SEO, Technology, Tests
We all know that generating relevant, high quality links to our content can help it to rank better in search, but there is still some debate as to whether outbound links have a similar effect.
On one side of the argument, linking out to authoritative sources on a subject is a way of associating your site with other high quality domains – kind of like buying a house in an upmarket neighbourhood. Also, providing users with immediate access to additional info from sites Google knows are high quality and accurate should indicate that your site is dedicated to providing a great user experience. Finally, most low-quality sites won’t bother linking out to other sources of information – it’s more effort, and spammy sites are often more likely to be caught up in the old-school mind set of hoarding link equity rather than linking out to other sites.
On the other hand, if outbound linking provided a significant benefit to rankings, it would be a very easy signal to game – getting a link from the BBC is extremely tough to do, while linking to them is a 5 second job.
So, we set out to conduct our own test to see if we could boost the rankings of a range of websites, simply by linking to authoritative sites in their niche.
As is often the case with tests like these, we needed some ‘clean’ test sites with no prior history that could interfere with the results, and a couple of keywords with limited competition. This keyword needed to be rare enough that our very basic sites could rank for it, but not so rare that there were no other relevant, high quality sites on the subject that we could link out to.
Through a highly scientific process of picking interesting words at random and checking the SERPs (Search Engine Results Pages), we settled on the term ‘Kihlepa’, which is a small pig farming village in Estonia.
The SERP looked pretty promising – there’s enough info on it for a knowledge box entry, but it’s obscure enough for Google to think it’s a miss-spelling. Perfect.
The plan was to:
Welcome to Kihlepa: Population…?
Once the sites were set up (basic WordPress sites on clean domains), the next task was to write some unique, engaging copy for each page, while ensuring the basic facts were accurate, and the same across all domains. These included:
Unfortunately, as there is relatively little info out there on Kihlepa, we had to be slightly creative with a few of the ‘facts’ we used, but the basic numbers remained consistent throughout. It was also hard to find suitable images for these pages, so we just added some random examples of what we thought Kihlepa might look like, based on what we’d read about it:
Once this was done, we set up GSC accounts for each domain and requested a crawl and index of each Kihlepa page, which resulted in all pages being indexed within a week.
To give the pages a bit of a boost, we also added a links to them from the Propellernet site, using the same anchor text for each. This eventually caused a fair bit of confusion internally, as well as a few calls from SEO agencies, eager to let us know that we had a mistake on our site.
How dare they? Home page links to single-page WordPress sites about obscure pig-farming Estonian villages is a crucial of our outbound linking strategy.
After a few months, the test sites had settled into fairly stable positions for the 3 main KWs we were monitoring:
As you can see, Test Site 17 is faring a little better than the others at this stage, though the differences between rankings are pretty small. It had crossed our minds that Google could end up grouping content under very similar domains together and just ranking one version, but all sites are ranking pretty well for ‘Kihlepa Facts’ so this seems unlikely.
Adding the outbound links
Next up, we added outbound links. It felt the fairest way to do this was to add the links in reverse ranking order, so that the site performing least well would have links to high quality sources added, the site performing second best would get lower quality links added, and the best performing site would be left to battle on alone without assistance. This way we could see whether the presence of outbound links was sufficient for the poorest ranking sites to leapfrog those ahead of it.
We chose the outbound link targets based on the three keywords we were trying to rank for, and added these in appropriate places in the pages:
You can thank me later for all the traffic Wikipedia.
Results! (sort of)
These links were added at the start of January and left to work their magic for 2 months. After this, I re-tested ranking positions. I’ve colour-coded the sites to show which had the links added, with Test Site 16 in green having the high quality outbound links added, Test Site 15 in yellow having the lower quality outbound links added and Test Site 17 in red having no links added.
Main ‘Kihlepa’ keyword:
‘Kihlepa facts’ keyword:
‘Kihlepa weather’ keyword:
In some respects, we did see an improvement in performance after adding outbound links to high quality, relevant websites. Taking Test Site 16 in isolation, the results look pretty positive:
The main ‘Kihlepa’ keyword has improved significantly, ‘Kilhepa Facts’ has maintained its strong position and ‘Kihlepa Weather’ has improved its already strong rankings further. In addition, the very similar performance of the Test Sites 15 and 16 (the two with outbound links added) might indicate that the links are having an impact.
However in the context of the overall test, the results have to remain inconclusive, as the performance of Test Site 17 (essentially a control subject with no links added at all) also improved, often in line with Test Site 16 (high quality links), so it’s difficult to tie improvements back to this single factor.
Making a better test
Part of the problem with this test was that the competition in the SERPs was pretty thin, thinner than we expected in fact. Within a very short amount of time, our test domains were dominating the SERP for the ‘Kihlepa Facts’ keyword, making it harder to ascertain any concrete improvement in rankings.
This was also a pretty small-scale test of something that has many variables, and as such, more data points would have helped in mitigating anomalies in the results. It would be interesting to try this with more domains, a slightly more competitive subject and a larger keyword set.
However, while the results were inconclusive, it’s also worth remembering that the value of outbound links is not just based on potential ranking improvements, but on user experience as well. Your site is unlikely to be able to provide enough in-depth info on every facet of your subject to satisfy every user without becoming bloated and unfocused, so it makes sense to link to other resources that might help provide additional information to your users.
In addition, the high quality sites you are linking to may well see the referred visits from your site and decide to investigate your content, and maybe even link to it as a useful resource themselves in the future.
With this in mind, and despite the inconclusive results of this test, we’d generally advise clients to link out to high quality authoritative sites in their niche to provide a better experience to their users, and if this helps improve rankings, it’s a welcome bonus.
If you have questions on anything raised by this test or on external links in general, get in touch.
BONUS – Hat tip to SEO Monitor for some pretty great PPC tactics
So while checking the SERPs as part of this test, we started noticing a pretty damn well-targeted PPC ad, asking us by name (sort of) how the Kihlepa experiment was going.
It’s a pretty great idea, and I’m not 100% sure if we’d been identified as a potential client first and then had ads targeted to the test we were running, or if there’s something more automated going on in terms of finding domains that contain terms like ‘test’ and then serving ads, but it’s a great approach.
With most tests like these being run on deliberately uncompetitive search terms, it provides a really low-cost way of getting ads in front of SEO Monitor’s ideal audience of search agencies.
Bravo SEO Monitor, we are now looking at a trial of your rank checking software, so job done!