Nofollow links passing value or rankings has been the underlying question since this inline link attribute was developed back in 2005. Many millions of websites have implemented this method to tell Google to not follow certain links that the website owner may not necessarily be able to “vouch” for. Nofollow has also been used to sculpt internal PageRank flow as websites continuously tighten their website’s internal linking and navigation paths to focus on the most important pages of their websites. Further, the nofollow link attribute is built into many popular CMS platforms like WordPress and Drupal, so it’s one of the most common and standardized outbound link attributes currently being used. So, considering the input from Google over the past five years and the larger percentage of Webmasters concluding nofollowed links were mainly pointless to pursue or acquire, this question is the one we were focused on answering in our experiment:
“Do rel=nofollow links pass or denote PageRank, rankings, or subsequent page value?”
The answer is yes to at least two of these — rankings and some page value.
I know you’re probably saying “Whoaaa! Hold on there partner. You can’t conclusively say that nofollowed links pass value and actual rankings without some sort of testing or proof.” We knew this question would be asked. So let me answer with proof of our findings and also back it up with what other well-respected SEOs are also concluding with their own findings for nofollow links and Google’s not-so-defined perception of them. It should also be noted that in many instances a nofollow link acts as a means of link discovery for Google. There are a lot of other nuances to Google’s algorithm for calculating links — such as internal navigation and anchor text used within a website for just a few short examples. Our goal was to see how far beyond link discovery some testing would go.
Our experiment focused on keywords not used only used within the website less than 10% of the time. The keywords we acquired nofollow external links for were not within any page attributes on our two domains used in this experiment — whether those site links were contextual, navigation, or URL page naming. There were variances of the keywords we used in our link building campaign found within the content on a few our our domain’s but the goal was to have as close to <10% keyword matching for on-page content as it corresponded to the keyword phrases used in nofollow backlinks we would be acquiring.
Nofollow Experiment in a Nutshell
To examine how only nofollowed links were valued in Google and how pages performed using nofollowed links we did the following:
- Our domains — Set up two brand new domains that were ambiguously named. These two domains were not an exact phrase match to our keyword for link building or really even close to relevant for our two industry niches as far as the keywords were concerned. We’ll call our domains we used “domain #1” and “domain #2.” These domains were built for niches but not limited to being local or geo-specific. We did not want to find in the end that we were ranking to easily because we were the only businesses in a tiny local niche. In other words, we wanted to make sure the competitive pool was larger, not smaller to help define and reach our objectives.
- Duration of experiment — We spent three to four weeks commenting only on blogs that included the nofollow link attribute for each of our domains. We measured a total of times during these weeks — although I peeked at some of the reports at times.
- On-page stuff — We made sure that we were not doing any lengthy on-page optimization for the keyword phrases we were targeting for the acquisition of the nofollow links. We felt it was better the nofollow links we acquired would be the primary and only real usage of of keyword strength or relevancy.
- Link building guidelines — Websites we were to acquire links from were related to the industry that our two domains were in. However, we also made sure we were not posting comments that were spam in nature and instead we really contributed to the author’s post and tried to be as helpful and constructive as possible for others. Further, we made sure we were posting on websites that were decently ranked and indexed regularly by Google (quicker cache dates). 80% of websites we acquired links from we considered in the upper 30% within their respective industries.
Both domains are active and each has approximately 8-10 pages in total. Both were indexed in Google as most new domains are within a few day or week period. Both had zero backlinks. Neither domain ranked for any of the five keywords each that we were going to build nofollow links for in the subsequent weeks. By not ranked, I mean not within the first 20 pages of Google for any keyword phrase.
We built 35 nofollow links for both domains combined using eight of the combined ten keyword combinations. The results at the end of week #2:
- Domain #1 was not ranked on the first 40 pages of Google for week #1, now three keywords are ranked — one on page #4, another on page #6, and the final keyword on page #8.
- Domain #2 was not ranked on the first 40 pages of Google for week #1, now all four of the keywords for that domain were ranked in Google. Two on page #3 and the other two on pages #7 and #8 of Google.
Note: All backlinks to this point are nofollowed links.
We built another 40 nofollow links for both domains combined using all ten keyword combinations. At this point we have a total of 75 nofollow links pointing to our domains, about evenly split between both domains. The results at the end of week #3:
- Domain #1 moved up again in Google, now with all five keyword phrases ranking between pages #2 and #5.
- Domain #2 also moved up in Google, now with all five keyword phrases ranking between pages #1 and #9 — with most keywords between pages #1 and #5.
Note: In this round of link building we purposely built around 15 links with a misspelling that was quite unique. The goal was to target an anchor text that was totally unique, i.e. Google has no current search results for it in their index. We also targeted several internal pages to see if we could actually get an internal page or two to rank for this misspelling.
Weeks #4 & #5 – The last week!! Yay!
Weeks #4 and #5 went more slowly with building nofollow links. For this two week period we only built another 15 links total, bringing our total of 100% nofollow links to 90. The results at the end of week #5:
- Domain #1 edged up again. All five keyword phrases improved and all were within pages #1 and #4.
- Domain #2 moved up slightly. All five keywords but two improved. Two keyword phrases were now on page #1.
Added Information to Conclude Nofollow Link Results We Found
It should be noted that we ran SEO Majestic, LinkScape, and another internal tool to quantify our finding and to rule out any other links, such as scraper sites linking to our two domains. We ran SEO Majestic and the other tools a total of 3 times during this experiment and nothing but our nofollow links were found.
Further, I think we should explain the type of keyword phrases we were ranking for on both domains. Here’s an example of anchor text and URL used and ranked for and how the anchor text was used on the corresponding website:
domain1.com/specific-air-conditioning-rating-type (not model like Trane, etc. but type — as in SEER 10 or SEER 16 rating types)
On the above page for domain #1 we used two different combinations of the external anchor text used for linking to this particular URL. The anchor text used was not a match for the URL keywords used. This means we did not use “specific-air-conditioning-rating-type” as noted in our URL above. Instead, we used several keyword combinations on both of our website’s pages that were less than a 10% match for keywords used for our link building. Furthermore, we did not have the keywords we used for link building on other pages of the site, such as sitemaps, navigation, or other main keywords in the titles, headings, or other locations on the pages on our domains. I did this to less our probability of getting ranked even more.
Final Results and Conclusions on Nofollow Links
Approximately 5 1/2 to 6 weeks after the nofollow links experiment started we found that, even though our domain names were not keyword relevant and our website content was <10% relevant to the keyword phrases used for our nofollow link building, we still ranked for those keywords after only acquiring nofollow links. The assumption by many is that nofollow links will not pass much, if any, link juice or anchor text value. But the nofollow link can still pass relevance… and slightly more in my opinion. Google may not “count” the link as a weighted backlink but this doesn’t mean they ignore the anchor text being used or the authoratative status of the website being linked from.
I find it interesting that Matt Cutts states that a nofollow link definitely does not pass anchor text, especially when the only keywords we ranked for and were using in our experiment described above were 100% nofollow links. Matt specifically stated this in his blog post above:
Nofollow links definitely don’t pass PageRank. Over the years, I’ve seen a few corner cases where a nofollow link did pass anchortext, normally due to bugs in indexing that we then fixed. The essential thing you need to know is that nofollow links don’t help sites rank higher in Google’s search results.
However, this does not correlate to our findings. What gives? Do nofollow links now pass some anchor text value in Google’s algorithm? It seems so. Coincidentally, there was a recent article on Google and Nofollow links over at Sphinn.com that tied in well with our experiment over the last month and a half. I thought a few viewpoints from several very knowledgeable and well-known SEO’s would be great to add to our findings above. Here are a few examples. Matt McGee stated:
…We both agreed that we’d seen this kind of scenario and that, in some cases at least, no-follow links were influencing rankings. So I’d say that, yes, in some cases where there aren’t a whole lot of other ranking signals to use, no-follow links are definitely worth more than Google says.
Jill Whalen also noted:
Too many are nofollowing all links, or some links without having a clue as to what they’re doing. Google has spent tons of time and energy in learning how to graph links in terms of their popularity and authority. There’s no way they would simply ignore all that data becuase a bunch of dopes stuck an attribute on their outgoing (or internal) links for “SEO purposes”!
Finally, Marcus Miller said:
When examining a site, they [Google] could use nofollow links to increase trust, they may use them to look for organic linking patterns and to compare these with any uncapped links. They may even use them to determine if they can trust the anchor text from true links if it is replicated in nofollow links. So, I am not sure that nofollow links alone can rank a site (I can feel an experiment coming on) as this would mean that google is lying to us all but I think they are certainly part of the bigger picture when it comes to trust of a site and relevance of ranking keywords.
I would suspect that Marcus is partly right in that Google is looking at more than just the nofollow link itself to signal trust. Perhaps authority, comment relevance, overall anchor text distribution for nofollow links, and many more markers now go into how and why any particular keyword phrase and/or pages would rank in Google — despite the keywords deriving solely from nofollow links. What do you think? Is there a silent shift going on at Google with nofollow links or are they not telling us the complete story? It would definitely not benefit Google to tell the web community to go out and get as many nofollow links as possible since it would open up doors for more spam and gaming in Google.