Cooladata – Consumerization of Big Data and Online User Behavior

Cooladata logoLast week CoolaData guys announced their platform general availability. The following post is based on few discussions I had with the company founders.
In his 20 years of experience working in the field of data analysis, Guy Greenberg, a co-founder of CoolaData, came to realize that there is no simple way to work out a Big Data without turning to professional services and integrating several systems. For instance, one might need to incorporate Hadoop with ETL and visualization tools.

 
However, these types of customized Big Data projects can take months to complete, and like all complicated IT projects, they will include a system maintenance team. CoolaData has been built as a result of these circumstances and customizes business needs as if they were developed in house.

The Case: Online Users’ Behavioral Analytics

The company’s CEO and other co-founder, Tomer Ben Moshe, explained that the company targets online sites that want to learn their users’ behavior. Although there is a world of knowledge and numerous tools to support tracking online user usage, the area is still painfully lacking. From personal experience, I can say that these tools are far from generating the transparency one need.  Calculating users’ habits and workflows can aid in the learning process of what generates customer satisfaction, ultimately leading to more business and higher revenues. Comprehensive online mega services such as gaming and ecommerce companies understand the significance here, and are still building their own solutions to the understanding consumer behavior.  While large numbers of users seem to complicate the situation, even as few as a couple hundred users constitute a big data challenge.

These tools provide a great start, however once an online service begins to grow, so does the pain. We found that web companies can launch these tools quickly, however once they start growing, the companies always seem to reach a point where they are forced to build their own in-house solution. Ben Moshe.

Multiple Data Sources

The Big Data challenge is generated from multiple sources of data. The use case of tracking online users’ behavior is not limited to your online product usage. Users appear all over the web and tracking behavior begins with monitoring their social activities and most recent interactions with your company. In addition with the wide selection of technical devices used to consume software products and services, it is hard to keep track of it all. Each user interface (UI) implies different experiences and habits, all of which generate a great amount of data.
According to Greenberg, business intelligence (BI) traditional tools are including functionalities such as pivoting, aggregations and visualizations. However qualitative analysis of an online user’s behavior is based on a sequence of events, each of which is individually delineated. For example, conclusions may be drawn from a user’s profile, workflow, time spent at a specific event and latest tweet.
Another interesting feature of CoolaData is their innovative session tracking – Sessionization. What I have learned from working with Kiss Metrics is that it is pretty challenging to define what the session route is, or when it begins or ends since online services always change. In addition, each assumption about how a session should look will most likely be incorrect. Therefore, you might want the system to build automatically, presenting and analyzing users’ behavior based on multiple sessions.
CoolaData’s platform includes a custom semantic query engine (CQL) that understands the connections between various data sources. It presents the relationships on a CoolaData “analytic document” that can be saved on Google Drive, embedded on other sites and shared. The CoolaData engine can also receive third-party data via API. All these make it fairly a mature product.
CoolaData’s platform is based on Google Cloud Platform (GCP), which is elaborated on in the Google Cloud Blog:

Google Cloud Platform as the foundation. Hadoop and Storm run on Compute Engine, and Google App Engine serves our Semantic Layer web API. Google BigQuery enables analysis of the huge datasets we collect from a variety of sources via Google Cloud Storage Learn more

When it comes to platform as a service (PaaS), the world is still underdeveloped. IMO CoolaData, however, is a nice example of how to leverage the advanced computational capabilities of Google’s infrastructure and frameworks in order to eliminate development and operational efforts that are not specific to your product values.

Final Words

CoolaData is a great example of how BigData evolution is increasingly becoming a tool for business growth. While CoolaData users may not be the most tech savvy individuals, company analysts or higher-level managers are most definitely next in line. For mega online services this capability is a must. In addition, from my experience over the last few years I tend to believe that also small organizations such as new startups with a few hundred users are also likely to find great value in tools like this, ridding themselves of the hassle associated with lower level analytics (such as Kiss Metrics).
Tracking user behavior is only one niche that companies like CoolaData can support. Big Data challenges occur with machine-generated data, leaving us at the very top of the long list to come when the cloud comes into play.

Related posts