A data scientist is responsible for mining through the
massive amounts of data that an organization has an access to, and
is required to come up with insights that could drive action. If your
business has an online presence, chances are a major chunk of business critical
customer data gets generated online. In order to analyze your
digital properties and their performance, you would already have deployed
a digital analytics platform like Adobe analytics or Google analytics.
These systems would be capturing metrics like visits, paths taken, time
spent, products browsed, number of logins and pages visited etc.
As an enterprise business, your business intelligence unit would
also be running their own series of analysis in order to assist the
business in making strategic decisions. Some examples of these analysis could
be assigning a propensity score for all the customers of the brand which
details their propensity to buy products or services from that brand.
Another analysis could be coming up with answers to questions like which
products or services are in demand by the consumers and should be launched on
priority by the organization. These analysis help the top management
to take decisions and shape the overall strategy and hence are
of utmost importance. The data points used for running these analysis
could come from customer databases(CRMs), transactional or order
management systems, product information management(PIM) systems
and digital analytics systems.
Personalization is the basis for becoming an
"experience business" and there is no bigger frontier to
personalization than a mobile phone. If you look at any ten random phones from
ten users, you would find that each would have their
own personalized unlocking pattern. This proves that mobile phones
are deeply personal devices and deriving personalization here is not
as simple as adding someone's name to your message while sending an email. In
order to personalize the content on your mobile app, you really need to
understand the consumer in detail. You need to combine the online profile of
the user with any other type of information that you have available for her and
use this intelligence in providing the personalized content and offer
to her.
In the past, I have been a business analyst as well as data
scientist for few Indian and global brands and nowadays as a management
consultant for digital initiatives, I get to speak to a whole lot of young
as well as experienced data scientists and digital marketers from all
sorts of organizations ranging from banks to media companies to ecommerce
portals. In this time, one thing which has remained constant across
every type of industry is that the digital analytics is not given its due
importance when it comes to data science.
This is resulting from the fact that the teams handling these two
branches of analytics are completely different in almost all type of
organizations. Digital analytics is primarily the domain of
digital marketers and they usually have few analysts within their teams to
come up with dashboards and distribute reports. Data science technologies
are usually housed within the business intelligence units and their
primary source of data input comes from all the offline systems. Now this
is not to say that the data science teams do not use online analytics data at
all but to suggest that the relative importance of online analytics data
is quite low when it comes to various data modeling activities. Some
of the new age companies, primarily in the ecommerce domain are trying to
bridge this gap by assigning a single person or team catering to both of these
different data sets. It makes sense for ecommerce guys to quickly realize the
importance of this as their business is primarily done online and they cannot
afford to ignore the critical online analytics piece.
In my opinion, there are two things which need to be done in order
to speed up this change. First is the change in mindset - A data scientist
is not the answer to all strategic problems for an organization but
a decision scientist is. A decision scientist combines the knowledge
of data science with the power of marketing communications and acts a
bridge between the marketers and data scientists. Another need is to use
those pieces of technology that could seamlessly connect the data
from digital systems to the data from offline world. The idea is lower the
overhead from putting in manual integrations in place which are usually not
fool proof and do not always work as expected. But this is easier
said than done. Even if a data scientist wants to build his model
using attributes from the digital analytics data, the data scientist
is either not familiar with the best in class technology which can make
the job of connecting different datasets easier or if known, is too technically
cumbersome for the data scientist to effectively use it.
One of the ways in which I have tackled this problem successfully
in the past is to use open source technologies like R or Python. R is one
of the most powerful languages for data analysis and provides the most up
to date innovations, in form of packages, in its field. One of these packages
that I have used extensively is "RSitecatalyst" which combines
the digital analytics data from Adobe analytics with R and makes web and
mobile app analytics variables available for modeling in R. There is a
significant advantage of using these technologies because it lets the data
scientist focus on the job at hand, which is to mine insights, and not worry
about the data accuracy and leakages resulting from inefficient
technology integrations.
If you would want to discuss more ways to make analytics power
meaningful insights to digital marketing strategies and for your
business, please feel free to reach out to me.
