Our Twitter data are not the equivalent for Twitter data

It feels like 5 years ago when we entered 40,000 Twitter accounts into our database - just as a test case to find out whether our business model for hyper-targeting through Twitter would be feasible. 

It was a hell of job as it took us lots of manual work to tag the accounts but it worked out extremely well. In such a way that our enriched Twitter database cannot be regarded as the equivalent for plain Twitter data. That’s what we want to explain to the jury of the #Promote Innovation Challenge. Our model contributes to all advertising workflows. In particular the B2B market, where it is more difficult to reach the right target group, can benefit from our advertising model and tailored audiences.

Twitter, what else?
Being a start-up in 2012 we wanted to focus on datamining and advertising. As we were convinced that ad models could be made more specific in targeting. Another goal was to have less waste.

 There were two reasons that we chose Twitter for this purpose:
•    Twitter has an open source database, which meant that we could access data such as profiles, followers, follow to and tweets and its content.
•    Its values as a news platform – no other platform is faster in spreading the news. In addition, people on Twitter are eager to know what’s happening in the world and are looking for news and developments. Their mindset is open for messages regarding their interest or affinity.

In 2012 we discovered the world based on 40k Twitter accounts
So, we started entering the 40,000 most followed Twitter accounts worldwide in our database. We wanted to know whom the accounts belonged to and what we would discover about the people who followed these accounts. 

The results of our 40k selection were spectacular:

•    The test appeared to contain a broad mixture and representation of relevant topics – news, health, agriculture, technology, IT related professions, music and entertainment, sport and many other industries.
•    By means of tagging all followers of the 40k selection we discovered that the 1% largest Twitter accounts were followed by 99% of all Twitter users. This meant that our goal of being able to pinpoint the right audience was nearby…
•    Manual tagging work paid off as we were able to really find someone’s profession and interests. This would enable a 1-to-1 approach in campaigns. We decided to stay away from models based on algorithms, which cannot go into details the way we envisaged.

5 years later we can find any tailored audience
We continued the tagging and found out more and more about people’s interests and the connections they make. The numbers of our Sudoku-database got filled in and completed in a way that was beyond our initial expectations: we now have 700 million enriched accounts in our pocket and can compose any tailored audience.

All these accounts are nicely listed in communities and sub audiences and ranked by most important influencers and their followers and follow to followers. You want to reach your customer? We will find him. And even more: we will detect other relevant target groups to expand your market with!