Ian Mullane posted on 26 March 2014
Gender distribution is one of the most basic data points for marketeers, yet it has never been available on Twitter as they do not require it in their sign up process. We understand how valuable it is, especially when you want to tailor messages to your followers so we have worked on providing a solution and have delivered a feature that provides Twitter Gender follower analysis accurately.
If you are just happy that you can now see the gender split of your Twitter followers then move on. If you have any interest in understanding how we achieve it, continue reading.
In the first step, we use a gender matching library based on first-name. The library has more than 40,000 defined first-names, but whilst this is a fast, it is not 100% efficient.
We then apply a second process driven through natural language classification. As an example, it has been observed that female names end with a certain few letters more frequently. Now, if you take the second to last letter also into account, further increase the probability of identifying the gender. This allows us to work backwards and assign a gender to a name with more certainty. If you really have to understand more detail, you can find reams of the stuff here.
There you have it! It would have been a lot easier if Twitter and Instagram had provided it in the first place.