Saturday, March 29, 2014

Push vs. Pull: Which Way is Best to Do Big Data & Analytics?

Take a Hybrid Big Data Approach: Centralized vs. Decentralized; Quick Win vs. Long Term; Push vs. Pull and Innovation Lab vs. Expertise Lab....

Big Data is in every forward-looking organization’s agenda, and a lot of companies are reorganizing themselves in order to push Big Data & Analytics mindset and agenda. There are many paths to do that as each company has its specific culture and history. But ‘Push’ or ‘Pull’: Which way is best to do Big Data & analytics?

A Big Data and analytics strategy must be aligned first with the business strategy, to ensure business challenges are addressed. To be an independent function or part of an existing department is the second piece that should be tackled. Businesses must start with "WHY" and "SO WHAT" before the "what" team, technology and analytic tools. Also, the more strategic the problem statement, the more you get the ear of senior officials in the organization, which is a prerequisite for obtaining the right allocation of resources to ensure the success of the initiative. Big data and analytics are just a new great tool to solve business problems, find a specific use case that delivers some tangible business value. Remember there is no benefit to adopting the latest hot technology if it doesn't solve your business problem. If you don't start with the challenges that keep top management awake, escalating costs, customer churn, high employee turnover, etc. you risk investing in a team and technology without really helping the organization.

It's possible you may have a hybrid approach of centralized vs. decentralized; quick win vs. long term and push vs. pull: A centralized team focuses on long term strategic questions, as well as small focused teams in specific departments tackle pressing business problems. The quick wins solve handy business issues, but Big Data road-map focuses on building long term business analytic capability, so the result of short term interests controlling assets are needed to prosper in the long-term. The "push" is "pull", it means if there is a business that requires immediate and valuable help, it will "pull" the analytics services, as the solid analytic capability underpins business strategy and improve business decision making scenarios.

Big Data Innovation Lab vs. Expertise Lab: Generally speaking, the “Innovative Lab” strategy works better because it is less bureaucratic. The team will be accountable and have the dedicated focus that is necessary to get things rolling from the ground-up and innovation always works best in an empowered centralized approach, but long term adoption and transformation in an organization comes best through a decentralized pull that each stakeholder can leverage based on specific needs & priorities. 

”Innovation Team" seems to be the best approach if 
1) Sponsor/Leader of the Lab is a business-oriented person, not a technical one, and who is recognized throughout the company, as a consequence, he/she knows how to find internal opportunities to demonstrate the technology and how to sell the projects to his/her peers. 

2) Base on this first assumption: quickness to find the first internal project. The pitfall of this team will produce an Ideal solution to your organization, but the transition from current ‘AS IS’ to ‘TO BE’ is expected to be Complex. And you may get a few "quick wins" due to focus, but it is difficult to sustain when working in a vacuum --separate from business. 

“Expertise Lab” approach has its merit. The experts who knows in and out of organization understand the complexity and will not make a move to go to Ideal (customized/assumed Ideal) state in quick time, Instead, they would like transition to happen in more phases over a long period. This group can act like a SWAT team and enable different parts of the organization to take advantage of big data. Folks in different parts of organization could feel empowered by this approach and would be more likely to share best practices with others. 

The pitfall of “Expertise Lab” is that the team may come up with the best criteria to evaluate which projects to go ahead, give occasional advice to the project teams, but they won’t be accountable to deliver projects themselves, and you probably end-up with a panel of judges on a talent show! Both approaches help transition, but it is up to organization to decide on how quick and how ideal their Big Data & analytics should be.

The power of “Pull":. The bigger and more strategic the problem statement, the greater the overall interest, and pull is main force. It might be quite worthwhile not to "push" at all, but instead, focus on making big data accessible to users. Who can turn massive data sets into manageable small ones, fast, and support users who want digestible, accessible data? Not big science. In all cases, any investment must be aligned with value. But the question is more about how to get started, the underlying issue is that either your innovation team or expertise team, need the expertise and skills to be successful with big data. The companies have to gain incredible insight, value, momentum, etc, big data can be truly transformational. Things like gaining a better understanding of your customer (customer 360), or even how to monetize big data. Drive and fund analytics on the back of a real project.

Big Data deployment scenario

1) get executive business sponsorship 
2) start with a small dedicated team 
3) identify quick wins (the "low hanging fruits") and go for them. 
4) put big data science results into production immediately; don't fall into the trap of doing one scientific study after another without generating business value (measured in hard money) . 
5) as soon as something is found to work, automate it (=software development) and repeat it as often as possible. 
6) grow the Big Data team either innovation lab or expertise lab according to the skillset required 
7) measure the result and make continuous improvement. 

Today, organizations know a much better HOW to build a "Data Intelligence Unit" and "Advanced Analytics" hub within organizations, what they should do to contribute to value generation, which tools they need, etc. So businesses can take a bit more "aggressive" tactics to build those teams faster and add value earlier. Be prepared to fail fast and collect learning if it’s not a great fit. You may run across a better fit once you understand the technology better

There is no one size fits all Big Data solutions, organizations have to experiment and learn; take hybrid approaches to solving key business issues, pulling resources and talent to make innovate solutions, and build analytic capabilities for businesses to compete for the future.


Analogica data is a one of the Best Big Data Services Provider Company in India, provide acumens on operations, products and customers. We also support predictive analysis,Big Data Services, master data management, and real time dashboards.

Analogica is a Big Data Analytics, Processing and Solutions company based in India. Our team has lived the evolutions and changes in the data analytics.

Analogica Data is one of the best Big Data Analysis Company in India, provide effective Big Data Solutions, efficient Big Data Analytics services in india.

Even if the information on this post is far from what i was looking for, i have to admit that the grammar used is just amazing. You have been very balancing among all readers, making sure that you offer information to all. Maybe what you need is Keywords Review Help, in order to make the page rank of your pages upgrade. Just check the link and find out more on the same.

Capsule theory is an excellent concept to talk about, but you can't ignore the relation of capsule theories with Google cloud big data services.

In Apache Spark a Resilient Distributed Dataset (RDD) is a read-only multi-partitioned collection of objects. As with other Spark concepts, RDDs are immutable. Learn More

Post a Comment