Download - Smart Ideas for Big Data in Social Media
The Bigger The Better: Smart Ideas for Big Data in Social Media
Aleksander Zawalich Analyst Sotrender Research [email protected] Mobile #10, Warsaw, 10.02.2016
About Sotrender
5 20 800 16 3Team
MembersYears Marketers Countries Research
Team
• Research company specializing in measuring social media, as well as developing it’s own analytics software with usage of big data technologies.
• Founders: Jan Zając (CEO), Paweł Kucharski (CTO), Dominik Batorski (Chief Scientist).
Our Software
Our Software
How are we doing it?
• Apache Big Data: Hadoop, Hive, Spark, HBase
• R-software Analyses: R-Studio, own algorithms, machine learning, decision trees and random forests, open-source packages (e.g. dplyr, ggplot2)
Data statistics
35 4 30k
PlatformsTB of data Profiles
We have BIG DATA.What else do we need?
Big & SmartIDEAS
Idea #1 Trends Reports
Trends Reports
• Monthly reports for marketers
• Facebook, Twitter, YouTube• 3 countries (PL, UK,
Indonesia)• Have been published,
developed and improved since 4 years.
sotrender.pl/reports
PL Fanpage Trends Report – 03.2011
Rank Name Fans Growth Growth %
PL Fanpage Trends Report – 01.2016
PL Fanpage Trends Report – 01.2016
% of Engaged users Engaged users
PL YouTube Trends Report – 01.2016
Ratings number % of positive ratings
PL Twitter Trends Report – 01.2016
Most often mentioned profiles
Idea #2 Facebook Fans Interests
Facebook Fans Interests
• On-demand analysis about your fans: where else are they active and what are their interests.
• Useful for planning communication strategy and choosing brand ambassadors.
Facebook Fans Interests
Hey, which cars do you like apart from Honda?
Facebook Fans Interests
Example data
Facebook Fans Interests
Example data
Coactivity with categories
Facebook Fans Interests
Idea #3 Paid Posts Prediction
Facebook Paid Posts Prediction
• Estimation of paid posts on fanpages made with machine-learning and random forests algorighms.
• Based only on public data derived from 14 000 profiles.
• >90% accuracy of the model.
$$$
Facebook Paid Posts Prediction
Example data
The share of paid communication vs reach per post N
umbe
r of a
ll po
sts
Number of paid posts
Facebook Paid Posts Prediction
Example data
Non-paidPaid
Paid vs non-paid communication usage
Facebook Paid Posts Prediction
Example data
Non-paidPaid
Efficacy of paid posts – mean number of activities
Idea #4 Personal e-marketing
trainer
Sombrero App – personal E-marketing trainer
• iOS and Android app aimed at helping beginners at the beginning of their social media adventure.
• Big-data and experience-based everyday simple tasks.
sombreroapp.com
Sombrero App – personal E-marketing trainer
Add profiles Complete tasks See results
Thanks for your attention Let’s stay in touch!
Aleksander Zawalich Analyst Sotrender Research [email protected]
sotrender.comfacebook.com/Sotrendertwitter.com/Sotrenderslideshare.net/FanpageTrender
Go Mobile #10, Warsaw, 10.02.2016