Data Driven Marketing

Unit 1

Big Data refers to the explosion in the quantity (and sometimes, quality) of available and potentially
relevant data, largely the result of recent and unprecedented advancements in data recording and
storage technology.

The five V: velocity, variety, volume, veracity, value

Structured data vs unstructured data: what can be in an excel table.

Data can be used as a tool (collection, development, improvement), as an industry (provide, deal or
store a lot of data), as a strategy (big data on the edge).

To create a strategy on data: choose the right data (internal data or new ones, IT support, non-
obsoletes structures, management of unstructured data, identify important data, integrate siloes of
information), build prediction models (mining isn’t enough, complexity can be issue, very complex
not practical, identify patterns, more targeted more valuable), transforming the organisational
capabilities (business relevant analytics, big data and analytics have to be sync, analytics aid decision
making, embed analytics into simple tools, enhance visualisation).
 Unit 2: Data Ethics and Surveillance Capitalism
Preparation:

Definitions

The age of surveillance capitalism, Shoshana Zuboff: A new economic order that claims human
experience as free raw material for hidden commercial practices of extraction, prediction, and
sales. / A parasitic economic logic of goods and services is subordinated to a new global
architechture of behavioural modifications. /A rogue mutation of capitalism marked by
concentrations of wealth, knowledge, and power unprecedented in human history. / The
foundational framework of a surveillance economy. / As significant a threat to human nature in the
twenty-first century as industrial capitalism to the natural world in the nineteenth and twentieth. /
The origin of a new instrumentarian power that assert dominance over society and presents startling
challenges to market democracy. / A movement that aims to impose a new collective order based on
total certainty. / An expropriation of critical human rights that is best understood as a coup from
above: an overthrow of the people sovereignty.

GDPR is the most important change in data privacy regulation in 20 years. => reshape the way data is
handled across every sector from healthcare to banking and beyond.

Video:

Surveillance capitalism is meant to be undetectable and bamboozling. Common misconception: we

give to them a part of our personal data and we have control. No, only the least valuable part of the
information they can get. Before, those data were wasted data (but in fact very predictive: the way
we drive, our favourite colour or how fast we type). Behavioural surplus: they feed models with our
data to predict our behaviour. Surplus? More than needed to improve the services => can identify
groups. Conducted at a level that is not accessible, not only online. Can use photos for feeding
physical recognition AI and selling it to China to imprison Uighurs. Two findings: can manipulate
behaviour and emotion thanks to online content and we can exercise it bypassing users’ awareness.
Pokemon Go good example paid Niantic (start-up but developed by google) for footfall,
establishments buy lure models to have users walk into their shops: bodies were the actual currency
for the shops, the value was to be able to predict efficiently the destination (punishment reward
game, insidious). System of security are also used to have new surplus: voices, conversations… What
if we refuse? Companies can say ok but no updates, no supporting of functionality … Hostage!
Investors are more prone to invest in a firm which makes an app AND surveillance dividend. With
android, google have access to 90%. Predict the emotional status of the users. Everything on the
shadow is never coming to us.

Click through and Footfall

Cathy O’Neil Ted Talks

Algorithms are everywhere => winners and losers. We need 2 things for algorithms to work: data
and a definition of success. The problem is that with this definition success is subjective => opinions
embedded in code. Cars for instance can go wrong without any bad intent => cannot predict when it
goes wrong. Need to audit: 1/ data integrity check 2/ Definition of success check 3/ accuracy 4/ Long
term effects of the algorithm. The algorithms are translations, coders are translators, they are no
truth seekers.

Digital Service Act

It’s a new regulation on cookie, online ads, deleting illegal content, more transparency algorithmic.
Digital services Board with Digital Services coordinators. => Transmits, stores data (strictest for more
than 45M monthly users).

GDPR

Limited changes, further concentration in companies that own third party trackers (monopoly), but
cannot predict that it won’t enable more changes in the future.

Data Relationship management

It’s a new approach that may be a solution. It is divided in four different aspects that have to be
respected in order to foster a better relationship between customers and companies. (Data
invitation, Data security centre, data dialogue, data value proposition)
 Unit 3
Preparation

Explore – Analyse – Communicate – Interpret

Converting raw data into understandable pictures => effective decisions. Visualisation serves 2
purposes: Exploration (insights from data) and Communication.

Need context, to explore individually the data and more complex analysis. Common sense, sceptical,
data may not be relevant.

Organise information: Location (sources, where goods are distributed for instance), Alphabet (huge
amount, so similar names), Time (if not framework), categories (goods in industries), Hierarchy (to
assign weight or value)

"The least you can do is not have the words and images competing with each other" by Bob Gill

Think about intent and narrative arch. To design a… [what is the format?] About… [what is the
story?] There is always more than one story from the same dataset That... [who is the audience?]
Can use to... [what will they know/do/think/feel as a result?]

Histograms and Bar Charts: data of different categories

Line Charts: relation of two variables, changes! Time series mostly.

Scatterplots: correlation between variables

Boxplots: some key descriptive data points

Pie Charts: Proportion

Tree Map: ????

World map: geographical insights

Function then flow and form at last!

Here are some extra tools: – https://flourish.studio/ – https://www.datawrapper.de/ –
https://rawgraphs.io/ – https://vizzuhq.com/ (interactive)

Visual Analytics is a way to connect human and machines… Make it easy to understand results and
insights

Article:

With the huge amount of data, need to take a step back: visualisation is a must-have skill. Internet
offers the possibility for everyone to make visual with less money; however sometimes don’t think
and that’s a pb. Need to think of reflection, not of a chart. Need visual grammar. But be careful:
visual communication is not alone, many activities which need their own strategy. 2 questions: Is it
conceptual or data driven? Is it exploring or declaring? If it is conceptual, need to simplify, to teach.
If it is data driven, you want to show, enlighten. If it is exploratory (prototypes, iterating, automating,
2 types: hypothesis or completely searching) you’ll either want to confirm or to discover. Declarative
= affirm.

Four types:
Clear communication, structure, logic… Need to capture the gist of the information, synthesis. It’s
about teaching.

Sketches, just to breakthrough and rough thinking. Nothing special and then illustrated again more
properly.
Depends on whether you are testing a hypothesis (limited, focused) or mining for patterns (huge
amount of data and really complex). Hypothesis: the objective is to start using everyday dataviz, to
quit the exploratory lands. Exploratory, mining patterns: don’t know what we are looking for, so
loads of data, inclusively, and sometimes with real time updates. It can also use interactivity from
managers.

SIMPLE

For all the categories, people from social grade ABC1 are twice more likely to have an opinion than
those from C2DE social grade.
 Unit 4
Preparation
Introduction Video

Insight: gain an accurate understanding of someone or something. Knowledge about a market,

consumers, companies or other stakeholders… Provide opportunities, changes in desires, perception
of the products and the environment. Marketing Insight different from Marketing Research (Insight =
discovery, research = a tool). Focus on secondary data, already collected, publicly or commercial.
More time and cost-effective than primary data. Market analysis = quantitative and qualitative
assessment of the market (size, segments, competitors, economic, political and cultural
environment). => Industry description and outlook/// target market /// Competitive analysis ///
Distributive channels /// Barriers to entry /// Regulation. (Resources – Databases ESCP recommends
Marketline, Euromonitor International, LS:N Global, Statista, xerfi knowledge).

Handout

To gain those insights, we can look back, focus on the present (real time) or look forward. The
marketing Research method is composed of 5 steps: Research problem or question formulation /
Determination of sources of information and research design / Data collection method selection /
Data collection / Analysis and interpretation. But distinguish Primary research (data collected by the
company for experimental and non-experimental purposes) and Secondary Research (has been
compiled and could be widely available, internal and external).

Industry description and outlook. Current state, projections, forecasts. Includes size, trends, life
cycle, projected growth… important for segments.

Target market. Identifies the consumer market. Described with demographics, psychographics,
needs, purchase behaviour and emerging expectations. (Size and value)

Competitive analysis. Overview of competitors, positioning, strengths, weaknesses, marketing and

distributing strategies.

Distribution channels. Existing ones, degree of connection with customers, emerging channels,
channel power structure, relevance and importance of the online channels.

Barriers to entry. Cost of entry, dangers, weaknesses related to your company.

Regulation. Governmental regulation, restriction. Compliance costs.

The Anatomy of a Trend

Trends are all around us. We see them in what we eat, where we live and how we connect with each
other. And whether we’re aware of it or not, they dictate the products we want and the services we
crave.

Carla Buzasi, Managing Director of WGSN

Trend forecasting is often regarded as an unscientific industry. For the world of fashion and design,
and not for ‘real’ industries. – and yet when you look at these so-called real industries (finance,
economics, science, business) far more guesswork, gut reaction, and intuition is used there than
you’ll find in the forecasting industries!

Anne Lise Kjaer, Founder of Kjaer’s Global

Intuition + Hard Data + Experience => Trend Forecasting

Trend comes from trendan which meant “turn in a specific direction”. Now: A CONSUMER TREND IS
A NEW MANIFESTATION AMONG PEOPLE—IN BEHAVIOR, ATTITUDE, OR EXPECTATION— OF A
FUNDAMENTAL HUMAN NEED, WANT, OR DESIRE.

Micro trends = early influencers => mass adoption in 1 or 2 years (important for fashion for instance
but often come directly from macro trends)

Macro trends = at least 5 years and impact a lot of industries. Evolve from year to year.

To detect the patterns that can persuade consumers to change behaviour (trends) need techniques
and an open mindset, sensitive to the aspects of the society approach. Identifies weak signals,
anomalies or disruptions. They find points of tension between what people want and what’s
available, but also Early Adopters.

To identify trends:

Weak signals: anomalies, vague and without purpose. Disconnected and random but meaningful
when in clusters across industries.

Cultural brailing: created by trend forecaster Faith Popcorn to detect the changes in the way people
live.

Horizon scanning: structured approach to detect shifts in social, technological, economic,

environmental, political, legal or ethical landscape (STAAPLE)

Questions:
To forecast trends: review previous ones, research about future events, establish/describe current
situation, analyse and consider unexpected events and Identify common threads.
Tools for Trend forecasters

The trend cartogram: the who (innovator) the what (trend or innovation) the where (place trend
begins and current impact in society) the why (drivers influencing the trend) the when
(consequences long and short-term).

The Consumer Trend Canvas: Framework to assist managers and executive to understand consumer
trends and allowing applications. Transforms insights into idea, solution, product, service,
technology (innovation). Divided in two parts: Analyse section (Inspirations, drivers, basic needs and
emerging expectations) and Apply section (how, where, when, for whom the trend will be applied
referring back to the insights of the analyse part).

Once you chose a trend, you need to start analysing. Inspirations: how are the other businesses
applying this trend? New products, experiences, business models… List innovation that raises the
expectations… Finally broaden your view (industries, regions, cultures, businesses). Basic needs:
What deep consumer needs & desires does this trend address? Social status, self-improvement,
entertainment, excitement, connection, security, identity, relevance, social interaction, creativity,
fairness, honesty, freedom, recognition, simplicity, transparency… Drivers of Change: Why is this
trend emerging now? What’s changing? Shifts (macro changes) and triggers (immediate changes).
Emerging Consumer Expectations: What new consumer needs, wants and expectations are created
by the changes identified above? Where and how does this trend satisfy them? Innovation =>
expectations. To get these expectations, look for expectation gap between what they have and what
they want. Monitor other industries that may have an impact on yours.

Then the application part. Innovation type: How and where could you apply this trend to your
business? Vision, business model, product/service/experience, marketing. Who: Which (new)
customer groups could you apply this trend to? What would you have to change? Customer
personas.

Article

We are surrounded by trends, but we fail to identify the interconnexions. For companies macro
trends are a huge stake, for they lasts decades. Need to implement them in their corporate DNA. To
predict trends: intuition, experience and hard data. 1or 2 years for a micro trend, more than 5 for a
macro one (touches diverse industries). WGSN compares the whole consumers behaviour to allows
its clients to proactively change before their consumers. Look at a lot of different information:
influencers, biotechnologies, demographic shifts, subcultures, geopolitics… Travel across the globe to
get the gist of everything that happens in the world… Then every local bureau shares with the others
and find similarities… Just validate with hard quantitative data.

Long video

The longer uber has been in a city the less people are willing to wait a car. Change their habits. In
business today feels overwhelming like being in an avalanche of new products/services… What will
my client want next? 1 ask customers, don’t know what they want and why they are going to want. 2
we are going to watch them, understand their behaviour, but too much time and human resources.
3 big data, insights, but better in validating. Nothing work so? Analysing the innovations!!!! The new
expectations that these innovations bring. Human needs do not change a lot. Innovations change our
way of serving these human needs. Expectations spreading = trends. Against whom are we
competing? the people who create the new expectations. Ai predicts delay for train in Sweden => ai
beneficial (add real value to services). REI close for Black Friday => Insider trading (change internally
the company culture). If innovation fails, the expectations don’t fail automatically.

Market description and outlook

Japan = 6.389.237.400,00 euros //// Sweden = 656.709.000 euros

14.6% -> 22% //// 17% -> 21% (2017 – 2022 E commerce retailer)

Other solutions for distribution in Japan: 4.4% pharmacies, General Merchandise stores 7.2%, 64.7%
Optical good stores. Sweden 76/79 % of Offline sales are in Optical good stores.

Johnson & Johnson: 13.2% market share in Japan, best in Sweden 19.1% Essilor Luxottica (EU vs USA)
…

10 first in Japan = 56.2% in Sweden 63.8%

Steve Jobs: Market research cannot lead to a jump, just incremental innovation, customers can only
give feedbacks. Breakthroughs come from instinct. Innovation can create needs but prediction is
maybe too deterministic.
 Unit 5
Segmentation – targeting – positioning (STP) approach

STP => better understanding their customers and also marketing effort efficient and effective.

Consumption patterns are no longer traditional demographics (choose their own patterns). So
traditional criteria are limited in identifying the people’s needs.

Segmentation

Market Segment: Group of customers with similar needs, behaviour, lifestyle.

Market Segmentation: process to identify market segments’ attractiveness and which one can be
served efficiently and effectively.

Short Term: Salesforce allocation/call planning, Channel assignment, Communication program,

Pricing, Today’s competitors and my current relative advantage to the customer.

Long Term: Emerging needs, New and evolving market segments to serve, Planning for segment
development/growth, Not in kind competition/threats (satisfying customer needs in different ways),
Lead user identification and management, Market driving (vs. customer focused)

 Small sample of customers but apply to all.

Bases: tell us why segments differ and respond differently.

Descriptor: helps profiling and reaching each segment (offerings, promotions, communications).

Cluster analysis: creating clusters of customers depending on different variables.

Targeting

Once evaluated, need to select one or more.

A useful tool is the GE/McKinsey Portfolio Matrix: 1. Specify drivers of each attractiveness
dimensions and weight each driver to reflect their relative importance to the company and the
company’s competitive advantages (competitive strengths) 2. Rate segments on each driver through
score 3. Multiply weights times rates for each segment 4. Plot results in a matrix (as in the picture) 5.
Review/sensitivity analysis using different rates and scores (there might no be consensus)
Positioning

Differentiation: tangible or intangible difference on attributes between a product and its

competitors.

Positioning: A set of strategies a company develops to differentiate its offering in the minds of its
target customers. A successful positioning results in a distinct, relevant, and sustainable position in
the minds of target customers.

Mapping: techniques to develop positioning and differentiation through visualisation of the

competitive structure of the market perceived by their customers. Derive from data of customers’
perception of products, attributes, similarities between brands, preference for product or
behavioural responses.

Perceptual maps show the positioning of the brands in a market with the customers’ point of view.
They are asked to compare brand to do this map. Can be 2 dimensional or multi-dimensional. The
distances = differences or similarities between brands. The axes show what was in customers’ mind
in order to compare. The vectors show products’ attributes.

Brand association map:

In class:

Cluster analysis (creating the clusters) is different from cluster profiling (describing the clusters).

Data are divided in three types: qualitative, ordinal and metrics (more important, divided in 2,
whether they have a natural zero or not. They just have the same distance between two ideas, like
degree, weight, age). There are L1 metrics (between two points the way is a city map) and L2 metrics
(direct line).

The number of clusters is a trade-off: homogénéity and number of cluster).

Methods can be with centroids or, closest or furthest point…

Global consumer types.

Axis of the map are chosen by the one who is doing the mapping.
 Unit 6
Introduction to customer acquisition

The customer journey = Acquisition => Conversion => Retention and Relationship Management

Customer acquisition is IMPORTANT => Make money to meet costs and reinvest, but also evidence
for outside parties (investors, partners, influencers). But becoming harder and harder, marketing
more expensive and customer less trustworthy

Customer Acquisition Cost (CAC): cost for one new customer (events, marketing, advertising).
Calculated for a campaign or a time. Important, because can measure a ROI and the effort put into
marketing. Calculate: CAC = MC/CA (MC = marketing costs; CA = number of new customers). Need to
add to marketing cost everything that has a link to it (events or advertising…)

New Product Development (focus: value and testing its perception): the phases

Any kind of object or service that can fulfil a need or a want.

A product has 3 levels: Core product (satisfies the need), Tangible product (quality, style, brand
name, packaging) and augmented product (additional services that make the product attractive).
New Product Development (focus: value and testing its perception): Measuring Customer Value

Value = Benefit – Cost. (Net customer value = (Functional, Psychological or economic advantages) –
(Monetary, Perceived risk or inconvenience costs).

Internal Engineering Assessment: Rely on evaluations made by firm’s managers, engineers, and
sellers, use of the products by its employees, or computer simulations. Success depends on the
understanding of the customers’ behaviours and patterns.

Indirect Survey Questions: ask company personnel about the effect of one or more changes in
existing offerings on certain aspects of their needs or problems. This technique is indirect, in that the
questions focus on the changes rather than the worth of the entire product offering.

Value-in-use: “dynamic phenomenon that is experienced and determined in usage processes that
are sequential and/or concurrent series of resource integrating activities in which actors apply the
available resources to reach their intended (individual and collective) goals.”
Exercise: 10$????

Focus Groups: a group of 5-10 people discuss with a professional about the usage of a product
(better understanding).

Direct Survey Questions: Sample of customers willing to complete it, description of potential product
offering or concept, indicate the value of it to them.

Importance Rating: popular for the unconstrained questions measures, few boundaries so can think
that everything is important.

Constraint of importance ratings: can give top scores to all the benefits (difficult to know the various
importance), difficult to measure the willingness to pay (no trade off nor indications), no direct link
to customers’ behaviour.

Conjoint analysis: widely used ‘constrained questions measures’. Employs a field research survey to
ask respondents to provide their overall ratings for each of a set of potential offerings. All the level
of these offerings vary systematically. Appears most frequently in new product development
processes.

Benchmarking: determine consumers’ willingness to pay for each attribute or feature. Respondents
indicate how much more they would be willing to pay for added products features to the benchmark
offering. Similarly, they might indicate how much less they would be willing to pay if the selected
offering is removed.

Choice models: use past information to estimate the value of product characteristics that might best
explain future/current behaviour.

Multinomial Logit model: simple/powerful model.

Data mining: Many organizations keep extensive records of customer purchases. Companies like
American Express find that segmenting their customer databases provides a better understanding of
customer purchase patterns and often highlight opportunities for cross selling, up- selling and other
marketing initiatives.
New Product Development (focus: value and testing its perception): A/B Testing

Definition: split audience to show them different versions of a campaign.

Benefits:

- Increased Website traffic

- Higher conversion rate
- Lower bounce rate
- Lower Cart Abandonment

Unit 7
Social listening:

Social media listening: conversation/trend about the brand, but also about the market. Different
from social monitoring (data driven, reactive responses), social listening (insights driven, proactive
decision) create the kind of content the followers actually want. Generate new ideas based on
industry trends. Improve the customer experience by interacting directly with customers.
Continuously adapt your marketing strategy to satisfy the current customer need. It’s What vs Why.

It's important because:

- Survey customers like: 83% brand respond question, 68% join conversation, 48% buy
products from responsive brands
- Fast moving, high intensity, high pace => go viral. One unhappy can become thousands.
Listening can help building a solid reputation.
- When loads of people start asking the same pb => need to take note. Listening => finding
ways to solve.
- Boost acquisition. Many opportunities. Not only loyal but also enjoy the content. They fall
upon social media handles. Easier to convert followers into customers than random people.

Why and How

 For brand intelligence. Can identify common customer questions, comments, complaints,
demographics and general sentiment around the brand, and easily share those insights with
the rest of the team.
 For competitive intelligence. Get a sense of the share. Why not satisfied. Content that
outperforms ours. Identify new solutions.
 Industry intelligence. Keep eye on everything disrupting the space that may impact
organisation. Key political and social issues. Find gaps that could be solved by a
solution/product. Look frequently asked questions.
  Campaign insights. Lot of time coming up with campaign to launch but no information
whether it will work. Can be solved with social media.
o Track the impressions and engagements around the campaign posts.
o Quickly gather general sentiment around specific campaigns.
o Identify the top influencers discussing the campaigns.
o Understand which of the audience demographic the campaign resonates with.
o Figure out the key themes within the campaign that are being used and whether
those are positive or negative.
o Track collaborative campaigns from a single source of truth.
 Content strategy. Discovering word research in google and trend in social media helping
knowing topics before they become popular.
 Customer service. If you experience a spike in messages, dig through the trends in those
conversations to see what happened. Work to fix what happened. If the issue is out of the
control, develop a communication strategy. Find the most frequently used keywords and
hashtags receiving complaints. Closely follow the developing narrative. Share information
internally on the top locations and demographics. Send messages to help pinpoint issues.
 Product research. Conversations on product lines, competitors, industry.
 Influencer marketing. Promoting products through individuals capable of driving audience.

TED Talks

What do we need from tech platforms? Problem? When large group of people. => Space influence
group’s behaviours. Encodes norm and structures.

Spaces in real life are structured, but the culture of silicon valley => unstructured is better.

Anomie!!!

3 ways structure: built environment, programming (what’s happening), mayoralty (possess the
space)

No more user design, but public design.

Could be an idea to have an immediate social negative reaction to curb user’s behaviour.

Incidental exposure = relating

Need planners, architects, places for the different social plateforms to learn from each others.

Unit 8
Normally companies group product same in product lines and product lines with same consumers
Strategic Business Units (SBUs)

Practitioners view offerings as how they perform vs real marketplace conditions. Using data-driven
methods to streamline R&D toward best opportunities for growth, new development and
profitability.

Review portfolio of product will: Improve competitive positioning / Maximize product investments /
Ensure that product investments align with business objectives / Identify strong and weak products
to clarify resource allocation / Improve communication and collaboration
Goals:

BCG portfolio matrix

GE McKinsey

Companies must regularly review their offerings through, for instance, A/B testing and Conjoint
Analysis.

Conjoint Analysis A method for assessing customers’ preferences for that decomposes preferences
for product bundles into preferences (value) for each product attribute option (partworths).

- Attributes: The product features being evaluated by the analysis. Examples of attributes for
Laptops: Brand, Size, Color, and Battery Life.
- Levels: The specifications of each attribute. Example of levels for Laptops include Brands:
Samsung, Dell, Apple, and Asus.
- Concept or profile: The hypothetical product or offering. This is a set of attributes with
different levels that are displayed at each task count. There are usually at least two to
choose from.
- Task: The number of times the respondent must make a choice. The example shows the first
of the 5 tasks as indicated by “Step 1 of 5.
- Partworths/Utility values: Partworths, or utility values, are how much weight an attribute
level carries with a respondent. The individual factors that lead to a product’s overall value
 to consumers are partworths. Example partworths for Laptops Brands: Samsung – 0.11, Dell
0.10, Apple 0.17, and Asus -0.16.
- Relative importance: Also known as “attribute importance,” this depicts which of the various
attributes of a product/service are more or less important when making a purchasing
decision. Example of Laptop Relative Importance: Brand 35%, Price 30%, Size 15%, Battery
Life 15%, and Color 5%.
- Profiles: Discover the ultimate product with the highest utility value.

1. Design
a. –
b. Language understood by customer, levels like degrees of characteristics and should
be precise, realistic, feasible, levels mutually exclusive within each attribute.
c. Too much => does not work. Maximum of 25 bundles but 16 or less is best. Still need
to be careful
2. Obtaining
a. Verbal or physique or picturial description. Pictures make it more interesting.
Physical cool but too expensive to be implemented. Then method to obtain. Rank all
products, grade them or pairwise evaluation.
b. The main characteristic distinguishing choice-based conjoint analysis from earlier
types of conjoint analysis is that the respondent expresses preferences by choosing
concepts (products) from sets of concepts, rather than by rating or ranking them. A
CBC question is often referred to as a "task." A set of products (concepts) is
displayed on the screen and the respondent chooses among the concepts.