figshare
Browse
Screen Shot 2013-08-10 at 11.11.10 PM.png (59.58 kB)

Custom Neural Networks Rescue Fan Engagement (Marketing Strategies For Businesses Suffering From Trolls Or Competitor Sabotage)

Download (0 kB)
Version 3 2013-08-11, 01:32
Version 2 2013-08-11, 01:32
Version 1 2013-08-11, 03:21
figure
posted on 2013-08-11, 03:21 authored by OS BH-LabsOS BH-Labs

After developing a series of nerual networks to analyze market demographic data and make suggestions regarding the ideal demographic to target, content, and ad design, we were able to grow the Facebook page of Indie Review Magaizine up from 100 likes to over 20k likes in a few days time. 

 

At the peak we were obtaining likes at a rate of around 2.5k a day, we were also creating ads with an average cost per like of 0.01 - 0.03 cents. 

 

We then began to experience problems with a troll, as we have a rather serious stalker situation. This person utilized an army of bot accounts to post offensive, pornographic, and disturbing content on our page, our posts, in the comments, and also used a tactic of liking our posts with bogus, offensive, and disturbing acconts. 

 

This had a significant impact on our page engagement and cause us to be placed on what we believe is Facebook's advertising blacklist. We have not had any confirmation from Facebook that any such list exists, but our data suggests that it does.

 

As outlined in a previous article, over 90% of our Facebook fan base came from advertisements for which we paid facebook (thus we are not suffering from the issues that plague pages who pay for bogus likes).

 

Unfortunately, a percentage of our likes which are "organic" include the bot accounts from the stalker, and those bot accounts have had a very serious impact on our page's performance. 

 

However, utilizing the software that've I've developed to do generic pattern recognition, we were able to recover the fan engagement of our Facebook page in less than 2 days by following targeting strategies, ad design strategies, and marketing strategies suggested by the software - even though we have been plagued by the trolling situation.

 

This suggests that the software can be used to optimize a wide range of advertising situations, and we plan to continue to investigate this. 

 

The figure here shows that were were able to rescue our engagement from around 3k back up to 10.5k in just a few days. 

 

At this time we have stopped all ads, and we are halting all experimentation as our Facebook page has been experiencing some rather serious technical difficulties. 

 

- Ads don't report stats 

- Stats will not update

- Ads are not functioning properly

 

Our editor will be speaking with Facebook about these issues early next week, and we will continue to post updated of our reserach progress on Figshare as the project moves forward. 

 

For questions you can contact me at Jamie@ITSmoleculardesign.com

History