So, over the last 15 months, we analyzed nearly 1 million backlinks. This is our understanding of how every link is useful or not, along with the June update. AMA about either of these or backlinks in general.
As the title says, we analysed nearly 1 million backlinks (989,176 to be precise) since last year's C3 Conductor in New York got cancelled, and we lost the initial set of clients in the first 2-3 weeks or so. So, we set up this experiment to get a better understanding of what kind of backlinks impact websites, and at what stages in their life journey, and what are the factors that you should measure to be sure of how good the potential backlink is.
That being the case, these are the factors that we now always look at before determining the quality behind any backlink that is acquired:
1. Organic Traffic of the website- Now, even though this is something that is obvious, for us to safely factor this in is an issue, since the only data that is available to us is from Ahrefs . Nevertheless, we take a 20% margin to determine that if a website doesn't have traffic, then getting a backlink from it doesn't make sense. And as the organic traffic increases, so does value of the link being acquired from it. However, there is a line of demarcation that needs to be set up, since there are websites that will artificially, or with the help of black hat SEO would gain traffic. On that front, we kept 100K sessions per month as that number which the website should have maintained for at least 6 months, in order to get that safe website badge.
2. Ahrefs Domain Rating (DR) range and the time it took to build that DR- We focused on DR here for our analysis, and once again made a demarcation point of 30 as that clearly safe zone beyond which we could say that the website. We also looked at the trend of how they have progressed in terms of their backlink acquisition in terms of Referring pages and Domains, along with Ahrefs Domain Rating (DR) as well. If there are any sudden upward or downward movements across any of the charts, that should be explored, since it mostly points to mass spam link related activity. Any drop or rise in the range of +/- 25% is considered as a case that needs further investigation, and anything above +/- 40% is immediately rejected.
3. Geographical relevance- A quick question to act as the example over here: If you are a business based out of USA, should you get a link from a site whose majority traffic share is from Germany, or Czech Republic, or Russia, or South Korea? The obvious answer here allows us to see whether or not it would make sense to acquire a link from a geographically relevant site.
4. Topical relevance- Following up on the same example as above, if the business that we mentioned is a dentist looking to get more leads, would it make sense to get it from the website that writes about Tech or Marketing? In this case, the grey area would be creating a topic which relates both these fields like, "Why dentists need to market their websites?", but usually the answer is No. In cases like these, you look for closest relevance, i.e. a website that is about Healthcare, and get the most relevant article from them that links back to you.
5. Integration of the link in the content- An important part of the any page is the integration of links in it. Not just the anchor text, but the content before and after it matters a lot. So if you are writing a few lines about how eating fish is good for eyes, and all of a sudden in the next line you link to an Optometrist, that won't be a considered as a good integration.
6. Non-repeated linking root domain (or repetition depending on the quality of the LRD with a diminishing return)- Consider a simple example: Look at the total backlinks a website has, and the number of domains from which all this links have been acquired, and divide Backlinks/Domains. If the ratio is anything greater than 20, check out if there are "x" number of domains that are providing a very high number of backlinks. Too many of these domains, and you are dealing with a website that would have a potentially high spam score, and is a website that you should ideally avoid.
7. Anchor text variations- There are 5 different categories of anchor texts. Let's say you are working for Samsung. In that case, you can come across the following anchor texts
a. Branded Anchor Text \- Where the term "Samsung" is used directly as the anchor text. Ideally, 50-70% of all links should have a branded anchor text.
b. Soft Anchor Text \- This is where the brand name is used with a few words to soften up the brand name. An example would be, "visit Samsung's site", and it should ideally be used 20-30% times when it comes to link acquisition
c. Nonsense Anchor Text \- If you are using a generic phrase like, "click here", "visit this page", that's a Generic or Nonsense Anchor Text, and should be used in less than 10% of the cases.
d. No or Naked Anchor Text \- Using a page URL directly is termed as Naked anchor text, and once again, should be used in less than 10% of the cases.
e. Hard Anchor Text \- Using a keyword that you want to rank for, and something that should be avoided as much as possible. An example would be "Samsung Galaxy S20" and ideally the number should be kept under 5% to be safe.
8. Follow and No-Follow link– This goes without saying but DoFollow links help your website much more than the NoFollow links. The only point to consider is that you should have a good mixture of NoFollow links to create an optimized backlink profile to ensure that it looks natural.
Looking at all of these factors allowed us gain a better perspective of which link works best for a particular website, and allowed us to quantify each factor so as to provide every backlink a score out of 10. The smaller the score, the higher the chances of it being a link that would harm your website in a long run. This helps us in identifying every link that we acquire, as well as every old link that is currently active for the purpose of doing a link audit.
As far as the June update is concerned, we saw a decent update across the board. A few notes before we started: We saw a few clients who started with losses and ended up gaining traffic, and vice-versa over the course of the 10-day period. I am only going to talk about the results that we saw at the end of this period and will not consider any fluctuations. I will bifurcate the results in terms of the niches that we serve to make the observations a little easier. And finally, any client that saw a traffic increase or decrease in the range of 5% would be considered a non-factor or stable clients as that is a general trend that could be observed in any case, with or without any update.
Dental niche – Our largest niche with 31 clients, we saw 19 of them gain rankings, 7 remained stable, while 5 of them lost traffic over the weekend. The average gain was 21% while the loss was 17% for the 5 clients. In this case, most of the clients who saw ranking gains had been updating their content regularly and had focused on increasing their link velocity at an average rate of 5% on a month-on-month basis (clients who were acquiring 50 links or more per month), while the ones who lost out on rankings were low on their Link Velocity, and their E-A-T markers weren't in place, and neither were the Core Web Vitals (CWV) optimized to the degree that should have been.
Eye Care niche – Our second largest niche with 21 clients, we had 15 winners, 2 who remained stable and 4 losers. The average gain was 24% in this niche, and the average loss of traffic was 14%. In this niche as well, we saw similar trends when it came to gains and losses as in the Dental niche.
Other Specialty Hospitals and General Hospitals (Minus Dental and Eye Care) – Among the last 27 medical clients that we have, we had 17 winners, 4 who were stable, and 6 losers, with 3 hospitals posting a loss beyond 30%. The average traffic gained for these clients stood at 19% while the average loss was at 24%. Most of these clients had poor CWV scores, and that makes us a bit scared going into June with the CWV update coming in.
Software As A Service niche – SAAS sites saw 11 out of 17 clients seeing a rise in traffic (27%), while 3 clients saw an average dip of 14%, and the last 3 observed changes of 3.1, 2.9 and 2.8% traffic respectively. Site architecture and Core Web Vitals (CWV) played the most important role for these guys, which was in sync with the December update as well, and their content production was close to 1-2 new content pieces a month. The ones who lost out had a poor Links/Domain ratio of 397/1, 318/1, and 273/1 respectively, along with poor maintenance of their Link Velocity and CWV scores.
Education niche – Education was where we hit gold this time around with 11 out of 15 clients seeing an average rise in traffic by 31%, and 3 clients losing on an average 9%, and the largest client saw a traffic uptick of 42.7%, and the last guy with no change at all (I mean if you consider 0.3% as a change). A huge focus on CWV and architecture here, since the last update wasn't too great for this niche, and we had been hounding them and their techies to get their shit together. Very pleasantly surprised and hopeful for the June update for this niche.
Finance niche \- We had 10 clients here, out of which 6 reported an average gain of 19%, while 2 reported a loss of 13%, while the remaining 2 stood at 3.7% and -2.1% respectively, with just 1 content piece being published a month and average Core Web Vitals and stable link velocity (5 links per month on an average). The content production was once again not that great in this niche except for 1 client and they increased their traffic by 29%. we basically saw that good CWV focus and a good architecture definitely helped the industry.
Other niches – Among other niches (real estate, tech platforms and e-mags, e-commerce, gaming sites, travel, restaurants, cannabis and a single funeral website), we saw 29 of our clients gain rankings among the overall 42, while 8 lost, and 5 remained stable. The overall growth of 24% for these clients was supported by a huge push for improving their CWV scores, an uptick in content production, followed by localized link acquisition with an increasing Link Velocity since the December update. The clients who saw their rankings take a dive by an average of 11% mostly had poor CWV scores for their major product or service page, and hadn't been updating their content properly.
That being said, if there is any question related to the experiment we ran, or the June update, or backlinks in particular, please feel free to ask. 30 💬🗨
chrisnddie How you rate premium backlinks? How much they are effective for any niche?
Whats the difference in Ahrefs Domain Rating (DR), traffic etc, if any website have it or not.
randomvar ✍️ It depends on the industry and geography. For every niche, we normalize them by determining the average traffic of the sites for that niche, otherwise it won't be a fair assessment.
Once a benchmark has been set, the rest of the variables come into the picture and we are able to determine the score for the backlink with all the fqctors that I had mentioned.
skipintroboi Who is the "we" you keep talking about? I don't read research results unless I can understand the source and their perspectives/incentives.
randomvar ✍️ My team and I, and I run my SEO agency/consulting firm.
attidude I too have been a digital marketing/SEO consultant with in-house and agency experience for roughly 11 years now and this is pretty similar advice and findings for my backlink portfolio building tasks.
This is great data to back-up what we've been practicing for our clients, particularly the focus on branded anchor text being the most dominate.
There are lots of rules to follow, but I do like a "shotgun" approach to quickly see which backlinks do/do not perform over a 3-6 month period.
To everyone in this thread: just do testing. Find qualitative backlink vendors and test them. Don't get those $5 – 1,000 backlink offers; go for niche's and qualitative backlinks with the parameters outlined above.
Great stuff!
HispanicNachs Can you elaborate why hard anchor text should rarely be used? If I have a page about a "classic apple pie recipe" and I'm trying to rank for that keyword, why is it bad to use "classic apple pie recipe" as the anchor text? What would make more sense instead randomvar ✍️ Hey man,
Thanks a lot.
And I completely agree, that there is no better way to learn in Search Engine Optimization (SEO) than testing and experimentation.
DisPater Is it really bad if the keyword appears in the anchor text? This is the case with most of my backlinks…
randomvar ✍️ Well, it's an obvious ploy to rank on that keyword, and is easily detectable. I would rather avoid it.
JustScray Do these same principales apply when link building for a specific sub-page?
For example I have several categories like "televisions", "PC's" and "phones". What we usually did when link building for one of those specific groups would be 65-70% of the links with a matching or similar anchor text to the keyword, about 20% branded and 10% generic phrases.
We have seen good results with this practice, however when I read your post it suggests we would see better results if we change these ratio's towards more branded anchor texts.
randomvar ✍️ Hey there,
From our analysis' perspective, yes, you would stand to benefit if you changed those ratios.
espressodude Thanks for this very detailed finding. I appreciate you compiling it.
Question — though this may be a bit of stretch — any advice on link exchanges?
A number of our link building efforts sometimes end up with the other party always asking for a link back.
Did you by any chance analyze this setup of exchanging links?
randomvar ✍️ Please don't get into link exchanges. All they do is hurt your backlink profile, and are easily detectable by Google.
espressodude But what if they are relevant links added to both parties? I mean links not just added for the sake of being there, but links that actually add value and those that make sense as a reference.
rpmeg This is incredible! You and your agency are absolute masters of your craft and I sincerely appreciate you sharing your original research with us. Thank you thank you thank you.
I got hit pretty hard in June. From speculation, too much hard anchor text. Would you suggest updating and/or removing those links, or just building more link types to get to the ratios mentioned above?
randomvar ✍️ Thank you so much. I really appreciate the kind words.
Well, I can't comment with absolute certainty without looking at the backlink profile, but I will go with you analysis as of now. I would say a bit of both: removal of the lowest quality links, and acquisition of the best quality links to offset the removal.
