Mistakes to Avoid When Setting Up a Big Data Firm Big Data is no more in hype cycle; it is clearly now mainstream.

You have been thinking about this great idea on Big Data. You are convinced in your heart of hearts that this is indeed what the world needs. Why, you think, isn't the timing just right? Big Data is no more in hype cycle; it is clearly now mainstream.Everyone is collecting so much data that they are literally drowning in their data lakes, and worse, they are not able to take advantage of this incredible asset they have.

Surely in such a situation, you think, your idea may be just what the world is waiting for.

But wait. Before you quit your cushy job and plunge full time into startup mode, stop for a moment, ponder on the following and, once you have figured it out, go ahead. The world is indeed waiting for your startup to take shape.

Be very careful of these pitfalls.

Every venture carries some risk and a Big Data startup is no different. However, there are some specific challenges in this emerging field which need good understanding. You need to know what you are getting into and you need a plan to avoid the pitfalls.

Difficult to find good people? Develop them.

While it is true for all the industries, there is a big shortage of Data Scientists, let alone good Data Scientists. And to compound that, there is not even a consensus on what defines a Data Scientist.

Should they be proficient in Mathematics/Statistics or should they know more of R and SQL? Should they have a deep domain knowledge or should they possess business analysis skills? Do you need a skillful data modeler or a master story teller? An artist with scientific training or a Scientist with an artistic streak?

As you guessed correctly, the right answer isa mix of all these. Keep looking for that perfect person, but in the meantime hire people, even people fresh out of college, with right attitude and then train them to acquire the right skills.

Do not hard sell your algorithm, tell what it will do for the client

Remember, each client will be at a different stage of the Big Data journey. While some may be in advanced stages of this journey, others may have just started it. You can talk about how powerful and efficient your algorithm isbut, what will really click is what the client can do with your algorithm. A story based on a practical application is any day better than the mathematical formulas or the pseudo code of the algorithm.

Therefore, put on your hat of consultative sales; work with the client to understand their pain points and help them articulate their problem statements. And then focus on the use-cases.

Difficult to calculate ROI? Build the business case with the client.

Big Data has not attained that level of maturity that enterprises can easily calculate the ROI. In some way, it is where eCommerce or Cloud Infrastructure were in their nascent years. In general, the CxO's know the benefits that will result from Big Data initiatives but it is difficult to quantify these benefits and hence,they not able to decide how much to invest in a Big Data initiative.

The best way to deal with this issue it is to help build compelling business cases. Define the benefits,for example, 3 per cent savings due to optimized inventory,accurate forecasting will reduce wastage by 5 per cent, prompt attention to customer complaints within 24 hours leads to reduction in customer churn by 3 per cent. In each case, ascribe numbers and monetary figures. It may not be perfect, but it still is a good start.

Data is everywhere and nowhere- triangulate within company and from outside

Everyone collects data and will collect more and more of it. However, the data models, applications, storage and archival practices have evolved over the years. Some data was collected for regulatory compliance, some because the ERP systems had a mandatory field in a transaction and perhaps, most of it was collected because the application were designed so. In modern days, every merger and acquisition, every new market or product line and every joint venture or partnership is making the data more and more complex.

The data scape is therefore not pristine but a rocky and rough terrain with gaps, inconsistencies, and many imperfections arising out of evolution of application and business landscape.

By default, such data does not lend itself easily for analysis. You may have to break down internal barriers, as well as triangulate this data with other external sources. Remember that the best way to find and collate the data is to focus on data discovery together with the client.

Data is private, the owners are supreme

Regulations pertaining to data security and data privacy are myriad and evolving. European Community is the frontrunner in this area and is drafting rules that more fiercely protect their citizens from data abuse. Privacy concerns make some part of the data scape unusable, and some parts become minefields. And you cannot circumvent these regulatory requirements, period!

On top of it, some of the data would be considered sensitive by the business users and hence restrictions will be imposed on its usage.

Work with the client to understand such restrictions and use techniques like anonymization, aggregation, masking, etc. to address these concerns.

So many problems?Should I do it or not?

Like I said earlier, every venture carries some risk and a Big Data startup is no different. However, the more you understand these risks, the better will be your risk mitigation strategies. And a thoughtful risk management plan is critical for evolution of a successful business.

One last word – analyze these challenges and have a plan to deal with them. But, do not let these problems deter you. Go ahead and make that leap of faith. After all, the world has been waiting for your idea for so long!