Increasing conversion via site Chat feature compiled with Analytics data

Have been thinking about this product feature and have to document it, do let me know how you feel about the idea.

I am taking example of typical eCommerce site which obviously tracks conversion as its most important goal.
(Goal and scenario may differ based on business)

User reaches product page and adds a product to cart, event is triggered (agent at backend/CRM is alerted and request is assigned to agents at backend in round robin fashion to the agent who is free, as user clicks on close button (or approaches it) other event is triggered this passes page info to backend/CRM. Now there are tools like one offered by webengage. Ideally as user approaches close tab site should be able to identify user behavior and prompt a chat window asking visitor if agent can help. I have read webengage is experimenting something similar notification event.

Now lets see what happens at backend; since almost 60% users are returning customers (who have registered in past) and their credentials are saved in cookies, agent at backend sees a screen which shows the product on which user currently is and product/s available in cart. now magic is in some special values which have been calculated based on analytics data.

This is what agent does and notes down while chat is triggered.

a) LTV of customer is visible on the screen 
ref: 1. http://goo.gl/cx3fUd 2. http://goo.gl/WDhN6m 3. http://goo.gl/KNeZy
b) Additional discount which can be offered on each of catalog/category/product are already in front of agenct, max discounts which can be offered depending on 3 or 4 simple scenarios (eg LTV< x, if catalog is Mobile discount is 1%, or Toys its 10%) UI can also be in form of a calculator where depending on user inputs (user inputs can again be automated is user is logged in) else user can be asked to provide email id to check best possible discount for visitor.
c) Based on customer LTV (higher the better) discounts can be offered to customer. Discounts can be for immediate purchase or as loyalty points for future purchases.
d) What were previous purchase-product/category/catalog
e) What was value of previous customer
f) How many loyality points (if any) exist with customer
g) If customer is not interested, info can be stored on CRM about product/category/brand preference and this data can be fed into system so that later customized mailers can be sent to remind/convert customer.
h) Get first hand feedback, why customer is not interested in purchasing product and reason of leaving can be fed back into CRM system. 

Advantages
1. This shall reduce customer exit from checkout process and product pages.
2. Giving discounts only to those customers who actually matter to business.   (This can also be new customers in case).
3. Customer first approach, this means instead of waiting for customer to visit again or email/call or chat prompt action can be taken by seller to reach out to customer and offer best possible help.
4. Collecting valuable customer feedback data eventually leading to Big Data (Huh!!) assuming daily traffic is approximately 1/4 Million and 30% traffic reaches product pages ( that's a lot of customers in an year ~27 Million) 
5. Data mining can provide valuable insights how certain buckets of customers are behaving. Certain type of products are leaving, others accept the discounts. Are certain customers just waiting and closing window so that they may get prompt for discount


I will keep on updating this post, depending on feedback and thoughts of process improvements which come to me.