fbpx

Facebook Experiment That Enhanced Organic Reach By 987%

I conducted an experiment a few years ago while I was trying to gain more insights on how Facebook’s algorithm worked and today I thought to share the results.

I can’t be more wrong with the timing of this blog though as Facebook has went after and sued a cloaking company that allowed the advertisers to cloak their real landing pages from Facebook’s review systems and instead served a “safe page”. So while I publish the results of this experiment, please know that I conducted it for research purposes, have not used the outcomes for monetary benefit, and publishing the findings here for educational purposes. I also encourage all of you to always use these platforms as defined by the terms of services. Now let’s dive in.

As I knew that Facebook is capping reach of certain domains and giving extra ordinary mileage to other domains, I went out on the search to understand deeply how the systems worked. We eventually found a way to leverage this while staying well within the TOS by using sub-domains. However, we didn’t just come up with sub-domains on the day 1. It took many other tests, one of which I’m going to publish here.

To conduct this test, I used a programmable link shortening service called Tr.im. I picked up an article from a domain that Facebook had given preferential treatment to and shortened the URL using tr.im. I then created a rule in the tr.im shortening system to continue forwarding all desktop traffic to the link I shortened but instead send mobile traffic to another dummy website hosting an exact replica of the same article.

I then picked up 2 Facebook pages with similar demographics, size and performance. I shared the dummy link directly on Page 1 and the tr.im shortened link on Page 2. In the next couple of minutes, we saw 987% more traffic coming on from the shortened link as it leveraged organic reach of the whitelisted domain but sent all mobile traffic our way.

This experiment set the foundation for us about how Facebook newsfeed is treating different domains and we eventually scaled and moved forward using the sub-domain method.