Summary
Summary Digital Marketing & Analytics - Exam Study (Dec 2018)
- Course
- Institution
Exam preparation notes. Combined summary of lecture notes, lecture slides and reading materials, focusing on the learning objectives.
[Show more]
Seller
Follow
vermartial
Reviews received
Content preview
WEEK 1 – INTRODUCTION TO DIGITAL MARKETINGDigital Landscape
- Online Marketing nowadays is very interactive
- Many platforms and touchpoints, complex journeys, a lot of input and output → Need for analytics
- Digital vs. Offline:
o Targeted (even personalized) → Individually tailored offers
o Real‐time (results, monitoring and adjustments) → Offline is retrospective
o Product information → customer contributes as well (product reviews)
o Less intrusive, interactive → More information exchange
o Measurability → Everything can be measured
Online Touchpoint Types
Paid Media: Not owned → Advertising
- Brand initiated contact, brand to consumer
communication
- e.g. Display ads, SEA, Online product placement
Owned Media: Owned by you → Content
- Ideally customer-initiated contact
- Interactive communication
- e.g. Your app, website, social media
Earned Media: Neither own, nor paid
- Created by others → Engagement
- Customer initiated communication between
customers (can be positive or negative)
- e.g. Social WOM, online reviews, SEO, articles
Two new forms of media suggested: ‘sold’ and ‘hijacked’ (Article 1.3 Edelman & Salsberg, 2010)
- Sold media: company sells advertising space to other companies (creates revenue)
- Hijacked media: large amounts of negative feedback (threatens revenue, could cause costs to fix)
o Should maybe just be seen as negative earned media and companies usually recover
Digital Data Types
On a broader level we categorise on the nature of the source → all complementary
- On-site data: Data that we can publicly see / extract / scrape
o Data that's available on the web (Likes, product listings, ratings)
o Technique to get it: "web scraping"
o Web crawling is looking at different data on the internet, scraping is about actually
extracting it for analysis
- Clickstream / Behavioural: What, when and how people do things online → actual behaviour
o What they click, what they clicked before etc.
o Can't see it on the website, but there are many providers (Google analytics)
o Clickstream gives you timing, device, location information
o TOUCH‐POINT DATA: When and where did this IP‐address see or click on one of my
ads/content?
o SESSION DATA: Everything you do when you visit my website. What pages you see, how
long, what you
- CRM data: Customer online history with the firm (complementary)
o Any extra information on customer
o Demographics, relationship history, customer service contacts etc.
,Digital Online Metrics
- Next to knowing the data types, it's even more important what metrics to use in different situations
- Marketing Metrics: Tools which help companies quantify, compare, and interpret the performance
of their marketing activities → Measuring the impact of marketing activities
- Best metrics depend on objective: Measure customer engagement? Are online ads driving sales?
- A good idea is to look at the marketing funnel
- KPIs are important for seeing what the metric is actually telling us (brand awareness, loyalty etc.)
- Impression Rate & In-View Rate
o Fine balance between impression & in-view rate
o Impression does not equal consumers seeing your ad, it means the content is potentially
visible on the device, might be out of reach though and you have to scroll first
o Real importance is the in-view-rate (50% or more of ad in-view for at least 1 sec), but hard
info to get and most marketers still work with impressions (Article 1.1 Fulgoni, 2016)
- Bounce Rate: Percentage of website visitors that immediately leave again
o Important to understand: web page effectiveness/success and engagement
- Click-through-rate: Clicks divided by impressions
o Has decreased a lot recently
o Simple, fast and easy to measure, but actually not a good indicator of engagement and
marketing effectiveness anymore (Fulgoni, 2016)
- Conversion rate: Conversions divided by clicks
Conversion Attribution I
- Conversion = consumer performs desired action (purchase, register, subscribe, app download etc.)
- Customer journey: starts with a consideration, has different steps, influences, process, might take
minutes, days, months, depending on what you buy
- Which touchpoints are more influential/valuable? → Attribution (Nisar & Yeung)
- Attribution models can be rule-based (use assumptions) or data-driven (better when data and
analytical skills are available)
Rule-based Attribution Models (use assumptions):
- Last Click = last click gets 100% of credit, assumes last click has most influence (→ industry standard)
o Criticism: can’t ignore the touchpoints that lead to this last point
- First Click = what matters is how the journey starts
- Linear = all touchpoints count equal, we should not omit the ones in the middle
o Also called uniformly distributed
o Good: no touchpoint is omitted / Bad: doesn’t distinguish between any level of contribution
- Time decay = people have a memory which fades, so closer to contribution is more influential
o But sometimes the decision is made early and the other stops are just for confirmation, time
decay disregards this
o Mostly used for short-lived deals or promotional offers
- Position-based = first and last click get most credit
o Combination of first, last click and linear model
o 80% to first and last, 20% to the rest
o But still, the real trigger might sit among the middle touchpoints
, Data driven Attribution Models:
- No assumptions! Focuses on real value of each online touchpoint (then we can make predictions)
- Most commonly used: Logistic Regression, Probabilistic Models (Shapley Value)
- Logistic Regression:
o Each observation is a click path that ends with a conversion or not (Y=1 or Y=0)
o Calculating the relative contribution (β) of each touchpoint (X) for each journey
o Pro: Easy to use, insights into channel effectiveness, additional explanatory variables can
easily be added (e.g. time on site etc)
o Con: Does not account for touchpoint order
- Probabilistic Models:
o Calculating the probability that one additional touchpoint will make a difference on a certain
type of journey → marginal contribution (using identical journeys except for one touchpoint)
o Pro: Easy and intuitive, accounts for touch-point order
o Con: Computation problem, as there are unlimited number of ‘possible paths’, how likely is it
that two people have the exact same click path? Bias for channels that are later in click path
- Shapley model:
o Applies algorithm to dataset based on Shapley Value concept from cooperative game theory
o Team = all marketing touchpoints, Team members = each touchpoint, Output = conversions
o Every different combination will have a different potential value
o The SV Attribution computes the counterfactual gains of each marketing touchpoint
→ compares the conversion probability of similar users, who were exposed to these
touchpoints, to the probability when one of the touchpoints does not occur in the path
o Main idea: Marginal Contribution of each touchpoint through in‐depth analysis of different
combinations / orders (permutations) of this touchpoint with other touchpoints
o Pro: Takes into account the position of the touchpoint in the sequence
Introduction to Data Analysis
- Before choosing the analysis method we need to know the variable type
- Categorical: Non-numerical data → Logistic Regression
o Nominal: Different situations (e.g. M/F, different countries, purchase yes/no)
o Ordinal: Categories that can be ranked, but the intervals between values are not quantifiable
(e.g. High/medium/low, Type of education)
- Numerical: Real numbers / Metrics → Multiple Linear Regression
o e.g. Sales, number of clicks, time of video watched
o Interval: Equal intervals between values, but not relative to zero (Likert scale, Hotel ratings)
o Ratio: Real quantities, continuous, relative to zero
- We focus on two types:
o When Y is metric/continuous/quantity → Linear Regression
o When Y is discrete/nominal → Logistic Regression
Regression I
Outcome Variable is Metric (Continuous) → Multivariate Linear Regression
Example: Drivers of Online Sales
Our Data: From an online retailer. Y= Total order
amount (sales) per customer (in Euros) in 2017
Question: What drives online purchases? Which type of
ads are more influential? Or is it the loyalty program?
Our Predictors: Total number of clicks (#) for various
online ads, and loyalty program membership status
The benefits of buying summaries with Stuvia:
Guaranteed quality through customer reviews
Stuvia customers have reviewed more than 700,000 summaries. This how you know that you are buying the best documents.
Quick and easy check-out
You can quickly pay through credit card or Stuvia-credit for the summaries. There is no membership needed.
Focus on what matters
Your fellow students write the study notes themselves, which is why the documents are always reliable and up-to-date. This ensures you quickly get to the core!
Frequently asked questions
What do I get when I buy this document?
You get a PDF, available immediately after your purchase. The purchased document is accessible anytime, anywhere and indefinitely through your profile.
Satisfaction guarantee: how does it work?
Our satisfaction guarantee ensures that you always find a study document that suits you well. You fill out a form, and our customer service team takes care of the rest.
Who am I buying these notes from?
Stuvia is a marketplace, so you are not buying this document from us, but from seller vermartial. Stuvia facilitates payment to the seller.
Will I be stuck with a subscription?
No, you only buy these notes for $6.96. You're not tied to anything after your purchase.
Can Stuvia be trusted?
4.6 stars on Google & Trustpilot (+1000 reviews)
52510 documents were sold in the last 30 days
Founded in 2010, the go-to place to buy study notes for 14 years now