2. 14th January, 2017Author: Kashif Saiyed
500,000
1,000,000
1,500,000
2,000,000
2,500,000
3,000,000
2010 2011 2012 2013 2014 2015
SalesNumber
Year
Two-Wheeler Export Trend
• Export trend stabilizing
over the past few years.
Leading to more ..
• Focus on customer
acquisition in the domestic
market
• But what leads to
customer acquisition? Two
wheeler market requires
ingenious products to win
over competition.
• Differentiation through
product design.
If We Know..
Source: SIAM
3. 14th January, 2017Author: Kashif Saiyed
The market is smart
• Male youth (target customers) are tech savvy, intelligent, resourceful
• The long average sales cycle means that they spend a good
amount of time on digital mediums both before and after
purchase
• For example, myself after a few months of considering a bike
purchase (Bajaj V), decided against it due to a dislike of the
front design.
• Were there more people in my category? Yes!!
I found many other posting a similar reason of not buying
the Bajaj V (if they represent a significant part of the
population is another question)
4. 14th January, 2017Author: Kashif Saiyed
How can we help?
• By providing valuable, integrated, easy to
consume information regarding consumer
wants.
• Information which can be consumed by product design
and engineering, marketing and services teams in an
efficient and in a customized manner.
• Importantly, information which traditionally has been
ignored i.e. unstructured data
5. 14th January, 2017Author: Kashif Saiyed
Web
Forums,
Call Centre,
Social Media
Product &
Marketing
Managers
1. Product
support/engineering
explore problematic
features
2. Product
Design:
Research for
potential new
features
How it currently works..
Users/buyers/enthusiasts
Data Consumers
3. Marketing:
Brand
building,
campaigns
• Manual monitoring (difficult)
or via external agency
(disconnect)
• Time consuming
• Excruciatingly boring for a user
to monitor hundreds of
threads
• Not empirical
6. 14th January, 2017Author: Kashif Saiyed
Web
Forums,
Call Centre,
Social Media
Product &
Marketing
Managers
Buzz Model:
ML/Natural Language
Processing
- Text Mining
- Information Extraction
- Summarization
Proposed Buzz Model
Users/buyers/enthusiasts
Web
Dashboard
1. Product
support/engineering
explore problematic
features
2. Product
Design:
Research for
potential new
features
3. Marketing:
Brand
building,
campaigns
7. 14th January, 2017Author: Kashif Saiyed
End goal – Customer Acquisition
Immediate goal
1. Capture product features customers discuss (negative or
positive)
2. Predict features of products in ‘Buzz’ or being talked about
most
3. Catch user reviews early
4. Identify product features customers are having trouble
with
5. Track attributes of competitors
Goals
8. 14th January, 2017Author: Kashif Saiyed
Uses
• Helps supplement market research in development of new product features
• Catch issues earlier missed: Product engineering updated about problematic
features through new mediums (unstructured)
• Culture of empirical research and exploration
• Less reliance on external agencies
• Track ‘Buzz’ of competitors and associated product features
• Model reduces effort to monitor hundreds of forums and pages
9. 14th January, 2017Author: Kashif Saiyed
5. Ad hoc exploration to mine unstructured data from web, social
media, call center
Using the model
1. A dashboard deployed on an R server/ SAP Lumira
2. Users can search for and subscribe to a product category.
4. Prediction of buzz topics and product features
3. Receive daily emails with summaries and link to relevant topics.