Comments (1 )
A new wave of IT service providers from India are ready to flood the market with fresh services. It has been variously labeled knowledge process outsourcing, business intelligence or analytics. These companies are ready to satisfy the big data and analytics services needs of companies in healthcare, financial services, consumer packaged goods and others. Service providers come in many flavors: traditional full-stack IT companies with analytics, new entrants that service all verticals, new entrants specializing in just one vertical and, finally, pure-play technology specialists.
How can clients prepare for and benefit from this new wave of services? We have identified the following requirements.
Prepare the organization: Becoming an analytics-driven organization requires more than the addition of software. The culture of the organization needs to be changed to one that is data driven in its decision making. Companies must be ambidextrous, with one leg exploiting the present and the other looking to explore the future. Companies like Google, GE, Netflix and others can be good examples to follow.
Understand the problem: If the problem is well understood and the techniques for solving them exist, then one can hire a single vertical vendor or a full stack vendor for end-to-end services. Examples include TCS, Cognizant, Tech Mahindra, HCL, iCreate, ZS or Fractal.
If the problem is an unstructured one - no clear problem statement or solution technique - it would be wiser to use a vendor that services all verticals. Examples include Mu Sigma, LatentView or the full stack vendors.
Sourcing intent: If the intent is to complement existing initiatives, clients need vendors that can provide them with analytics capacity on-demand. Most of the full-stack service providers have built a deep bench of data scientists that can be put to use on projects.
When setting up an R&D or experimentation center, it is better to form a strategic partnership with a multi-vertical or full-stack vendor and take a holistic approach to problems.
Finally, when creating your own centers for excellence on analytics, technology vendors can be great partners. This coupled with your own ability to recruit data scientists would be a good approach. Examples of such technology vendors are Tableau, QlikView, Splunk and Cloudera.
Build absorptive capacity: In order to make use of the vendor provided capabilities, companies should be building their own internal capabilities. This could be in the form of small, centralized centers for excellence. Lessons learned by this group should be widely publicized.
What to look for in vendors? Just as in traditional IT projects one needs to look for references. In addition, certain capabilities can be vital.
Lab environments: Many vendors are setting up visualization labs to demonstrate what your decision making environment would look like. This could be very helpful in assessing their capabilities before selection.
Decision styles and other biases: We like to think that we are rational in our decision making. In reality, there are many types of biases in our daily decision making. Look for expertise in cognitive psychology or at least an understanding of the work done by experts like Kahneman and Taversky.
Analytics organization: Vendors are still not clear where they should place their analtyics organization. Some have multiple units each within a vertical. Others have created a separate unit. In either case, talent sourcing and retention are proving to be difficult. The pay scales for data scientists, career paths and management of resources are quite different from the traditional programmers. Ask your vendor about their approach to managing data scientists.
Data scientists: Internal universities are being setup to train data scientists. It would make sense to assess their curriculum. In particular, check to see if they have good programs for statistical techniques, research methods and communication.
Model management Capabilities: In addition to the traditional capabilities for modeling, assess their ability to manage models that they create. Just as knowledge management helped with traditional IT projects make gains in productivity, model management will have a big role within analytics-driven organizations.
Connectors: Vendors (especially full stack and multi-verticals) have many divisions along with the organization. It is very important to bring to bear expertise from these divisions. Look for connector roles within these organizations that have experienced people with modeling, domain and communication skills.
Order takers vs. scientists: When solutions to problems are clear, it is easy to hire data scientists from these vendors. When dealing with unstructured problems, make sure to qualify the vendors on the scientific method and also good communication skills.
The business analytics market is still at its infancy. Like many other industries, much is yet to be learned. As others have opined, profitability and risk mitigation awaits the fast learners.
Comments (2 )
A former student of mine stopped by to discuss an idea that he was planning to implement. This conversation got me thinking about the general idea that he was utilizing -- telemetry. A definition that I found on the web about telemetry states:
automatic transmission and measurement of data from remote sources by wire or radio or other means.
The idea is simple from a technology perspective. Build an architecture that has dumb sensory nodes that can communicate simple information to a centralized hub using wired or wireless communication technologies. These sensors send information on an entity of interest and transmit that to a central location. This allows entities to be globally distributed but for information to be centralized. This allows for scarce experts to be concentrated and better utilized in the hub while the spokes can have low-cost dumb sensors.
The data itself can be about the social interactions of the object, attributes of the object or the basic components of the object. It is a continuum between social and deep information. Statisticians can analyze this data and discover hidden patterns that can be forwarded to scientists or domain experts. These scientists can now create hypothesis to be tested and use the patterns purposefully
This core idea can be used to collect information about people, machines or other every day objects (like cars,) The collected data must be cleaned up for quality purposes and then be used for monitoring, gaining insights, or predictions.
Some questions to consider. Who owns the data? What policies should be set for information usage? Is it necessary to collect deep information? What behavioral information should be collected? What about data quality (timeliness, frequency, accuracy, etc.)? Is there a need to integrate with third-party data? What techniques to apply to get information from the data? Any societal benefits? What new roles and responsibilities need to be created?
This idea is being applied to healthcare, telematics, personal analytics, manufacturing and services. General patterns of use will emerge over the next few years. Big data specialists and data scientist are going to be very busy creating business models around the idea.
Comments (2 )
Organizations are seeking to extract value from the large volumes of data that they are collecting. This data deluge has been spurred by the availability of cheap sensors, lower storage costs, explosion of social media tools and the drive to be data-driven or evidence-based.
Many of the success stories that we hear are centered around A/B experiments for product features or go/no-go decisions based on customer adoption. Developments in text mining has also resulted in computing customer sentiment for a product or company using unstructured data.
All these techniques result in innovation that are interesting and important. However, they are not of the breakthrough kind. GE may use the internet of things to decide on what product features to highlight or Mariott may choose to exploit a particular sentiment for short term gains. These are incremental in nature.
These are what one would term localized exploitation. I am still waiting to hear about data was used to redefine the business scope of an organization or new product introductions. In some sense this is not a fair expectation. Discontinuous or disruptive ideas are not a linear extrapolation of the past. As such, past data cannot be used for these ideas.
To paraphrase Ditty Harry -- "A Data Analyst has got to know his limitations."
A new generation of Indian entrepreneur has arrived. They have a unique perspective on how to run a business. One such trail blazer is Venkat Vishwanathan of LatentView.
LatentView is part of the white hot business analytics space. This company is very representative of the knowledge economy in India. It has employees that have business, math and programming expertise. It has all the trappings of a modern day company – a great location, fabulous offices, creative workspaces and a learning environment. They even have their on 5F model (product, customer, employees, sales, and environment) model they use to find signals that support decision making.
The thing that stood out for me was its adoption of the Chennai value system. Venky, while very accomplished, is very humble and determined. They want to be known for their quality and integrity in the marketplace. If this means turning down projects they are not afraid to do so. Everything about them is measured and deliberate. They aim to grow at a predictable pace of between 80 and 120% year over year. They recruit from a small set of highly acclaimed schools and choose their associates to fit well with their value system. Another thing that makes them unique is they are self-funded. This gives them the freedom to grow at their own pace and not lose their values.
Their employees hold stock options and even have an internal market that enables the management to buy back shares and allow employees to gauge value. This company will go places in this marketplace and with their value system and capabilities. If I were an employee and holding stock in this company, I would hold.
There is so much to learn from this company. There is a saying that if you want to get anywhere you need to know where you are from. This company has deep roots and will be going places. I hope to make this a regular stop during my visits to Chennai .
Comments (7 )
It was a glorious summer day in Bangalore and Mu Sigma was celebrating the class of 2010. The entire batch was being fed a traditional South Indian Brahmin meal cooked by the famous Arasuvai Natarajan. The servers were none other than the executive team lead by company founder Dhiraj (dressed in traditional South Indian garb).
What makes this company so successful at what they do? While the answer to that may be too complex to analyze, here are some things that stuck out during my conversations with the team.
While they understand the competitive landscape very well, they are not obsessed with the moves made by them. As their chief fulfillment office Ambiga put it – we are just focused on our company culture and executing to our plan. She feels that as long as they maintain a small company feel, while they expand and grow, they will be just fine. They have made investments in capabilities that allow free flow of information and allow associates to locate the right expert for solving problems.
They have created a unique cadre of knowledge workers that they label decision scientists. These are individuals who can move effortlessly between the business domain, mathematics, statistics and computer science domains. Their in-house university ensures that they have the ability to create and mentor this talent. Their selection process is tailored to detect talent who are good problem solvers with a curiosity and appetite to take on challenges.
When I asked one of their senior managers Anuj as to what they do, he replied “we solve customer problems.” They follow the scientific approach to problem solving and have adopted a structure for problems (referred to as DNA). They listen to the problems/symptoms/opportunities as described by the client. This is followed by a list of variables that they find critical to the exercise. They then perform a mind mapping exercise to breakdown the problem statement. They look at all factors that may affect the variables that they identify. This is followed by a pattern matching exercise, where they look for prior art that matches the current situation. This is followed by a similar search through the literature and conversation with experts to find approaches to solve the problem. It is now time to iteratively prototype a solution. There are two supporting elements to their approach. One, get the client organization ready for decision sciences. Two, use a holistic approach to problem solving.
This organization has continued to grow stronger over the years. It is being attacked from all directions. Traditional IT vendors are moving up the stack by introducing anaylytics to their portfolio. Management consulting firms are moving down the stack with these offerings. Other pure play vendors are taking them head-on. Finally, niche players are nibbling away at their positions one vertical at a time.
Emerging markets go through a lot of turmoil with new entrants, failures, mergers and acquisitions. The business analytics marketspace is no different, but Mu Sigma feels like it is built to last and the one to beat.
Comments (2 )
My good friend Sukumar is the chief innovation officer at Cognizant. His group has done excellent work in the area of knowledge management and social networks and they won many an award to prove that. This time around, he explained to me how they have created an internal app marketplace to solve the legendary IT development backlog problem.
Organizations struggle with projects that run over budget, fall short of the required needs and exceeds the time allocated. In addition, project and resource allocation is a complex process, often shrouded in secrecy but alway a political process. Cognizant has solved many of these issues by going to an app based platform and governance model for their internal application needs.
Given their size and complexity, this is a great accomplishment. They now have a transparent model for project selection, approval and retirement. Users can propose ideas for apps in their internal marketplaces. Developers can choose to work on one or more of these apps. A governance body makes budgetary and go/no-go decisions based on data. Since every app is API based, CTSH can provide the vital statistics on each one of them.
A side benefit of this model of development (that follows many of the principles of lean thinking) is that each one is carefully scoped and covers bounded functionality. When many such components exist, users can repurpose them to create new applications as needed, using concepts like mashups. This reduces development and testing time by orders of magnitude.
In my opinion, this may be the largest collection of apps behind the corporate firewall. In addition, many organizations can benefit from learning about their governance model.
Comments (2 )
One of my stops today was with Ananth Krishnan, the CTO for TCS. Just a few seconds with him and you will conclude that this IIT graduate runs his group as a scientist carefully formulating a theory. A true history buff, he has studied various R&D centers like the one at Xerox, GE, MIT, P&G, etc. and adapted them to the context faced by TCS. Our conversation was about the right mix of exploration vs exploitation within the R&D group at TCS.
Within TCS they move projects from the explore phase to the expand phase and finally to the exploit phase. Individuals specialize within each phase and are managed and compensated differently. The most interesting role I found was that of the evangelist. This person’s role was to socialize each innovation within the company, extended enterprise and with their customers.
With annual budget plans, customer forums, co-innovation networks and workshops for prospects and clients, TCS generates enormous feedback for its innovation. Project approvals require input from internal experts, service line heads, the CTO and from academic experts. Solicitation of input from various sources help TCS manage the proportion of project that are solely academic with those that are totally applied.
Most IT companies today are being utilized for automation projects. TCS has begun to find customers who come to them with their problems and use the expertise within TCS to solve them and then automate them. This portends well for TCS and its business lines.
Comments (2 )
Thomas Simon, the recruiting head of TCS, recently told me that TCS has figured out how to recruit in scale and quality. Here is a company which employs over 275,000 associates and has a industry leading 10% attrition rate. It is not unusual for TCS to set hiring goals of 50,000 in a quarter. Thomas had me intrigued and I decided to meet with his team and learn about their approach from his associate Sudeep.
What TCS has done is to create a Facebook like application (Know Me) inside, with profiles, badges, gamification and social networks. They have opened this to their qualified colleges (~450). This system (Campus Connect) allows students to create profiles, participate in discussions, solve problems/challenges. It has features of Innocentive, OpenStack, Smarterer, LinkedIn and discussion boards.
As students participate and contribute to this system, they build their credentials, learn and grow. They get to understand various challenges that TCS faces and how their experts go about solving them. TCS has around 600 of their in-house experts mentor students and develop their problem solving skills.
Behind the scenes, an analytics engine creates a leaderboard and helps TCS sift through profiles quickly. As a result, they have a good idea of people they would live to interview even before they enter the campus.
The next version of the system will include faculty participation. Faculty can be rated just as students are currently being done. The analytics engine helps them run various experiments that allow them to predict future performance of students based on their current behavior.
Once this system is fully developed, it can be a gold mine for finding talent and developing a pipeline to meet future needs. For example, if they anticipate a spurt in demand for Java programmers, they can advertise courses in Java and provide incentives for students to take them.
The platform goes beyond just mining for talent. Once the talent is found, it can be given advanced access to the internal knowledge management systems and proprietary IP developed by TCS. While hiring in bulk is certainly supported, niche hiring for R&D can be done by looking at the challenges section and singling out people who have been creative problem solvers.
Platforms like these will be very hard for competitors to replace. By being the first to market, students would have spend time building their profiles and learning through the familiar platform over their four-year education.
With thousands of students with varying degrees of competency applying for jobs at TCS, here is a filter that selects the best fit for TCS. With Campus Connect, TCS may have indeed solved the puzzle of hiring with scale and high quality.
Comments (2 )
I have always enjoyed my conversations with Mr. Lakshminarayanan of Cognizant. He always seems in a Zen state of mind. My conversation with him today was about the role of strategy in their portfolio of services.
As Google's chairman once pointed out in 2002, they have no long term business strategy. The way they chartered their course was through a series of experiments and allowing the data to guide them. Lakshmi pointed out a similar way for technology projects. When a company expresses interest in moving to the cloud, CTSH does the transformation and makes them aware of the possibilities this enables.
For example, if a pharma company would like to allow its retailers to experiment with their supply chain, they can do that with the App marketplace that clouds enable. CTSH shows the possibilities that technology enables by pointing to other companies (even in other sectors). Using their technological capabilities, they develop prototypes that demonstrate how the idea works. This can then be copied by the client for their sector. In some sense, this is like exploring feasible regions in the nearby search space for interesting ideas, as opposed to looking at distant horizons. We should label this approach as lean strategy.
What this has allowed CTSH to do is to move up the stack as the lower levels are getting commodotized and continue to get competitive returns for its clients. Traditional strategy firms do not have the technical skills to experiment and demonstrate the possibilities.
The race to commodotize may be on in the IT services marketplace, but some are breaking out with lean strategy and experimentation.
Comments (2 )
It is now common to hear of companies trying to create a vision of integrating their internal assets and competing as OneCompany. The recent announcement of Microsoft and the one by Pepsi in the past are examples. Pepsi couldn’t bundle Frito-Lay and its soda from their side and exploit synergies on the customer side. The same has happened to Microsoft. They have many products and services but are unable to bundle and transfer efficiencies and experience to the customer.
Microsoft’s reorganization should go a long way in fixing that problem. In the near future, customers will get a more integrated feel with MSFT’s products and services. This strategy is to make it more like Apple. Apple, however, has a great internal innovation culture that MSFT never had or lost. Expectations around the clock speed of innovation are very high today. That is the problem that MSFT needs to fix.
MSFT should think of itself as a sophisticated utility (unlike Amazon which is a basic infrastructure utility) and allow their customers to participate in the delivery of innovative products. The current trend in the market place is do it yourself. While organizing itself for internal innovation by packaging its internal assets into Lego like blocks, MSFT should think about ways to allow third parties to access these assets and innovate.
While there is no guarantee that this strategy would work, it has a better shot that trying to accomplish everything internally as OneMicrosoft. They learned this with Xbox, now it is time to do it for Xcompany.
Comments (2 )
Previous Page | All Posts | Next Page