Last Updated: April 24, 2024
Roughly 4 years ago, at the start of 2017, I used to use expired Tumblr blogs to rank my portfolio of affiliate websites.
For those who may not be aware, Tumblr is a web 2.0 site, which can be utilized as a blog. However, often people will abandon their blog after years, and so the blog username and URL will become available to register. How does this benefit us? Well, sometimes these blogs have amassed a large number of backlinks and are great for us SEOs, for link building purposes.
Therefore, we can register these available Tumblr blogs, create content on them and link to our own sites to harness their existing domain authority and page authority.
And this technique worked like a treat!
I’d use a tool like FCS Networker (which no longer exists) to help automate the process, and it really was quite easy.
Indexing the Tumblr accounts was a little more difficult, but I had ways to do this, using tier 2 links and force indexation techniques. This is an entirely separate blog post.
Anyway, I used this method up until the end of 2020, when the December 2020 Core Update hit.
And my websites dropped like flies! At first, I thought, it has to be the link-building methods…. But I was very wrong.
I began cross-referencing our portfolio sites, which we use the same link-building strategy on, and some have been boosted, whilst others had been hit.
It took about 6 months to begin seeing patterns, and to review data from other SEOs in the affiliate space. It turns out those sites that had been hit had almost no informational content compared to commercial content.
Informational content is simply content written for the purpose of informing, whilst commercial content is created to convert/have affiliate links (at least for affiliate marketers).
Therefore, our ratios were all wrong. And so the sites just looked like spammy affiliate sites to Google.
During those 6 months, I took a step back from link building and began honing in on fixing the ratio of commercial to informational content.
It’s now been 2 years (boy has time flown) and the sites are recovering well.
However, after re-visiting the glory days of seeing old portfolio sites ranking again, I’m looking to now focus on link-building.
Because of this, it got me thinking… How effective are expired Tumblr links in 2022? And is the newly implemented Tumblr re-direct still passing link equity or not?
The answer is, I don’t know… But I’m going to find out.
So what’s the plan?
Now, since I can’t find much information from other SEOs about this, I wanted to run a test and create an SEO case study to document my progress.
For reference sake, this test is beginning on the 7th of April 2022, and I’ll be running it for the next 3 months.
I’ll only be using Tumblr backlinks, and the domain in question that I’m trying to rank has existed since July 2021. It has 21 articles and has no backlinks (other than maybe 5-10 really spammy ones which are created by scraper sites). The site has not been updated since 23rd August 2021 when the latest article was uploaded.
All articles were created and optimized using MarketMuse, in conjunction with Frase… So they’re ultra-high quality. But the site has no link authority at all.
With that being said, I would consider this to be a new domain, making this test very interesting. The last time I ran a test like this (back in 2018), I tanked the domain completely… I suspect I had engaged what was a more primitive, yet still effective version of the Google Panda algorithm.
Now, I won’t be sharing the actual domain, because direct traffic, backlinks and referral traffic (through clicks – if I were to link to it) would invalidate the test. However, you can rest assured that I have no agenda here, I’m as interested as you are in whether this test will result in a positive or negative ranking change.
I’ll be creating new sections within this post after each month, showing the ranking changes for main target keywords.
Therefore, the test starts today!
Hypothesis
Do expired Tumblr backlinks positively affect organic search rankings in 2022?
Test Design
Firstly, this test isn’t going to be perfect, there are limitations. And I’m going to have to make some assumptions here, based on previous experience with tanking a site.
Backlinks
There are 22 pages on the target site (homepage and articles, I’m discounting privacy policy/contact/etc).
I propose to use 110 expired Tumblrs which have are PA50+ to link to 16 pages on the site.
I’ve chosen 10 as I want a clear indication as to whether the Tumblr links are passing authority, and this will be easier to see if there are more URLs that are being targeted.
However, I need a minimum of 11 pages to remain as the control group.
20 links will be reserved for the homepage, otherwise, I see this as being a huge red flag if only inner pages are getting links.
The remaining 90 links will be distributed equally between 10 pages, which are the article/blog posts.
The limitation here is that those 20 links reserved for the homepage will likely pass authority to those 11 control group pages, as will the other links reserved for test group URLs, as the link equity passes through the site architecture.
However, I’ve read that this kind of in-direct link equity takes time to affect rankings, so there should be a lapse where I see the effects on the test group URLs, but not on the control group.
Anchor Text
We want to keep each page’s anchor texts as natural-looking as possible. Therefore we’ll be using a ratio of 1:9, exact-match to natural/random anchor text, to keep exact-match percentages at approximately 11%.
Exact match anchors will include the exact phrase we’re looking to rank the page for, where natural/random anchors will be the URL and examples like “here”, “Link to page”, “read more”, etc.
DoFollow/Redirect
Tumblr has limited the placement, and amount (with reason) of dofollow links that you can create on an account.
Until recently, we were able to bypass this re-direct by using a specific custom Tumblr theme. Therefore, we were able to create contextual dofollow links.
However, Tumblr have recently rolled our more platform updates, which mean this theme no longer works to.
Therefore, all of the links we create from Tumblr will be Tumblr re-directs.
Link Velocity
Since we’ll be starting with adding backlinks around the 7th of April, we’ll have approximately 90 days to complete the task. I want to have all links created in the first 60 days, as I suspect we’ll see the “Google Dance” taking place, this is described in their patent here as:
Changing a rank of a document by applying a rank transition function
It’s this rank transition function that is a temporary fluctuation (in either a positive or negative direction, or both) after links are built to a page, and link equity is flowing. It’s essentially a way in which Google tests whether the website owner has built the links, because a less experienced SEO or novice webmaster who has built those links, might see the fluctuations and then remove them in hopes that rankings return to their old position. This is essentially a trap and something that needs to be waited out. A Google dance can last anywhere from a few days to 1 month, depending on the amount of links/link equity sent to a page. So the trick is to be patient.
Therefore, I think that the last 30 days where links are not being created provides a sufficient amount of time for us to see results.
With that being said, we’ll need to build 90 links over 60 days.
However, to make things simpler, this will be the schedule:
Week 1: 1 link per URL
Week 2: 2 links per URL
Week 3: 1 link per URL
Week 4: 2 links per URL
Week 5: 1 link per URL
Week 6: 2 links per URL
April 2022 – Expired Tumblr Ranking Case Study Progress
01st April – We’re currently in the process of purchasing the Tumblr accounts. And the turn-around time for us to get hold of these accounts will be around 7-10 days.
Therefore, we should be able to start linking to the test site by the 7th of April at the latest.
We’ll usually use Scrapebox to find these kinds of accounts, but considering we want to start the test now, and we’ve used this provider in the past, this is the specific Fiverr gig we’ve used.
7th April – We’ve got the file full of 110 tumblr accounts, and are nearly ready to begin posting, due to the schedule which is shown further above. However, so that we’re not having to manually write the content which will be placed on these Tumblr accounts, we’ve used the VIP SEO Content-on-demand service via SpinRewriter (with having a Gold membership) to request 20 pre-spun articles to be sent over to us. This content is provided in Spintax form, so we can further spin it if necessary, but generally, the output that SpinRewriter’s team provides is considered unique content. Yes, this content is not going to be the best, but we just don’t have the time to write 110 blog posts for this SEO test.
Whilst we’re waiting for this content, we’ll begin filtering the tumble accounts by metrics such as DA/PA/Moz Rank/Number of links and provide an even split of accounts, when it comes to their SEO metrics) to be directed towards each target URL on the test site.
The next update here will be in a few days, so be sure to bookmark this page and check back then!
7th – 10th April (Week 1)
Week 1’s links have been created, and this includes 1 link to each URL, with there being 11 URLs in total
Although it’s a little too soon to see ranking changes just yet, we can see from the image below that average rankings across the site have increased since beginning this link-building test. It’s very difficult to say whether this is just because of the link building, however, we’ve made no other changes to the site during this time, or since mid-last year.
We’re also still in the process of getting Google to crawl the pages and hopefully index the content. Even if Google doesn’t index the content, links should still pass link equity. During week 2 we’ll begin to drip-feed these 11 links to a range of Facebook pages to force crawl/index them.