Post on 15-Apr-2017
Making sense of Consumer Generated Content
Recognising, Analysing and Interpreting
Images in Social Media
A ThinkBumblebee presentation
User generated content
Pure
Spontaneous
Unsolicited
Free flowing
Honest
The Social Media Image Revolution
Brands aren’t able to capture
the complete consumer
story on social media
Lesser writing, more posting
Consumers do not always describe the brand, the category, or the product
85% of photos with logos are currently missed
Brand advocates are also missed because most don’t‘follow’ brands behaviour or proclaim they are ‘fans’
No reference toBolthouse Farms
No reference to Uber
No reference toPampers
Marketers are currently missing the significance of these images
@benandjerrys followers
237,000
People who have a B&J photo and follow brand
3,555 (1.5%)
People showing B&J in their photos
123,725
29,625People who
mention “Ben”or “Jerry” in
text
Marketers are currently missing the significance of these images
Every picture has a story
Contextual (social, cultural, physical, functional)
Expression of an emotion, a mood, a tone
Life passions or aspirations and fantasies
Most importantly, real life pictures –current live, real time experiences or memories
Technology
can
recognise
visuals,
elements in
the visual
and analyse
for
associations
Image A
nalysis
Qualitative analysis can
help decode the meanings
at a deeper level
Combine the textual and visual data to arrive at a holistic picture
Analyse and interpret using semiotics analysis frameworks
How Can Our Brand harness the power of social
listening of visual data?
DATA TECHNOLOGY Recognising the Visual Data
HUMANIZING DATAAdding qualitative texture and finding meaning
1A. Category : spot
early trends via top
consumer stories
What ?Identify stories by category, geography and more using over 50 different filters
How?Which posts are driving the most interest in my category ? And what are they saying ?
1B. Category : Find the
Influencers and what
they are saying
What ?Identify the top influencers , advocates, people who drive opinions and what opinions?
How?Number of posts + Content of posts
INFLUENCER LEADER BOARD Who are the most influential people who post photos of your brand?
1C. Understanding
Category Codes
What ?Unearthing category codes, brand codes of category
How?Mine the conversations, posts and visuals generated and analyse with qualitative semiotic framework to unearth the codesPresent in an engaging video format
Cultural Re-entry, today
Food Clothes Nails
1D. Segment Mapping
and Profiling
What ?Identifying consumer profiles /segments
Identifying variables basis conversations that they would differ on
How?Reading visuals for
• Age• Gender• Ethnicity• Brands/product they not only
like, but have a depth of engagement with
• Rituals• Geography• Feedback on category
2A. Context Mapping
for Products
What ?Identify multiple contexts and interaction of products/ categories within the contexts
How?PHOTO MENTIONS How does your product or brand appear today in images people share?
FACE OF THE CUSTOMER What are the demographics of the people who post photos of your product?
PROMINENT SECONDARY BRANDS/ products What other brands appear in these people’s photos?
WORD CLOUD OF INTERESTS What words are used most often with photos of your product?
GEOGRAPHIC HEAT MAP Where are photos of your product most often taken?
2B. Product pairing /
grouping into communities
to give directions for
packaging/ formats?
What ?Understand the associated product clusters in a category
How?Photo clustering techniquesAffinity analysis
Cultural Re-entry, today
Food Nails
Product Reviews : Own
Brand + Competition
What ?
Track product review sites like Amazon to get more completeView of customer opinion
Evaluate product perception over time Using up to two years of back data
How?Track for textual + visuals Qual analysis for conversations
Food Clothes NailsUnderstand what product attributesConsumers and press are discussingon social networks
Complexity
All of what you have seen so far – Our Brand needs the analysis in near real time.
Solution – Call the experts
High level block diagram for the use case
Data sources
Text data
Image Data Analyzed
Image
Mtadata
Enriched data
with scores
KafkaData
Classifier
Image Analysis
Image analysis model
Data Modeller (H2O, R, Dato)
Prebuilt model
Storage
Visualization/Dashboards
Instagram, etc