Lecture 7 Flashcards
Paid search, also referred to as search advertising, search engine marketing (SEM), or pay-per-click (PPC), is the process of
advertising on search engines such as Google, Bing, and Yahoo!
Search marketers use paid search in conjunction with SEO practices and software to form
a comprehensive search marketing strategy
Why is Search Advertising Important?
Allows advertisers to place their product in front of people who are already looking for it
Search Advertising is important because…
Targeted ad based on people’s searches reduces the
audience to those already seeking out a product or service
• 3.5 billion searches performed per day on Google
SEO is about
optimizing your website to get better search result rankings (organic listings)
SEM is
a form of marketing focused on increasing a website’s visibility in search engine result pages through optimization and advertising
How are Search Ads Priced?
Auction style cost per click model, not based on impressions
Bid on keywords (similar to eBay auctions)
Ranking and CPC depends on variety of factors:
Competition Bid Amount Landing Page Relevance Ad Relevance Quality Score
Quality Score
CTR - Are users clicking on ad at an efficient rate?
Landing Page - Does the destination URL identify with the ad copy?
Historical Perform
Ad Relevancy - Does the ad copy relate to the keyword query?
Keyword Relevancy
Quality Score =
The Ad Rank of the person below you / Your Quality Score + $0.01
Search Advertising Trends
Mobile Prominence
Personalized PPC
Keyword Importance Waning
Mobile Prominence:
In May 2015, Google reported that more searches are performed on mobile devices than PCs in the US
Personalized PPC:
ads will continue to become more data driven, resulting in an increase in remarketing campaigns
Keyword Importance Waning:
Experts predict that keywords will be knocked off as top priority in PPC strategies with the adoption of product listing ads (PLAs) and search engines’ growing focus on semantic search
Search Advertising Software
Help businesses advertise on search engines
such as Google, Bing, and Yahoo!
Typically used by marketers to identify,
target, and bid on relevant keywords,
gaining more prominent positioning in SERP
for engaged users
Leveraged to create and optimize ads, as
well as track conversion and return on ad
spend (ROAS)
It can include 1st party (Google, Microsoft, and
Yahoo!) and 3rd party platforms (to manage
campaigns across multiple engines)
Customer Modeling
Process of predicting and forecasting behavioral aspects of customers’ future perspectives
Includes identification of marketing and campaigning targets and optimizing predictive analysis
Framework for Model Developing
Describe the case and the define the goal. Ex: Who will subscribe?
Select all available data types (internal and external) and remove points that are obsolete, redundant, incomplete
Select variables with predictive power and transform them if needed to avoid overfitting
Run the model and validate the data
Implement the model or select different data
Types of Customer Modeling
- Reactive
- Proactive
- Recommendation Engine
Reactive Modeling
Bayesian
Artificial Neural Network
Business Rules
Proactive Modeling
Logistic
Time to Event
Bayesian
Recommendation Engine Modeling
Content Based
Collaborative
Hybrid Systems
Bayesian:
based on Bayes’ theorem which provides a
mathematical framework for forming inferences through probability. Allows for decision making and market research evaluation under uncertainty
ANN:
AI representations inspired by the way in which the human brain functions. Interconnected elements (nodes) process simultaneously the information, adapting and learning from past example
Time to Event:
recognizes the importance of time and the tenure/position in the customer life cycle. The outcome of interest is not only whether or not an event occurred but also when
Logistic:
statistical model that uses a logistic function to
model a binary dependent variable
Content Based Filtering
Process of recommending content based on its characteristics
Used in a number of applications, including information retrieval (i.e. search engines) as well as recommender systems (i.e. Netflix, Pandora)
Content Based Filtering Advantages
- Highly relevant results
- Transparent recommendation
- Users can get started more quickly
- New items can be recommended immediately
- Easier to implement
Content Based Filtering Challenges
- Lack of novelty and diversity
- Scalability
- Attributes may be incorrectly or inconsistently applied
Collaborative Filtering
- Based on the idea that people who share an interest in certain things will probably have similar tastes in other things as well
- Method of making automatic predictions (filtering) about the interests of a user by collecting preferences or taste information from many users (collaborating)
Collaborative Filtering Advantages
- Benefits from large user base
- Flexible across domains
- More diverse recommendations
- It can capture more nuance around items
Collaborative Filtering Challenges
- Complexity and expense
- Data sparsity
- “Cold” start issue
Content vs Collaborative Filtering
Content Filtering: Recommends products based on similar products
Collaborative Filtering: Recommends products based on similar customers
Hybrid Systems
Combination of collaborative and content-based filtering that can provide more accurate recommendations
Hybrid Systems can be implemented in different ways:
- Making content-based and collaborative-based predictions separately and then combining them
- Adding content-based capabilities to a collaborative-based approach (and vice versa)
- Unifying the approaches into one model
Netflix uses a
hybrid system with recos based on
comparing the watching habits of similar users
(collaborative filtering) as well as by offering movies that share characteristics with films that a user has rated highly (content-based filtering)
Chanel-Product Level Modeling
Which channel should I spend my next marketing dollar?
How effective is a promotional event?
How do I balance between product and brand spending?
Solution: Marketing Mix Modeling Algorithmic Cross Channel Attribution Integrated Marketing Attribution Modeling
What is Attribution?
- Practice of allocating partial value to different touchpoints within the customer journey which have influenced a desired outcome
- “Give credit where credit is due”
Marketing Channel Attribution Modeling
Last-Touch Attribution Model
Multi-Touch Attribution Model
Last-Touch Attribution Model
Attributes entire value of a conversion to the last ad clicked or seen
Miscalculates ROI by assuming the consumer doesn’t see or click any other ads along the way
Multi-Touch Attribution Model
Attributes value to observed touch point along the path to conversion
Provides a more realistic assessment of ROI by capturing a holistic picture of the customer journey across channels
Marketing Channel Attribution Modeling
Same Touch
Linear
U-Shaped
J-Shaped
Time Decay
Same Touch:
100% of the credit goes to the hit where the conversion
occurred
Linear:
each touchpoint in the conversion path (Natural Search, Paid Search, Social and Email) shares equal credit (25% each) for the sale
U-Shaped:
40% of the credit goes to both the first (Natural Search) and last interaction (Email), the remaining 20% is divided among any interaction in between (Paid Search and Social)
J-Shaped:
60% of the credit to the first interaction (Natural Search), 20% to the last interaction (Email), the remaining 20% is divided among any interaction in between (Paid Search and Social)
Time Decay:
follows an exponential decay with a custom half-life parameter
Multi-touch Attribution Limitations
Custom MTA models can still arbitrarily distribute fixed credit to each touchpoint
Event though they are considered superior to other rule-based models they are still limited when trying to build a complete measurement and performance picture
Multi-touch Attribution–What is the Right Tool?
Algorithmic Weighting
Sophisticated Approach
Integrated Attribution
Algorithmic Weighting:
use machine-learning and sophisticated algos to calculate probability-based weightings
Sophisticated Approach:
analyzes all media touch points of all users (even
non-converters), incorporates non-media factors that play a role in the purchase behavior (i.e. economy), uses a statistical algo to distribute credit
Integrated Attribution:
“Holy Grail”, integrates multiple events stream to offer
a more complete and accurate view of the customer decision journey
Next Generation of Attribution Modeling
Econometric top-down (marketing mix)»_space;>
Bottom-up (algo att models)»>
Machine learning (agent-based models)
