The Last Great Hope for Math PhDs: Kontagent, Wrangler Of Big Data

Garrett Wilkin
Feb. 01 2012, 02:31PM EST

Have you ever heard of a Chief Science Officer? I recently spoke to Kontagent’s CSO, Josh Williams, and was excited to learn that Kontagent’s world of big data processing is one of the few places in the world in which academic exercise is directly applicable to industry.  Kontagent is a user analytics platform which can easily be integrated into any web site or mobile app.  The ease of integration results from the decision to provide a RESTful Kontagent API.

In the computer science field,  we see that nearly all workers are a bit over educated for their roles.  Many programming positions do not require a masters level understanding, or even a full grasp of undergraduate course material.  I would guess that there are many under challenged programmers out there who only wish that they could do more.  But this fate will not fall on doctoral students of Mathematics.  No, no, Kontagent has problems in the big data world that can only be addressed by those with advanced theoretical understanding.

Kontagent is a big data playground.  Instead of monkey bars to hone strength and agility they have huge multidimensional data sets to train mathematical acuity.  As big data processors, Kontagent is one of the biggest in the field, being orders of magnitude larger than more recent players in the space such as Apsalar, whom we covered recently.

The basic idea behind the Kontagent API is that it is used to submit events that should be tracked.  For example you may wish to track application launch, close, and more detailed events around actual app usage.  You could find out which features users are exploring and which they are ignoring by simply giving each feature its own event.  This is how both mobile and web developers contribute to the big data deluge that happens every day at Kontagent..

One key difference between Kontagent and many other mobile engagement companies is that advertising is not sold hand in hand with analytics.  You may be aware that advertising companies are using high performance real time bidding systems (such as OpenRTB) to serve custom tailored and targeted ads to mobile phone users.  Here’s a little overview of how that works.  Engagement companies collect information on the use of a collection of applications.  All information is tied to an identifier that is unique to your particular device.  It’s called the UUID and it’s a way of creating an ‘anonymous’ profile of activity linked to a unique user.  At the time that an application which displays ads is launched, the UUID is hashed with a publcly known hash function and sent to the ad network.  In the case of OpenRTB, this would allow for advertisers to bid for the priviledge of serving ads to this particular user.  The tricky bit is that advertising companies can use the hashed UUID to check their own list of user engagement profiles.  If they find the hashed version of the UUID, voila, they suddenly understand how engaged this user is and whether or not they are likely to make in-app purchases.  The amazing and frighteningly impressive thing is that all this ad negotiation takes place in well under a second, quite possibly while the app appears to be loading.

Because of the intrinsic link between user engagement and adverstising value, Kontagent has decided to occupy the moral high ground by only providing analytics but not selling advertising based on the result of these analytics.  Since there is a motivation to change engagement calcuation to result in a larger value, Kontagnet thought it would be best to simply provide the engagement data while allowing users of the Kontagent service to interact with Ad networks if they would choose to do so.  This is a great example of limiting the business strategy based on the minimally relevant service offering.

Garrett Wilkin

Comments

Comments(2)

[...] a leader in the world of big data.  In our previous conversation we touched on the potential for PhD level mathematics students to apply their knowledge to the challenges in the big data space.  By the end of this second conversation it became [...]

[...] a few conversations with Kontagent earlier this year I was inspired to call them “the last great hope for math PhDs,” because the work they’re doing requires advanced statistics theory.  This contest is a [...]