of the sites LDNfashion linked out to had nothing to do with fashion, style or even anything vaguely linked. LDNfashion doesn’t sell links, accept advertorials or anything of that nature – all outbound links were editorial and served a purpose in the article they were included in.
LDNfashion linked out in a completely natural way to sites as far removed as expert audio equipment, an artisan cheese manufacturer, a US based book shop, a graffiti site and a shared office space in Brighton.
By seeing LDNfashion as a tiny part of the huge, tangled tapestry of the internet as a whole, we can see how little influence relevancy really has on how the internet is inter-connected, as websites simply do not exclusively link to relevant sites.
Proximity – A New Way of Seeing Relevancy
We visualise niche relevancy in somewhat isolated silos, where Manson Guitars and all of the guitar/instrument sites linking to them fit into one neat, topically connected section. However, as we’ve seen, the lines separating and defining domain relevancy are a lot fuzzier and hard to define.
Rather than placing sites into connected boxes, as is visualised by services such as Majestic and their topical trust flow, the reality of the way the internet links together is much more like a chaotic Jackson Pollock painting, where different niches are splattered across multiple sites, seeping into one another in a maelstrom of different topics.
Therefore, just as the reality of domain connections and niche relevancy is far more abstract than most imagine, so too must be the way Google sees and measures relevancy, with any site’s “sphere of relevance” being far more fluid and far reaching than we would initially expect.
For example, http://cheesesonline.co.uk/ is one of the sites that LDNfashion.com links to.
If a fashion site could gain a link on that site, we may look at cheesesonline.co.uk and think it’s a poor, irrelevant choice.
However, to an algorthim that has not only crawled the links from Ldnfashion to cheesesonline.co.uk, but has crawled all the links pointing at Ldnfashion, and at cheesesonline.co.uk as well as every site linked to them in anyway, these two domains would be related by linking proximity if by nothing else.
Therefore, another link to a fashion site from cheesesonline.co.uk would be no more unnatural to Google than any of the 73% of links Mansons Guitars got from irrelevant sites.
We’ve looked at a number of sites, all of which have similar linking patterns to LDNfashion and Mansons Guitars, and while this is by no means a representative sample size, based on the information we have, it’s sensible to estimate that of all the links across the whole internet, a large proportion of links are pointing at sites that are utterly irrelevant to each other.
Therefore, when link building for a niche specific money site, we shouldn’t necessarily be concentrating solely on relevancy, but on link proximity – how close a site is, in terms of links, to another site in your niche.
By taking this into account, our links will be replicating the natural linking relationships of existing sites.
This has huge advantages for any link builder as it widens the net of powerful domains that can be prospected. By building links with a close linking proximity to other sites in our niche, we can build a link profile including sites with irrelevant content, just as natural sites do.
Testing the Theory
It’s all well and good writing about these theories, but how does this kind of link profile actually perform?
To demonstrate its effectiveness, we’ve taken two long term campaigns across competitive niches.
Site 1: Furniture Niche
This client is a big designer furniture ecommerce site, with a multi-million pound annual turnover. They are competing against huge sites such as Ikea, John Lewis and other big brands for many of their core keywords.
We can’t show our full link profile, for obvious privacy reasons, but here is a random sample of the “topical relevancy” of 20 of their linking domains.
As a furniture site, the received wisdom would be that most domains should have a high trust flow in the “home” or perhaps “shopping” topic.
The links we used, while high metric and powerful, were across a wide spectrum of tangentially related sites, from lifestyle blogs to fashion sites and business sites..
Over the course of the campaign, the site has seen an overall keyword increase of 2,302 places across 100 high volume, competitive keywords.
The number of individual keywords in the index has increased from 500 to 9,200 – an increase of 1669%.
Their organic traffic has increased by 290%, with revenue through organic increasing 313% from £131k to £542k.
Site 2: Property Sales Niche
Just like with the above example, we built links that were only loosely linked to the client’s niche.
Just like with the above example, this resulted in good links from a wide variety of different niches.
Within 6 months of beginning the campaign, organic traffic was up 93.91% compared to November 2016, with July receiving 1,146 compared to 591 in November.
Similarly, the number of new users had increased by over 85.43%, with 853 new visits in July compared to 460 in December.
The number of leads generated by organic traffic increased by 191.6% between November and July, with the site receiving 210 leads in July compared to 72 in November.
The conversion rate improved by 50.41%, despite the site not being redesigned and no conversion rate optimisation taking place. This increase shows that the traffic our actions were bringing to the site were highly targeted and relevant to the site.
Each fully converted lead landed the client £1000s. Assuming each conversion netted exactly £3000, with a conversion rate from online leads of roughly 0.5%, this meant that organic went from making £1080 per month to £3150. Extrapolated across the year, this is an increase of £24,840 per year through SEO, with a total of £37,800 brought in through organic SEO.
How Google Perceives Domain Relevancy
Real websites link to a huge variety of niches, and receive links from a huge variety of niches.
Therefore, algorithmically determining niche relevancy from backlinks alone is extremely difficult, as each individual site has to be measured within the context of the expanded network of sites linking to it. Making sense of such a huge extended network would be extremely hard to do, which is why Google relies so heavily on cues from on-page content and anchor text to determine relevancy.
How To Apply This To Your Link Building
The sole aim of building links is to rank your money site in a way that is penalty-proof in the long term.
As a result, your link profile should mimic the way sites naturally link to one another as closely as possible. By looking at the link profiles and outbound links of real, natural sites, we’ve shown that websites do not solely link to other sites in the same niche – that of all the links across the whole internet, a significant proportion of them are pointing at sites unrelated to each other.
Therefore, it follows that your link building should be similarly varied.
Instead of trying to find links explicitly related to your site’s niche, your profile can benefit from adding lots of high authority domains that already have links from, and therefore an established proximity to, sites closely related to your niche.
By doing this, not only will you be able to widen your net and find high-authority links, but you will better replicate natural linking patterns.