Another day, another flip-flop from Google on the functionality and specification of nofollow… Lets look back at all the times it hasn’t changed in the last decade:
Timeline
-
January 2005: Google, Yahoo & MSN Announce NoFollow
As reported in detail by Danny Sullivan (yes, now Google Search Liaison) NoFollow was introduced to combat link spam manipulating results, particularly in comment boxes etc.
Here’s the original definition, as reported:
We later found out that point (1) in that list wasn’t strictly true, Google still used the nofollow links for URL discovery (ie. they did technically ‘follow’ the link, even though they never passed PageRank.
-
June 2009: Google Announce Changes to NoFollow
Again, reported by Danny (you really couldn’t make this up), Matt Cutts announced contemporaneously at SMX Advanced that PageRank sculpting by way of NoFollow no longer worked:
I’ve tried to check the original thread, but it has since been deleted (or my access has been removed, take your pick), luckily Danny recorded it for posterity on the article linked above.
It’s pretty fair to say though, that the wording here explicitly confirms that a followed link to a page on site that is of low search value, is “wasted”.
Furthermore, Bruce Clay managed to record some Q&A with Matt following his session at SMX, I’ve highlighted the bit about changes:
Src: https://www.bruceclay.com/newsletter/volume68/mattcutts-nofollow.html
March 2018: Google Announce NoFollow is the Same as its Always Been:
At SMX West last week, Nathan Johns, Google Search Engineer announced in a session that NoFollow has always remained the same since its inception:
Src: http://www.thesempost.com/nofollow-has-not-changed-since-google-launched/
I’d just like to point out, this probably IS A SURPRISE to most people that may have taken Google at face value over the years as to the definition of PageRank.
Some may argue that this is a minor detail, that NoFollow definitions are only to do with spam links, and that if you payed attention to Matt Cutts five years ago, you would have abandoned PageRank Sculpting anyway – but here’s the thing:
“PageRank Sculpting” Never Went Away
Back in 2015 this topic came up (again), and Matt Cutts confirmed that it still didn’t work, and that people investing time and effort into ferrying googlebot and PageRank round their sites were essentially wasting their time.
This has never been true, at least of enterprise sites with bucketloads of accrued link equity (like, the sites Ive worked on throughout my career).
I even wrote a piece about it back then (Yes, PageRank Sculpting Still Works), and that logic still has not changed: in an environment where PageRank is divided between links, removing some of those links ultimately results in increased PageRank being assigned to the remaining links.
What Should SEOs Do About PageRank Sculpting in 2018?
Based on all my experience, (including the last ten years of running SEO on some of the largest sites online, attracting millions of organic visitors per day):
I will continue to strip out inefficient links from site architecture, siloing content together and reducing the total number of navigational links required to access all the pages, and benefitting from being able to direct PageRank as I deem most efficient for both surfacing new content, and passing PageRank into deep URLs.
All I can do, is suggest you do too, and not take Google’s word for granted on these matters. Do your own testing, make your own mind up.
I’d love to hear your comments below! Do you think it’s changed? Should Google get their Messaging Straight?
So there is basically no value passed throught nofollow links?
What about Wikipedia backlinks? These still seem to have an effect.
There’s been lots of talk about this over the years. For some time the perceived logic was that certain “respected” sites passed some value through nofollow links, but that this was limited to Anchor text, not PageRank itself. I *think* it was Loren Baker that wrote about it first, but we’re talking 10+ years ago now.
I also seem to recall, Matt Cutts went on some years later to corroborate this, but said that it was unintended and a bug, that had been resolved.
Other than that, the messaging is that they pass no value at all.
But experiment and choose your own path!
My impression of pagerank sculpting is that it didn’t work. I always believed Google to calculate Pagerank using the total number of links on the page in the denominator – regardless of whether they’re followed or nofollowed. So in that scenario, your followed links still get the same amount of pagerank they would if all the links were followed.
So a simplified formula would be Pr(Followedlink) = SitePr / numTotalLinksOnPage
That certainly is what they’re saying now – but it doesn’t match with Matt’s statements historically about it changing.
Removing the total number of links on a page, rather than nofollowing them, still results in the old objective of sculpting PageRank flow anyway, so Im going with that! 🙂
Yeah you can still sculpt by not putting links to the stupid stuff (or doing them with onclicks instead of links) but I don’t believe the nofollow method of sculpting ever worked.
I have to disagree with your assessment that it never worked, Ryan. Before the 2009 revamp of their nofollow policy, I participated with a colleague on an experiment using an established ecom site. While it fell short of conclusive, it gave all indications of working to boost selected pages… enough so to convince us, at least.
And while it certainly won’t work today the way it did then, I think there’s still some benefit to be derived from it.
I do think it’s of lesser benefit than a host of other things we can do, though.
I was under the impression that Page Rank Sculpting didn’t work because the Page Rank that used to be re-distributed via the other links on the page once a link was nofollowed, now “evaporates”. That’s what I thought the change was with the functionality of the nofollow that made it not work well for that purpose. In other words, the saved Page Rank from the nofollow link doesn’t get redistributed. This certainly wouldn’t preclude people from using other ways to create links that the engine can’t see but that are visible to the user and in that case I see no reason why it wouldn’t still work just based on the math of how Page Rank works.
This sums it up best —> “I will continue to strip out inefficient links from site architecture, siloing content together and reducing the total number of navigational links required to access all the pages, and benefitting from being able to direct PageRank as I deem most efficient for both surfacing new content, and passing PageRank into deep URLs.”