Sign in Get started

DATA SCIENCE MACHINE LEARNING PROGRAMMING VISUALIZATION

Customer Life Cycle

Business is built around customers. Every business needs customers to survive. They are the source
of revenue. Needless to say, the success of a business is directly proportional to its ability to acquire
customers, nurture them, make them happy, solve their issues, and consequently make more money
from them. But for that to happen, the business needs to identify the right potential customers.
They have to figure out the who, what, why, and how.
Who are the potential customers demanding their products?
What do they want?
Why do they want this particular product?
How are the customers making their buying decisions?
How does a business go about doing this?
Typically all businesses have customer-facing people, like sales, marketing, and support who keep
talking to their customers.
They become the face of the company, but it’s impossible for a business to contact every potential
and current customer individually from time to time to understand their asks.
Imagine that there are serious challenges in keeping tabs on what customers think. When the target
Sign inone-on-one
markets are large, for instance, a million individuals or more, it is difficult to show Get started
attention. Also, with most businesses going online, the business do not have any direct contact with
the customers. DATA SCIENCE MACHINE LEARNING PROGRAMMING VISUALIZATION
They are scattered all over the world. We should also consider the competition that can reach out to
customers much faster than before. The traditional barriers of geography and language are gone. A
business in China can easily sell to a customer in the US, like Alibaba.
The competition is getting bigger and smarter. Customers today have more options for any product
or service, and the barriers to switch to vendors is becoming smaller.
This brings businesses to a situation where they need to understand and plan for what their
customers might do in the future.
For that, they need to predict customer behavior. Therein comes predictive customer analytics. To
be ahead of the curve and act proactively, businesses need to know what their customers might do
in the future.
Will they buy your product?
Will they switch?
Are they happy with the product?
Are they going to be dissatisfied?
Will they buy more?
Businesses need answers to these questions in order to identify the right customer, the right channel
to contact them, and the right way to offer to help them.
They need predictive analytics. Predictive customer analytics uses customer data to build models.
These models help predict future behavior. It helps businesses to target prospects who will convert
and identify additional products the customer might buy.
When customers have problems, predictive customer analytics will help businesses to identify the
right resources to solve the problems. It will help identify customers who might switch and make
them offers to redeem them.
And with predictive customer analytics, businesses can achieve this with lower cost and higher
effectiveness than traditional means.
The opportunity for doing predictive customer analytics today is better than ever before. There are
rich data sources that are widely available, such as web blogs, social media, chat, transactions, and
voice transcripts. Also, today’s big data technologies are capable of doing large-scale data
processing, integration, and storage in cost-effective ways.
We have multiple ways to understand what the customers think and feel, and be able to mine the
data to build effective models. We can then use these models for driving business towards the right
customers and making them stay with your business.
The relationship between a customer and a business goes through a lifecycle. Businesses provide
products and services that a customer wants, and a customer looks at a business to provide his
wants and needs. All starts with a need. Imagine that. A customer needs a laptop, so he goes about
looking for options based on the technical specs, deals, and customer service ratings.
Then the customer selects a store to buy the laptop. This is the purchasing process. Once the
customer has bought a product, he or she may purchase additional items, like cables, chargers,
storage, et cetera. The customer then continues to use the product for a length of time. He may face
issues with repairs and upgrade parts. After the laptop’s lifetime, the customer might want to buy
from the same business or go to a different business based on his previous experience and current
options. Sign in Get started

DATA SCIENCE MACHINE LEARNING PROGRAMMING VISUALIZATION

From a business perspective, the first stage in the cycle is acquiring customers. This cycle involves
advertisements, targeted marketing deals, et cetera, to pull the customer to their store. Once the
customer has bought the product the business might want to up-sell other items like cables, service
contracts, et cetera. It costs them significantly less to sell to an existing customer than to a brand
new customer. The business needs to insure that the customer is happy with the product through
proper service and support.
When the life of the product comes to a close, the business wants to regain the customer by
providing priority deals. Remember, it costs less to keep a customer than to acquire a new customer.
So we see that the business cycle is similar for both the customer and the business. The customer
will focus on the value of the product while the business is more concerned about revenue and
retention.
Apply analytics to the customer life cycle
Acquisition
Customer analytics can identify customers who are most likely to buy your products and services
through attribute and behavioral analytics. It can identify the right channels like e-mail, phone or
social media to contact these prospects. It can also be used to determine price points and discounts
that will most likely attract prospects.
Up-sell
When a customer buys a product they most likely want complimentary products. For instance, a
laptop buyer would need cases, cables and warranty. Predictive customer analytics can help identify
the products and brands that a buyer of a given product brand would buy and help the business to
recommend those. Sign in Get started
If you have used Amazon or eBay, you have already experienced this. At the bottom of each product
page you will generally find a section called Customers Who Bought This Product Also Bought.
Service DATA SCIENCE MACHINE LEARNING PROGRAMMING VISUALIZATION
Analytics can help predict failure patterns in products and guide the business to take proactive
actions. It will also help choose the right channel for providing service to various types of
customers. For example, some customers feel more comfortable with phones, while others like
email.
It can hook up customers with the right contact center agent It can be used to identify customers
who are most likely to be dissatisfied and reach out to them
Retention
Predictive customer analytics can help in preventing customer attrition. It can identify customers
who are most likely to leave and also predict their timing. This helps businesses to take actions at
the right time. Predictive customer analytics can help identify the right programs for attention
based on customer types.
So, predictive customer analytics can help businesses in engaging customers in a mutually
productive way in all stages of a customer’s life cycle.
Customer Data
Predictive customer analytics needs data that carries signals about the customers’ intentions and
behavior. It is important to identify the right data sources that provide these signals and use them
for model building.
The customer-business relationship consists of four entities:
the customer himself,
the product or service the customer buys or uses,
the channels of communication, like email, web chat, and finally,
the customer-facing agents representing the company, like sales or support personnel.
There are two types of data that are required for customer analytics.
the attributes of the entities,
the events in which the entities participate
Entities can be customers, products, channels, or agents.
Attributes of the customer include demographics, income, age, gender, location, etc.
Attributes of the product include type, price, quality, etc. Attributes of the channel include type,
frequency of use, and response times.
Attributes of the agents include handle times, resolution rates, etc.
These attributes includings what customer will buy and how much will pay.
The second type of data is interactions data. These are events related to what the customer does
with the company. Events come with a timestamp. Events include interactions like browsing,
emails, chats, phone calls.
They also include transactions like purchasing, filing a complaint, delivery of a product, payment,
etc.
They also include social media events where the customer might tweet or post on facebook about
the product or respond to a message, like liking or forwarding.
For example, Customer browsed the company’s website for 15 minutes yesterday, looking at
product or service. He or she was tweeting yesterday, asking for his friend’s opinion on a suitable
product or service. All this points to his intention to buy a specific product or service very soon.
The business can take advantage of this by offering him or her chat to answer any questions and/or
providing some leads to close the sale. Sign in Get started
The success of predictive customer analytics is determined the ability of the business to identify
multiple data sources that represent attribute and events data, acquire, mine, and integrate them to
DATA SCIENCE MACHINE LEARNING PROGRAMMING VISUALIZATION
build a customer 360.
The more complete the data is, the better predictions are going to be.
Customer Analytics Process
Predictive Customer Analytics is a continuous improvement model. It needs a well laid-out process
within the business to work and deliver.
The process starts with identifying use cases where predictive customer analytics can help the
company. Once the use cases are identified, architects need to identify the data sources that can be
utilized for this purpose then data pipelines need to be built to acquire, process, integrate and store
the data.
Then data scientists need to get to work to mine the data and build models. Models need to be
tested for accuracy before they can be deployed.

The performance of the models needs to be monitored and the models need to be fine-tuned. As
time goes on, additional data elements can be added in for better prediction work. For any of this to
work, there needs to be management buy-in. Once management is on board, we will be able to find
resources to undertake the project.
If you found this post useful, stay tuned for Part II ,Part III , Part IV and Part V
Other Articles:
Service Management for Data Scientist
Architecture for the Complete Data Pipeline